diff --git a/CMakeLists.txt b/CMakeLists.txt
old mode 100755
new mode 100644
index 61a85bf4637f99157c7762b69e9d884edf90b874..634c581b203cbd77f1346a1689abc7bc7a1691cb
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -12,7 +12,7 @@
# Vladimir Klement
# Jakub Klinkovsky
-cmake_minimum_required( VERSION 3.4 )
+cmake_minimum_required( VERSION 3.5.1 )
project( tnl )
@@ -21,28 +21,25 @@ set( tnlVersion "0.1" )
set(CMAKE_MODULE_PATH "${CMAKE_SOURCE_DIR}/cmake")
include( OptimizeForArchitecture )
-include( UseCodeCoverage )
####
# Settings for debug/release version
#
if( CMAKE_BUILD_TYPE STREQUAL "Debug")
- set( PROJECT_BUILD_PATH ${PROJECT_SOURCE_DIR}/Debug/src/TNL )
- set( PROJECT_TESTS_PATH ${PROJECT_SOURCE_DIR}/Debug/src/Tests )
+ set( PROJECT_BUILD_PATH ${PROJECT_SOURCE_DIR}/Debug/src )
set( PROJECT_TOOLS_PATH ${PROJECT_SOURCE_DIR}/Debug/bin )
set( LIBRARY_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Debug/lib )
set( EXECUTABLE_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Debug/bin )
set( debugExt -dbg )
else()
- set( PROJECT_BUILD_PATH ${PROJECT_SOURCE_DIR}/Release/src/TNL )
- set( PROJECT_TESTS_PATH ${PROJECT_SOURCE_DIR}/Release/src/Tests )
+ set( PROJECT_BUILD_PATH ${PROJECT_SOURCE_DIR}/Release/src )
set( PROJECT_TOOLS_PATH ${PROJECT_SOURCE_DIR}/Release/bin )
- set( LIBRARY_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Release/lib)
- set( EXECUTABLE_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Release/bin)
+ set( LIBRARY_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Release/lib )
+ set( EXECUTABLE_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Release/bin )
endif()
# set Debug/Release options
-set( CMAKE_CXX_FLAGS "-std=c++11 -Wall -Wno-unused-local-typedefs -Wno-unused-variable" )
+set( CMAKE_CXX_FLAGS "-std=c++11 -pthread -Wall -Wno-unused-local-typedefs -Wno-unused-variable" )
set( CMAKE_CXX_FLAGS_DEBUG "-g" )
set( CMAKE_CXX_FLAGS_RELEASE "-O3 -march=native -mtune=native -DNDEBUG" )
#set( CMAKE_CXX_FLAGS_RELEASE "-O3 -march=native -mtune=native -DNDEBUG -ftree-vectorizer-verbose=1 -ftree-vectorize -fopt-info-vec-missed -funroll-loops" )
@@ -54,7 +51,19 @@ set( CMAKE_SHARED_LIBRARY_LINK_C_FLAGS_RELEASE "" )
get_filename_component( CXX_COMPILER_NAME ${CMAKE_CXX_COMPILER} NAME )
if( CXX_COMPILER_NAME MATCHES "icpc" )
message( "Intel compiler detected..." )
- set( CMAKE_CXX_FLAGS "${CXXFLAGS} -DHAVE_ICPC ")
+ set( CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -DHAVE_ICPC -wd2568 -wd2571 -wd2570")
+ #####
+ # Ckeck for MIC
+ #
+ if( WITH_MIC STREQUAL "yes" )
+ message( "Compile MIC support..." )
+ set( MIC_CXX_FLAGS "-DHAVE_MIC")
+ # build all tests with MIC support
+ set( CXX_TESTS_FLAGS ${CXX_TESTS_FLAGS} -DHAVE_MIC )
+ set( WITH_CUDA "no")
+ else()
+ set( MIC_CXX_FLAGS "")
+ endif( )
endif()
#####
@@ -67,12 +76,19 @@ if( WITH_CUDA STREQUAL "yes" )
set(CUDA_ATTACH_VS_BUILD_RULE_TO_CUDA_FILE OFF)
set(BUILD_SHARED_LIBS ON)
set(CUDA_SEPARABLE_COMPILATION ON)
+ # Use the CUDA_HOST_COMPILER environment variable if the user specified it.
+ if( NOT $ENV{CUDA_HOST_COMPILER} STREQUAL "" )
+ message( "-- Setting CUDA_HOST_COMPILER to '$ENV{CUDA_HOST_COMPILER}'" )
+ set( CUDA_HOST_COMPILER $ENV{CUDA_HOST_COMPILER} )
+ else()
+ message( "-- Setting CUDA_HOST_COMPILER to '${CMAKE_CXX_COMPILER}'" )
+ set( CUDA_HOST_COMPILER ${CMAKE_CXX_COMPILER} )
+ endif()
set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} ;-DHAVE_CUDA)
# disable false compiler warnings
# reference for the -Xcudafe flag: http://stackoverflow.com/questions/14831051/how-to-disable-compiler-warnings-with-nvcc/17095910#17095910
# list of possible tokens: http://www.ssl.berkeley.edu/~jimm/grizzly_docs/SSL/opt/intel/cc/9.0/lib/locale/en_US/mcpcom.msg
- set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} ; -Wno-deprecated-gpu-targets --expt-relaxed-constexpr -Xcudafe "\"--diag_suppress=code_is_unreachable --diag_suppress=implicit_return_from_non_void_function\"")
- #AddCompilerFlag( "-DHAVE_NOT_CXX11" ) # -U_GLIBCXX_ATOMIC_BUILTINS -U_GLIBCXX_USE_INT128 " )
+ set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} ; -Wno-deprecated-gpu-targets --expt-relaxed-constexpr --expt-extended-lambda -Xcudafe "\"--diag_suppress=code_is_unreachable --diag_suppress=implicit_return_from_non_void_function\"")
set( ALL_CUDA_ARCHS -gencode arch=compute_20,code=sm_20
-gencode arch=compute_30,code=sm_30
-gencode arch=compute_32,code=sm_32
@@ -115,9 +131,6 @@ if( WITH_CUDA STREQUAL "yes" )
set( CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} ; ${CUDA_ARCH} -D_FORCE_INLINES )
# TODO: this is necessary only due to a bug in cmake
set( CUDA_ADD_LIBRARY_OPTIONS -shared )
- # TODO: workaround for a bug in cmake 3.5.0 (fixed in 3.5.1)
- set( CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} ; -ccbin "${CMAKE_CXX_COMPILER}" )
- set( CUDA_HOST_COMPILER "" )
####
# Check for cuBLAS
@@ -234,7 +247,7 @@ if( ${SYS_TIME_INCLUDE_DIR} STREQUAL "SYS_TIME_INCLUDE_DIR-NOTFOUND" )
message( "Missing header file sys/time.h" )
set( HAVE_SYS_TIME_H "//#define HAVE_SYS_TIME_H 1" )
else()
- include_directories( ${SYS_TIME_INCLUDE_DIR}/tnl-${tnlVersion} )
+ #include_directories( ${SYS_TIME_INCLUDE_DIR} )
set( HAVE_SYS_TIME_H "#define HAVE_SYS_TIME_H 1" )
endif()
@@ -246,7 +259,7 @@ if( ${SYS_RESOURCE_INCLUDE_DIR} STREQUAL "SYS_RESOURCE_INCLUDE_DIR-NOTFOUND" )
message( "Missing header file sys/time.h" )
set( HAVE_SYS_RESOURCE_H "//#define HAVE_SYS_RESOURCE_H 1" )
else()
- include_directories( ${SYS_RESOURCE_INCLUDE_DIR}/tnl-${tnlVersion} )
+ #include_directories( ${SYS_RESOURCE_INCLUDE_DIR} )
set( HAVE_SYS_RESOURCE_H "#define HAVE_SYS_RESOURCE_H 1" )
endif()
@@ -258,15 +271,20 @@ if( ${SYS_IOCTL_INCLUDE_DIR} STREQUAL "SYS_IOCTL_INCLUDE_DIR-NOTFOUND" )
message( "Missing header file sys/time.h" )
set( HAVE_SYS_IOCTL_H "//#define HAVE_SYS_IOCTL_H 1" )
else()
- include_directories( ${SYS_IOCTL_INCLUDE_DIR}/tnl-${tnlVersion} )
+ #include_directories( ${SYS_IOCTL_INCLUDE_DIR} )
set( HAVE_SYS_IOCTL_H "#define HAVE_SYS_IOCTL_H 1" )
endif()
if( WITH_TESTS STREQUAL "yes" )
- find_package( GTest )
- if( GTEST_FOUND )
- set( CXX_TESTS_FLAGS "-DHAVE_GTEST" )
- endif( GTEST_FOUND )
+ enable_testing()
+
+ # build gtest libs
+ include( BuildGtest )
+
+ if( WITH_COVERAGE STREQUAL "yes" AND CMAKE_BUILD_TYPE STREQUAL "Debug" )
+ # enable code coverage reports
+ include( UseCodeCoverage )
+ endif()
endif( WITH_TESTS STREQUAL "yes" )
find_package( PythonInterp 3 )
@@ -327,9 +345,6 @@ if( OPTIMIZED_VECTOR_HOST_OPERATIONS STREQUAL "yes" )
AddCompilerFlag( "-DOPTIMIZED_VECTOR_HOST_OPERATIONS " )
endif()
-set( CXX_TEST_FLAGS "-fprofile-arcs -ftest-coverage" )
-set( LD_TEST_FLAGS "-lgcov -coverage" )
-
set( configDirectory \"${CMAKE_INSTALL_PREFIX}/share/tnl-${tnlVersion}/\")
set( sourceDirectory \"${PROJECT_SOURCE_DIR}/\" )
set( testsDirectory \"${PROJECT_TESTS_PATH}/\" )
@@ -342,7 +357,6 @@ endif( PYTHONINTERP_FOUND )
#Nastavime cesty k hlavickovym souborum a knihovnam
INCLUDE_DIRECTORIES( src )
-INCLUDE_DIRECTORIES( tests )
INCLUDE_DIRECTORIES( ${PROJECT_BUILD_PATH} )
LINK_DIRECTORIES( ${LIBRARY_OUTPUT_PATH} )
@@ -350,7 +364,10 @@ LINK_DIRECTORIES( ${LIBRARY_OUTPUT_PATH} )
add_subdirectory( src )
add_subdirectory( share )
add_subdirectory( tests )
-add_subdirectory( examples )
+
+if( WITH_EXAMPLES STREQUAL "yes" )
+ add_subdirectory( examples )
+endif()
set(CPACK_PACKAGE_DESCRIPTION_SUMMARY "Template Numerical Library")
set(CPACK_PACKAGE_VENDOR "MMG")
diff --git a/INSTALL b/INSTALL
index ab759d74a7ea9dc57225f4f0fac1479dc58e7ac7..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 100644
--- a/INSTALL
+++ b/INSTALL
@@ -1 +0,0 @@
-/usr/local/share/automake-1.11/INSTALL
\ No newline at end of file
diff --git a/README b/README
index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..f4804b5e42323862dc09b426cf601af825ae6bfd 100644
--- a/README
+++ b/README
@@ -0,0 +1,80 @@
+Installation
+============
+
+ Requirements:
+
+ To install TNL, you need:
+
+ cmake 3.4 or later (https://cmake.org/download/)
+ GNU g++ 4.8 or later (https://gcc.gnu.org/)
+ CUDA 8.0 or later (https://developer.nvidia.com/cuda-downloads)
+
+ For image processing problems, you may optionally install:
+ DCMTK (http://dicom.offis.de/dcmtk.php.en)
+ libpng (http://www.libpng.org/pub/png/libpng.html)
+ libjpeg (http://libjpeg.sourceforge.net/)
+
+ The latest release of TNL can be downloaded as:
+
+ wget tnl-project.org/data/src/tnl-0.1.tar.bz2
+
+ Unpack it as:
+
+ tar xvf tnl-0.1.tar.bz2
+ cd tnl-0.1
+
+ Executing command
+
+ ./install
+
+ will install TNL to a folder ${HOME}/.local . You may change it by
+
+ ./install --prefix=<TNL prefix>
+
+ During the installation, TNL fetches latest version of Gtest and install it only
+ locally to sub-folders Debug and Release. At the end of the installation, the
+ script is checking if the prefix folder is visible to your bash and your linker.
+ If not, it informs you how to change your ${HOME}/.bashrc file to fix it.
+
+How to write a simple solver
+============================
+
+To implement your own solver:
+
+ Create and go to your working directory
+
+ mkdir MyProblem
+ cd Myproblem
+
+ Execute a command tnl-quickstart
+
+ tnl-quickstart
+
+ Answer the questions as, for example, follows
+
+ TNL Quickstart -- solver generator
+ ----------------------------------
+ Problem name:My Problem
+ Problem class base name (base name acceptable in C++ code):MyProblem
+ Operator name:Laplace
+
+ Write your numerical scheme by editing a file
+
+ Laplace_impl.h
+
+ on lines:
+ 34, 141 and 265 for 1D, 2D and 3D problem respectively with explicit time discretization
+ 101, 211 and 332 for 1D, 2D and 3D problem respectively with (semi-)implicit time discretization
+ Compile the program by executing
+
+ make
+
+ for CPU version only or
+
+ make WITH_CUDA=yes
+
+ for a solver running on both CPU and GPU. Run it on your favourite HW architecture by executing
+
+ ./MyProblem
+
+ and following the printed help.
diff --git a/build b/build
index bd644b6a111fd298803ba4e281747b28e0c27307..e3fbfe5e470bb80e09038269423ea2cfe2a0d798 100755
--- a/build
+++ b/build
@@ -2,35 +2,51 @@
TARGET=TNL
PREFIX=${HOME}/.local
+INSTALL="no"
+ROOT_DIR="."
+DCMTK_DIR="/usr/include/dcmtk"
+BUILD=""
+BUILD_JOBS=""
+CMAKE="cmake"
+CMAKE_ONLY="no"
+HELP="no"
+VERBOSE=""
+
WITH_CLANG="no"
WITH_CUDA="yes"
+WITH_CUDA_ARCH="auto"
WITH_TESTS="yes"
+WITH_COVERAGE="no"
+WITH_EXAMPLES="yes"
-WITH_CUDA_ARCH="auto"
-WITH_TEMPLATE_INSTANTIATION="yes"
+WITH_TEMPLATE_INSTANTIATION="no"
INSTANTIATE_LONG_INT="no"
INSTANTIATE_INT="yes"
INSTANTIATE_LONG_DOUBLE="no"
INSTANTIATE_DOUBLE="yes"
INSTANTIATE_FLOAT="no"
OPTIMIZED_VECTOR_HOST_OPERATIONS="no"
-CMAKE="cmake"
-CMAKE_ONLY="no"
-HELP="no"
-VERBOSE=""
-ROOT_DIR="."
-DCMTK_DIR="/usr/include/dcmtk"
-BUILD_JOBS=""
for option in "$@"
do
case $option in
--prefix=* ) PREFIX="${option#*=}" ;;
+ --install=* ) INSTALL="${option#*=}" ;;
+ --root-dir=* ) ROOT_DIR="${option#*=}" ;;
+ --dcmtk-dir=* ) DCMTK_DIR="${option#*=}" ;;
--build=* ) BUILD="${option#*=}" ;;
+ --build-jobs=* ) BUILD_JOBS="${option#*=}" ;;
+ --cmake=* ) CMAKE="${option#*=}" ;;
+ --cmake-only=* ) CMAKE_ONLY="${option#*=}" ;;
+ --verbose ) VERBOSE="VERBOSE=1" ;;
+ --help ) HELP="yes" ;;
--with-clang=* ) WITH_CLANG="${option#*=}" ;;
- --with-tests=* ) WITH_TESTS="${option#*=}" ;;
+ --with-mic=* ) WITH_MIC="${option#*=}" ;;
--with-cuda=* ) WITH_CUDA="${option#*=}" ;;
--with-cuda-arch=* ) WITH_CUDA_ARCH="${option#*=}";;
+ --with-tests=* ) WITH_TESTS="${option#*=}" ;;
+ --with-coverage=* ) WITH_COVERAGE="${option#*=}" ;;
+ --with-examples=* ) WITH_EXAMPLES="${option#*=}" ;;
--with-templates-instantiation=* ) WITH_TEMPLATE_INSTANTIATION="${option#*=}" ;;
--instantiate-long-int=* ) INSTANTIATE_LONG_INT="${option#*=}" ;;
--instantiate-int=* ) INSTANTIATE_INT="${option#*=}" ;;
@@ -44,13 +60,6 @@ do
INSTANTIATE_FLOAT="no"
WITH_CUDA_ARCH="auto" ;;
--optimize-vector-host-operations=* ) OPTIMIZED_VECTOR_HOST_OPERATIONS="yes" ;;
- --with-cmake=* ) CMAKE="${option#*=}" ;;
- --build-jobs=* ) BUILD_JOBS="${option#*=}" ;;
- --cmake-only=* ) CMAKE_ONLY="${option#*=}" ;;
- --verbose ) VERBOSE="VERBOSE=1" ;;
- --root-dir=* ) ROOT_DIR="${option#*=}" ;;
- --dcmtk-dir=* ) DCMTK_DIR="${option#*=}" ;;
- --help ) HELP="yes" ;;
* )
echo "Unknown option ${option}. Use --help for more information."
exit 1 ;;
@@ -61,15 +70,18 @@ if test ${HELP} = "yes";
then
echo "TNL build options:"
echo ""
- echo " --prefix=PATH Prefix for the installation directory. ${HOME}/local by default."
echo " --build=Debug/Release Build type."
- echo " --with-tests=yes/no Enable unit tests. 'yes' by default (libcppunit-dev is required)."
+ echo " --build-jobs=NUM Number of processes to be used for the build. It is set to the number of available CPU cores by default."
+ echo " --prefix=PATH Prefix for the installation directory. ${HOME}/local by default."
+ echo " --install=yes/no Enables the installation of TNL files."
+ echo " --with-mic=yes/no Enable MIC (Intel Xeon Phi). 'no' by default (Intel Compiler required)."
echo " --with-cuda=yes/no Enable CUDA. 'yes' by default (CUDA Toolkit is required)."
- echo " --with-cuda-arch=all/auto/30/35/... Choose CUDA architecture."
- echo " --with-templates-instantiation=yes/no Some TNL templates are precompiled during the build. 'yes' by default."
- echo " --full-build Instantiate all -- long int indexing, float and long double floating point arithmetics."
- echo " --with-cmake=CMAKE Path to cmake. 'cmake' by default."
- echo " --build-jobs=NUM Number of processes to be used for the build. It is set to a number of CPU cores by default."
+ echo " --with-cuda-arch=all/auto/30/35/... Choose CUDA architecture. 'auto' by default."
+ echo " --with-tests=yes/no Enable unit tests. 'yes' by default."
+ echo " --with-coverage=yes/no Enable code coverage reports for unit tests. 'no' by default (lcov is required)."
+ echo " --with-examples=yes/no Compile the 'examples' directory. 'yes' by default."
+ echo " --with-templates-instantiation=yes/no Precompiles some TNL templates during the build. 'no' by default."
+ echo " --cmake=CMAKE Path to cmake. 'cmake' by default."
echo " --verbose It enables verbose build."
echo " --root-dir=PATH Path to the TNL source code root dir."
echo " --dcmtk-dir=PATH Path to the DCMTK (Dicom Toolkit) root dir."
@@ -89,10 +101,12 @@ echo "Configuring ${BUILD} $TARGET ..."
${CMAKE} ${ROOT_DIR} \
-DCMAKE_BUILD_TYPE=${BUILD} \
-DCMAKE_INSTALL_PREFIX=${PREFIX} \
+ -DWITH_MIC=${WITH_MIC} \
-DWITH_CUDA=${WITH_CUDA} \
-DWITH_CUDA_ARCH=${WITH_CUDA_ARCH} \
-DWITH_TESTS=${WITH_TESTS} \
- -DPETSC_DIR=${PETSC_DIR} \
+ -DWITH_COVERAGE=${WITH_COVERAGE} \
+ -DWITH_EXAMPLES=${WITH_EXAMPLES} \
-DDCMTK_DIR=${DCMTK_DIR} \
-DWITH_TEMPLATE_INSTANTIATION=${WITH_TEMPLATE_INSTANTIATION} \
-DINSTANTIATE_FLOAT=${INSTANTIATE_FLOAT} \
@@ -109,7 +123,7 @@ fi
if test ${CMAKE_ONLY} = "yes";
then
- exit 1
+ exit 0
fi
if [[ -n ${BUILD_JOBS} ]]; then
@@ -128,14 +142,21 @@ else
echo "Building ${BUILD} $TARGET ..."
fi
-make ${VERBOSE}
+if [[ "$INSTALL" == "yes" ]]; then
+ # install implies all
+ make_target="install"
+else
+ make_target="all"
+fi
+
+make ${VERBOSE} $make_target
if test $? != 0; then
echo "Error: Build process failed."
exit 1
fi
-if test WITH_TESTS = "yes";
+if test ${WITH_TESTS} = "yes";
then
make test
if test $? != 0; then
diff --git a/cmake/BuildGtest.cmake b/cmake/BuildGtest.cmake
new file mode 100644
index 0000000000000000000000000000000000000000..5567e76c881bdd5359cffeb2c9fa07a40ec40231
--- /dev/null
+++ b/cmake/BuildGtest.cmake
@@ -0,0 +1,45 @@
+# Gtest developers recommend to build the gtest libraries directly from
+# the projects' build systems, see
+# https://github.com/google/googletest/tree/master/googletest#incorporating-into-an-existing-cmake-project
+
+#find_package( GTest )
+#if( GTEST_FOUND )
+# set( CXX_TESTS_FLAGS "${CXX_TESTS_FLAGS} -DHAVE_GTEST" )
+#endif( GTEST_FOUND )
+
+
+# compatibility with the GTest package
+set( GTEST_BOTH_LIBRARIES gtest gtest_main )
+set( CXX_TESTS_FLAGS ${CXX_TESTS_FLAGS} -DHAVE_GTEST )
+
+
+# Download and unpack googletest at configure time
+configure_file(cmake/Gtest.cmake.in googletest-download/CMakeLists.txt)
+execute_process(COMMAND ${CMAKE_COMMAND} -G "${CMAKE_GENERATOR}" .
+ RESULT_VARIABLE result
+ WORKING_DIRECTORY ${CMAKE_BINARY_DIR}/googletest-download )
+if(result)
+ message(FATAL_ERROR "CMake step for googletest failed: ${result}")
+endif()
+execute_process(COMMAND ${CMAKE_COMMAND} --build .
+ RESULT_VARIABLE result
+ WORKING_DIRECTORY ${CMAKE_BINARY_DIR}/googletest-download )
+if(result)
+ message(FATAL_ERROR "Build step for googletest failed: ${result}")
+endif()
+
+# Prevent overriding the parent project's compiler/linker
+# settings on Windows
+set(gtest_force_shared_crt ON CACHE BOOL "" FORCE)
+
+# Add googletest directly to our build. This defines
+# the gtest and gtest_main targets.
+add_subdirectory(${CMAKE_BINARY_DIR}/googletest-src
+ ${CMAKE_BINARY_DIR}/googletest-build)
+
+# The gtest/gtest_main targets carry header search path
+# dependencies automatically when using CMake 2.8.11 or
+# later. Otherwise we have to add them here ourselves.
+if (CMAKE_VERSION VERSION_LESS 2.8.11)
+ include_directories("${gtest_SOURCE_DIR}/include")
+endif()
diff --git a/cmake/Gtest.cmake.in b/cmake/Gtest.cmake.in
new file mode 100644
index 0000000000000000000000000000000000000000..762cff74ab9f856066a69c97c24fe69bfd076f66
--- /dev/null
+++ b/cmake/Gtest.cmake.in
@@ -0,0 +1,22 @@
+# vim: ft=cmake
+
+# This is a separate template for CMakeLists.txt to build gtest as a separate project
+#
+# Copied from upstream documentation:
+# https://github.com/google/googletest/tree/master/googletest#incorporating-into-an-existing-cmake-project
+
+cmake_minimum_required(VERSION 2.8.2)
+
+project(googletest-download NONE)
+
+include(ExternalProject)
+ExternalProject_Add(googletest
+ GIT_REPOSITORY https://github.com/google/googletest.git
+ GIT_TAG master
+ SOURCE_DIR "${CMAKE_BINARY_DIR}/googletest-src"
+ BINARY_DIR "${CMAKE_BINARY_DIR}/googletest-build"
+ CONFIGURE_COMMAND ""
+ BUILD_COMMAND ""
+ INSTALL_COMMAND ""
+ TEST_COMMAND ""
+)
diff --git a/cmake/UseCodeCoverage.cmake b/cmake/UseCodeCoverage.cmake
index d66c9a31a224ff35f96df47720eb3d2841812284..0e4b7f46f5c560a6169fdcbf72473517a5973830 100644
--- a/cmake/UseCodeCoverage.cmake
+++ b/cmake/UseCodeCoverage.cmake
@@ -1,31 +1,25 @@
#http://www.cmake.org/pipermail/cmake/2010-March/036063.html
-OPTION( ENABLE_CODECOVERAGE "Enable code coverage testing support" )
-
-if ( ENABLE_CODECOVERAGE )
-
- if ( NOT CMAKE_BUILD_TYPE STREQUAL "Debug" )
- message( WARNING "Code coverage results with an optimised (non-Debug) build may be misleading" )
- endif ( NOT CMAKE_BUILD_TYPE STREQUAL "Debug" )
-
- if ( NOT DEFINED CODECOV_OUTPUTFILE )
- set( CODECOV_OUTPUTFILE cmake_coverage.output )
- endif ( NOT DEFINED CODECOV_OUTPUTFILE )
-
- if ( NOT DEFINED CODECOV_HTMLOUTPUTDIR )
- set( CODECOV_HTMLOUTPUTDIR coverage_results )
- endif ( NOT DEFINED CODECOV_HTMLOUTPUTDIR )
-
- if ( CMAKE_COMPILER_IS_GNUCXX OR CMAKE_COMPILER_IS_GNUCXX )
- find_program( CODECOV_GCOV gcov )
- find_program( CODECOV_LCOV lcov )
- find_program( CODECOV_GENHTML genhtml )
- add_definitions( -fprofile-arcs -ftest-coverage )
- link_libraries( gcov )
- set( CMAKE_EXE_LINKER_FLAGS ${CMAKE_EXE_LINKER_FLAGS} --coverage )
- add_custom_target( coverage_init ALL ${CODECOV_LCOV} --base-directory . --directory ${CMAKE_BINARY_DIR} --output-file ${CODECOV_OUTPUTFILE} --capture --initial )
- add_custom_target( coverage ${CODECOV_LCOV} --base-directory . --directory ${CMAKE_BINARY_DIR} --output-file ${CODECOV_OUTPUTFILE} --capture COMMAND genhtml -o ${CODECOV_HTMLOUTPUTDIR} ${CODECOV_OUTPUTFILE} )
- endif ( CMAKE_COMPILER_IS_GNUCXX )
-
-endif (ENABLE_CODECOVERAGE )
\ No newline at end of file
+if ( NOT CMAKE_BUILD_TYPE STREQUAL "Debug" )
+ message( WARNING "Code coverage results with an optimised (non-Debug) build may be misleading" )
+endif ( NOT CMAKE_BUILD_TYPE STREQUAL "Debug" )
+
+if ( NOT DEFINED CODECOV_OUTPUTFILE )
+ set( CODECOV_OUTPUTFILE cmake_coverage.output )
+endif ( NOT DEFINED CODECOV_OUTPUTFILE )
+
+if ( NOT DEFINED CODECOV_HTMLOUTPUTDIR )
+ set( CODECOV_HTMLOUTPUTDIR coverage_results )
+endif ( NOT DEFINED CODECOV_HTMLOUTPUTDIR )
+
+if ( CMAKE_COMPILER_IS_GNUCXX )
+ find_program( CODECOV_GCOV gcov )
+ find_program( CODECOV_LCOV lcov )
+ find_program( CODECOV_GENHTML genhtml )
+ add_definitions( -fprofile-arcs -ftest-coverage )
+ link_libraries( gcov )
+ set( CMAKE_EXE_LINKER_FLAGS ${CMAKE_EXE_LINKER_FLAGS} --coverage )
+ add_custom_target( coverage_init ALL ${CODECOV_LCOV} --base-directory . --directory ${CMAKE_SOURCE_DIR} --no-external --output-file ${CODECOV_OUTPUTFILE} --capture --initial --quiet )
+ add_custom_target( coverage ${CODECOV_LCOV} --base-directory . --directory ${CMAKE_SOURCE_DIR} --no-external --output-file ${CODECOV_OUTPUTFILE} --capture --quiet COMMAND genhtml --quiet -o ${CODECOV_HTMLOUTPUTDIR} ${CODECOV_OUTPUTFILE} )
+endif ( CMAKE_COMPILER_IS_GNUCXX )
diff --git a/examples/CMakeLists.txt b/examples/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/examples/heat-equation/CMakeLists.txt b/examples/heat-equation/CMakeLists.txt
old mode 100755
new mode 100644
index 4b5bc34dba38e819d901cf46f311283c645d9c1a..9614bb21a6c453f80c6222b51899b7b33e74150d
--- a/examples/heat-equation/CMakeLists.txt
+++ b/examples/heat-equation/CMakeLists.txt
@@ -12,8 +12,9 @@ IF( BUILD_CUDA )
ELSE( BUILD_CUDA )
ADD_EXECUTABLE(tnl-heat-equation${debugExt} tnl-heat-equation.cpp)
ADD_EXECUTABLE(tnl-heat-equation-eoc-test${debugExt} tnl-heat-equation-eoc.cpp)
- target_link_libraries (tnl-heat-equation${debugExt} tnl${debugExt}-${tnlVersion} )
+ target_link_libraries (tnl-heat-equation${debugExt} tnl${debugExt}-${tnlVersion})
target_link_libraries (tnl-heat-equation-eoc-test${debugExt} tnl${debugExt}-${tnlVersion} )
+ TARGET_COMPILE_DEFINITIONS( tnl-heat-equation${debugExt} PUBLIC ${MIC_CXX_FLAGS} )
ENDIF( BUILD_CUDA )
diff --git a/examples/heat-equation/tnl-heat-equation-eoc.h b/examples/heat-equation/tnl-heat-equation-eoc.h
index e228a8ca98ed78b49b98537d61945616f82f7c99..f9876fb9adf66479282f1223bd48fb10da759ae7 100644
--- a/examples/heat-equation/tnl-heat-equation-eoc.h
+++ b/examples/heat-equation/tnl-heat-equation-eoc.h
@@ -53,16 +53,16 @@ class heatEquationSetter
typedef Device DeviceType;
typedef Index IndexType;
- typedef Containers::StaticVector< MeshType::meshDimension, Real > Point;
+ typedef Containers::StaticVector< MeshType::getMeshDimension(), Real > Point;
static bool run( const Config::ParameterContainer& parameters )
{
- enum { Dimension = MeshType::meshDimension };
+ enum { Dimension = MeshType::getMeshDimension() };
typedef Operators::LinearDiffusion< MeshType, Real, Index > ApproximateOperator;
typedef Operators::ExactLinearDiffusion< Dimension > ExactOperator;
- typedef Functions::TestFunction< MeshType::meshDimension, Real, Device > TestFunction;
+ typedef Functions::TestFunction< MeshType::getMeshDimension(), Real, Device > TestFunction;
typedef HeatEquationEocRhs< ExactOperator, TestFunction > RightHandSide;
- typedef Containers::StaticVector < MeshType::meshDimension, Real > Point;
+ typedef Containers::StaticVector < MeshType::getMeshDimension(), Real > Point;
typedef Operators::DirichletBoundaryConditions< MeshType, TestFunction, Dimension, Real, Index > BoundaryConditions;
typedef HeatEquationEocProblem< MeshType, BoundaryConditions, RightHandSide, ApproximateOperator > Solver;
SolverStarter solverStarter;
diff --git a/examples/heat-equation/tnl-heat-equation.h b/examples/heat-equation/tnl-heat-equation.h
index 3b289a8e614beebf317f6cfd4bc1011dfd92f390..1a0041f93e3f6edeb73b8bfe9d632ed9538f9f41 100644
--- a/examples/heat-equation/tnl-heat-equation.h
+++ b/examples/heat-equation/tnl-heat-equation.h
@@ -68,7 +68,7 @@ class heatEquationSetter
static bool run( const Config::ParameterContainer& parameters )
{
- enum { Dimension = MeshType::meshDimension };
+ enum { Dimension = MeshType::getMeshDimension() };
typedef Operators::LinearDiffusion< MeshType, Real, Index > ApproximateOperator;
typedef Functions::Analytic::Constant< Dimension, Real > RightHandSide;
diff --git a/examples/inviscid-flow/CMakeLists.txt b/examples/inviscid-flow/CMakeLists.txt
index 634cce16141410c692254c2b7385b2cac5acd5a7..19f8d03003c876c4c510a7d5ed5c616a3e2e0888 100644
--- a/examples/inviscid-flow/CMakeLists.txt
+++ b/examples/inviscid-flow/CMakeLists.txt
@@ -6,19 +6,19 @@ set( tnl_inviscid_flow_SOURCES
euler.cu )
IF( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE(tnl-euler-2d${debugExt} euler.cu)
- target_link_libraries (tnl-euler-2d${debugExt} tnl${debugExt}-${tnlVersion} ${CUSPARSE_LIBRARY} )
+ CUDA_ADD_EXECUTABLE(tnl-euler${debugExt} euler.cu)
+ target_link_libraries (tnl-euler${debugExt} tnl${debugExt}-${tnlVersion} ${CUSPARSE_LIBRARY} )
ELSE( BUILD_CUDA )
- ADD_EXECUTABLE(tnl-euler-2d${debugExt} euler.cpp)
- target_link_libraries (tnl-euler-2d${debugExt} tnl${debugExt}-${tnlVersion} )
+ ADD_EXECUTABLE(tnl-euler${debugExt} euler.cpp)
+ target_link_libraries (tnl-euler${debugExt} tnl${debugExt}-${tnlVersion} )
ENDIF( BUILD_CUDA )
-INSTALL( TARGETS tnl-euler-2d${debugExt}
+INSTALL( TARGETS tnl-euler${debugExt}
RUNTIME DESTINATION bin
PERMISSIONS OWNER_READ OWNER_WRITE OWNER_EXECUTE GROUP_READ GROUP_EXECUTE WORLD_READ WORLD_EXECUTE )
INSTALL( FILES run-euler
${tnl_inviscid_flow_SOURCES}
- DESTINATION share/tnl-${tnlVersion}/examples/inviscid-flow-2d )
+ DESTINATION share/tnl-${tnlVersion}/examples/inviscid-flow )
diff --git a/examples/inviscid-flow/LaxFridrichsContinuity.h b/examples/inviscid-flow/LaxFridrichsContinuity.h
index ebc7f8b07e5eb7df6ca819126effd68de8714420..657230695a5ec9cf84aaa84cd8b82d65e572df90 100644
--- a/examples/inviscid-flow/LaxFridrichsContinuity.h
+++ b/examples/inviscid-flow/LaxFridrichsContinuity.h
@@ -109,14 +109,14 @@ class LaxFridrichsContinuity< Meshes::Grid< 1, MeshReal, Device, MeshIndex >, Re
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 1, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 1, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 1, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
const RealType& velocity_x_west = this->velocity.template getData< DeviceType >()[ 0 ].template getData< DeviceType >()[ west ];
const RealType& velocity_x_east = this->velocity.template getData< DeviceType >()[ 0 ].template getData< DeviceType >()[ east ];
return 1.0 / 2.0 * this->tau * this->artificialViscosity * ( u[ west ] - 2.0 * u[ center ] + u[ east ] )
@@ -168,18 +168,18 @@ class LaxFridrichsContinuity< Meshes::Grid< 2, MeshReal, Device, MeshIndex >, Re
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 2, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 2, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 2, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
//rho
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
const RealType& velocity_x_west = this->velocity.template getData< DeviceType >()[ 0 ].template getData< DeviceType >()[ west ];
const RealType& velocity_x_east = this->velocity.template getData< DeviceType >()[ 0 ].template getData< DeviceType >()[ east ];
const RealType& velocity_y_north = this->velocity.template getData< DeviceType >()[ 1 ].template getData< DeviceType >()[ north ];
@@ -235,21 +235,21 @@ class LaxFridrichsContinuity< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, Re
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 3, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 3, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 3, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
//rho
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const RealType& hzInverse = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
const RealType& velocity_x_west = this->velocity.template getData< DeviceType >()[ 0 ].template getData< DeviceType >()[ west ];
const RealType& velocity_x_east = this->velocity.template getData< DeviceType >()[ 0 ].template getData< DeviceType >()[ east ];
diff --git a/examples/inviscid-flow/LaxFridrichsEnergy.h b/examples/inviscid-flow/LaxFridrichsEnergy.h
index fa0438f8c152007cbd7053d1b185abb3322801d0..d0af1de01db93d20a7ef800003b15b39228b9718 100644
--- a/examples/inviscid-flow/LaxFridrichsEnergy.h
+++ b/examples/inviscid-flow/LaxFridrichsEnergy.h
@@ -111,14 +111,14 @@ class LaxFridrichsEnergy< Meshes::Grid< 1, MeshReal, Device, MeshIndex >, Real,
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 1, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 1, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 1, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
const RealType& pressure_west = this->pressure.template getData< DeviceType >()[ west ];
const RealType& pressure_east = this->pressure.template getData< DeviceType >()[ east ];
const RealType& velocity_x_east = this->velocity.template getData< DeviceType >()[ 0 ].template getData< DeviceType >()[ east ];
@@ -176,17 +176,17 @@ class LaxFridrichsEnergy< Meshes::Grid< 2, MeshReal, Device, MeshIndex >, Real,
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 2, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 2, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 2, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
const RealType& pressure_west = this->pressure.template getData< DeviceType >()[ west ];
const RealType& pressure_east = this->pressure.template getData< DeviceType >()[ east ];
const RealType& pressure_north = this->pressure.template getData< DeviceType >()[ north ];
@@ -249,20 +249,20 @@ class LaxFridrichsEnergy< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, Real,
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 3, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 3, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 3, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const RealType& hzInverse = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
const RealType& pressure_west = this->pressure.template getData< DeviceType >()[ west ];
const RealType& pressure_east = this->pressure.template getData< DeviceType >()[ east ];
diff --git a/examples/inviscid-flow/LaxFridrichsMomentumX.h b/examples/inviscid-flow/LaxFridrichsMomentumX.h
index c63d155428392ed3090179f38f18ccb2fe938943..e054ad6ad261fbf53ef5c2e3a41475be3d992fd8 100644
--- a/examples/inviscid-flow/LaxFridrichsMomentumX.h
+++ b/examples/inviscid-flow/LaxFridrichsMomentumX.h
@@ -62,14 +62,14 @@ class LaxFridrichsMomentumX< Meshes::Grid< 1, MeshReal, Device, MeshIndex >, Rea
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 1, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 1, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 1, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
const RealType& pressure_west = this->pressure.template getData< DeviceType >()[ west ];
const RealType& pressure_east = this->pressure.template getData< DeviceType >()[ east ];
const RealType& velocity_x_east = this->velocity.template getData< DeviceType >()[ 0 ].template getData< DeviceType >()[ east ];
@@ -134,18 +134,18 @@ class LaxFridrichsMomentumX< Meshes::Grid< 2, MeshReal, Device, MeshIndex >, Rea
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 2, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 2, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 2, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
const RealType& pressure_west = this->pressure.template getData< DeviceType >()[ west ];
const RealType& pressure_east = this->pressure.template getData< DeviceType >()[ east ];
@@ -215,20 +215,20 @@ class LaxFridrichsMomentumX< Meshes::Grid< 3,MeshReal, Device, MeshIndex >, Real
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 3, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 3, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 3, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const RealType& hzInverse = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
const RealType& pressure_west = this->pressure.template getData< DeviceType >()[ west ];
const RealType& pressure_east = this->pressure.template getData< DeviceType >()[ east ];
diff --git a/examples/inviscid-flow/LaxFridrichsMomentumY.h b/examples/inviscid-flow/LaxFridrichsMomentumY.h
index 0a2847c6356b81cf443e945bd9698dfdcaee3221..ddf7b022c31831bde9c0c0d877883db63ec697cd 100644
--- a/examples/inviscid-flow/LaxFridrichsMomentumY.h
+++ b/examples/inviscid-flow/LaxFridrichsMomentumY.h
@@ -62,9 +62,9 @@ class LaxFridrichsMomentumY< Meshes::Grid< 1, MeshReal, Device, MeshIndex >, Rea
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 1, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 1, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 1, "Wrong preimage function" );
- //const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ //const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
return 0.0;
}
@@ -123,18 +123,18 @@ class LaxFridrichsMomentumY< Meshes::Grid< 2, MeshReal, Device, MeshIndex >, Rea
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 2, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 2, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 2, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
const RealType& pressure_north = this->pressure.template getData< DeviceType >()[ north ];
const RealType& pressure_south = this->pressure.template getData< DeviceType >()[ south ];
@@ -204,20 +204,20 @@ class LaxFridrichsMomentumY< Meshes::Grid< 3,MeshReal, Device, MeshIndex >, Real
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 3, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 3, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 3, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const RealType& hzInverse = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
const RealType& pressure_north = this->pressure.template getData< DeviceType >()[ north ];
const RealType& pressure_south = this->pressure.template getData< DeviceType >()[ south ];
diff --git a/examples/inviscid-flow/LaxFridrichsMomentumZ.h b/examples/inviscid-flow/LaxFridrichsMomentumZ.h
index a28e18156f5dff70c25111c8b6d73fc73d89e667..7f9213c4d512b79202734f10033d648a0b8cdeeb 100644
--- a/examples/inviscid-flow/LaxFridrichsMomentumZ.h
+++ b/examples/inviscid-flow/LaxFridrichsMomentumZ.h
@@ -62,9 +62,9 @@ class LaxFridrichsMomentumZ< Meshes::Grid< 1, MeshReal, Device, MeshIndex >, Rea
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 1, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 1, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 1, "Wrong preimage function" );
- //const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ //const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
return 0.0;
}
@@ -123,9 +123,9 @@ class LaxFridrichsMomentumZ< Meshes::Grid< 2, MeshReal, Device, MeshIndex >, Rea
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 2, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 2, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 2, "Wrong preimage function" );
- //const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ //const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
return 0.0;
}
@@ -184,20 +184,20 @@ class LaxFridrichsMomentumZ< Meshes::Grid< 3,MeshReal, Device, MeshIndex >, Real
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 3, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 3, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 3, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const RealType& hzInverse = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
const RealType& pressure_up = this->pressure.template getData< DeviceType >()[ up ];
const RealType& pressure_down = this->pressure.template getData< DeviceType >()[ down ];
diff --git a/examples/inviscid-flow/eulerRhs.h b/examples/inviscid-flow/eulerRhs.h
index 97dc9f07b391dc52020094be9d9eed279cbd1562..51d4e024398d579f49c158292e2890536a1e319c 100644
--- a/examples/inviscid-flow/eulerRhs.h
+++ b/examples/inviscid-flow/eulerRhs.h
@@ -6,7 +6,7 @@
namespace TNL {
template< typename Mesh, typename Real >class eulerRhs
- : public Functions::Domain< Mesh::meshDimension, Functions::MeshDomain >
+ : public Functions::Domain< Mesh::getMeshDimension(), Functions::MeshDomain >
{
public:
diff --git a/examples/mean-curvature-flow/CMakeLists.txt b/examples/mean-curvature-flow/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/examples/mean-curvature-flow/tnl-mean-curvature-flow-eoc.h b/examples/mean-curvature-flow/tnl-mean-curvature-flow-eoc.h
index f978f811af95fb7d5f66150e95cf8b8acff103e0..4d852a7d8358666d66b4482232649616cb9b5a32 100644
--- a/examples/mean-curvature-flow/tnl-mean-curvature-flow-eoc.h
+++ b/examples/mean-curvature-flow/tnl-mean-curvature-flow-eoc.h
@@ -70,7 +70,7 @@ class meanCurvatureFlowEocSetter
typedef Index IndexType;
typedef typename MeshType::PointType Point;
- enum { Dimension = MeshType::meshDimension };
+ enum { Dimension = MeshType::getMeshDimension() };
static bool run( const Config::ParameterContainer& parameters )
{
@@ -79,9 +79,9 @@ class meanCurvatureFlowEocSetter
typedef FiniteVolumeNonlinearOperator<MeshType, OperatorQ, Real, Index > NonlinearOperator;
typedef NonlinearDiffusion< MeshType, NonlinearOperator, Real, Index > ApproximateOperator;
typedef ExactNonlinearDiffusion< ExactGradientNorm< Dimension >, Dimension > ExactOperator;
- typedef TestFunction< MeshType::meshDimension, Real, Device > TestFunction;
+ typedef TestFunction< MeshType::getMeshDimension(), Real, Device > TestFunction;
typedef MeanCurvatureFlowEocRhs< ExactOperator, TestFunction, Dimension > RightHandSide;
- typedef StaticVector < MeshType::meshDimension, Real > Point;
+ typedef StaticVector < MeshType::getMeshDimension(), Real > Point;
typedef DirichletBoundaryConditions< MeshType, TestFunction, Dimension, Real, Index > BoundaryConditions;
typedef MeanCurvatureFlowEocProblem< MeshType, BoundaryConditions, RightHandSide, ApproximateOperator > Solver;
SolverStarter solverStarter;
diff --git a/examples/mean-curvature-flow/tnl-mean-curvature-flow.h b/examples/mean-curvature-flow/tnl-mean-curvature-flow.h
index 6a28b84210db9432d0b302b9f7f09aee6b160869..f2abd1671010c14207818909c0ca4c6e5e82b792 100644
--- a/examples/mean-curvature-flow/tnl-mean-curvature-flow.h
+++ b/examples/mean-curvature-flow/tnl-mean-curvature-flow.h
@@ -71,7 +71,7 @@ class meanCurvatureFlowSetter
typedef Index IndexType;
typedef typename MeshType::PointType Point;
- enum { Dimension = MeshType::meshDimension };
+ enum { Dimension = MeshType::getMeshDimension() };
static bool run( const Config::ParameterContainer& parameters )
{
@@ -102,7 +102,7 @@ class meanCurvatureFlowSetter
{
typedef OneSidedNonlinearDiffusion< MeshType, NonlinearOperator, Real, Index > ApproximateOperator;
typedef Constant< Dimension, Real > RightHandSide;
- typedef StaticVector< MeshType::meshDimension, Real > Point;
+ typedef StaticVector< MeshType::getMeshDimension(), Real > Point;
String boundaryConditionsType = parameters.getParameter< String >( "boundary-conditions-type" );
if( parameters.checkParameter( "boundary-conditions-constant" ) )
diff --git a/examples/narrow-band/CMakeLists.txt b/examples/narrow-band/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/examples/narrow-band/main.h b/examples/narrow-band/main.h
index 47d8648bbf7646cd9580134f55be6002359cf21e..e9f55656f1128eb268214eb85a4bc7fe26ec5773 100644
--- a/examples/narrow-band/main.h
+++ b/examples/narrow-band/main.h
@@ -47,26 +47,26 @@ int main( int argc, char* argv[] )
if(dim == 2)
{
- tnlNarrowBand<tnlGrid<2,double,tnlHost, int>, double, int> solver;
+ tnlNarrowBand<tnlGrid<2,double,TNL::Devices::Host, int>, double, int> solver;
if(!solver.init(parameters))
{
cerr << "Solver failed to initialize." << endl;
return EXIT_FAILURE;
}
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cout << "-------------------------------------------------------------" << endl;
cout << "Starting solver..." << endl;
solver.run();
}
// else if(dim == 3)
// {
-// tnlNarrowBand<tnlGrid<3,double,tnlHost, int>, double, int> solver;
+// tnlNarrowBand<tnlGrid<3,double,TNL::Devices::Host, int>, double, int> solver;
// if(!solver.init(parameters))
// {
// cerr << "Solver failed to initialize." << endl;
// return EXIT_FAILURE;
// }
-// checkCudaDevice;
+// TNL_CHECK_CUDA_DEVICE;
// cout << "-------------------------------------------------------------" << endl;
// cout << "Starting solver..." << endl;
// solver.run();
diff --git a/examples/narrow-band/tnlNarrowBand.h b/examples/narrow-band/tnlNarrowBand.h
index e2817424a1feb0f24bcbf8a759d37e8c9697977d..7d3d19bc03b43247f735cf04c5c82368360a748d 100644
--- a/examples/narrow-band/tnlNarrowBand.h
+++ b/examples/narrow-band/tnlNarrowBand.h
@@ -17,10 +17,10 @@
#define TNLNARROWBAND_H_
#include <TNL/Config/ParameterContainer.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
#include <TNL/Containers/StaticVector.h>
#include <functions/tnlMeshFunction.h>
-#include <core/tnlHost.h>
+#include <TNL/Devices/Host.h>
#include <mesh/tnlGrid.h>
#include <mesh/grids/tnlGridEntity.h>
#include <limits.h>
@@ -55,7 +55,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -135,7 +135,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
tnlNarrowBand();
diff --git a/examples/narrow-band/tnlNarrowBand2D_CUDA_v4_impl.h b/examples/narrow-band/tnlNarrowBand2D_CUDA_v4_impl.h
index e532bca7774c9c42648bf43f648a07ebbf20eb78..310d0fb239260fcc740a0f2bc56b37a36662a2f6 100644
--- a/examples/narrow-band/tnlNarrowBand2D_CUDA_v4_impl.h
+++ b/examples/narrow-band/tnlNarrowBand2D_CUDA_v4_impl.h
@@ -159,7 +159,7 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
int n = Mesh.getDimensions().x();
@@ -168,17 +168,17 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
dim3 numBlocks2(statusGridSize ,statusGridSize);
initSetupGridCUDA<<<numBlocks2,threadsPerBlock2>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initSetupGrid2CUDA<<<numBlocks2,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
/*dim3 threadsPerBlock(16, 16);
dim3 numBlocks(n/16 + 1 ,n/16 +1);*/
initCUDA<<<numBlocks2,threadsPerBlock2>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cout << "Solver initialized." << endl;
@@ -209,7 +209,7 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: r
cout << "Hi!" << endl;
runCUDA<<<numBlocksFS,threadsPerBlockFS>>>(this->cudaSolver,0,0);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cout << "Hi2!" << endl;
while(time < finalTime)
{
@@ -218,30 +218,30 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: r
runNarrowBandCUDA<<<numBlocksNB,threadsPerBlockNB>>>(this->cudaSolver,tau);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
time += tau;
cudaMemcpy(&reinit, this->reinitialize, sizeof(int), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
if(reinit != 0 /*&& time != finalTime */)
{
cout << time << endl;
initSetupGridCUDA<<<numBlocksNB,threadsPerBlockNB>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initSetupGrid2CUDA<<<numBlocksNB,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initCUDA<<<numBlocksNB,threadsPerBlockNB>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
runCUDA<<<numBlocksFS,threadsPerBlockFS>>>(this->cudaSolver,0,0);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
}
@@ -277,31 +277,31 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: u
int subgridID = i/NARROWBAND_SUBGRID_SIZE + (j/NARROWBAND_SUBGRID_SIZE) * statusGridSize;
if(cudaStatusVector[subgridID] != 0 && i<Mesh.getDimensions().x() && j < Mesh.getDimensions().y())
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real value = cudaDofVector2[Entity.getIndex()];
Real a,b, tmp;
if( i == 0 /*|| (i/NARROWBAND_SUBGRID_SIZE == 0 && !(cudaStatusVector[subgridID] & 9))*/ )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()];
else if( i == Mesh.getDimensions().x() - 1 /*|| (i/NARROWBAND_SUBGRID_SIZE == NARROWBAND_SUBGRID_SIZE - 1 && !(cudaStatusVector[subgridID] & 17))*/ )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
else
{
- a = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] );
+ a = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] );
}
if( j == 0 /*|| (j/NARROWBAND_SUBGRID_SIZE == 0 && !(cudaStatusVector[subgridID] & 3))*/ )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()];
else if( j == Mesh.getDimensions().y() - 1 /* || (j/NARROWBAND_SUBGRID_SIZE == NARROWBAND_SUBGRID_SIZE - 1 && !(cudaStatusVector[subgridID] & 5)) */)
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
else
{
- b = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] );
+ b = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] );
}
@@ -317,7 +317,7 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: u
}
-__global__ void initCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
@@ -346,11 +346,11 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
int i = threadIdx.x + blockDim.x*blockIdx.x;
int j = blockDim.y*blockIdx.y + threadIdx.y;
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
int gid = Entity.getIndex();
@@ -394,18 +394,18 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
// {
// if(cudaDofVector[Entity.getIndex()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1111(i,j);
// else
// setupSquare1110(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1101(i,j);
// else
// setupSquare1100(i,j);
@@ -413,16 +413,16 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1011(i,j);
// else
// setupSquare1010(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1001(i,j);
// else
// setupSquare1000(i,j);
@@ -431,18 +431,18 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0111(i,j);
// else
// setupSquare0110(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0101(i,j);
// else
// setupSquare0100(i,j);
@@ -450,16 +450,16 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0011(i,j);
// else
// setupSquare0010(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0001(i,j);
// else
// setupSquare0000(i,j);
@@ -497,7 +497,7 @@ Real tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: f
-__global__ void runCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int i)
+__global__ void runCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i)
{
@@ -603,7 +603,7 @@ __global__ void runCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double
-__global__ void initSetupGridCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initSetupGridCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
__shared__ double u0;
int gx = threadIdx.x + blockDim.x*blockIdx.x;
@@ -641,7 +641,7 @@ __global__ void initSetupGridCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int
// run this with one thread per block
-__global__ void initSetupGrid2CUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initSetupGrid2CUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
// printf("Hello\n");
if(solver->cudaStatusVector[blockIdx.x + gridDim.x*blockIdx.y] == 1)
@@ -670,7 +670,7 @@ __global__ void initSetupGrid2CUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, in
-__global__ void runNarrowBandCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, double tau)
+__global__ void runNarrowBandCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, double tau)
{
int gid = (blockDim.y*blockIdx.y + threadIdx.y)*solver->Mesh.getDimensions().x()+ threadIdx.x;
int i = threadIdx.x + blockIdx.x*blockDim.x;
@@ -688,44 +688,44 @@ __global__ void runNarrowBandCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int
if(status != 0)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(solver->Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(solver->Mesh);
Entity.setCoordinates(Containers::StaticVector<2,double>(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
double value = solver->cudaDofVector2[Entity.getIndex()];
double xf,xb,yf,yb, grad, fu, a,b;
a = b = 0.0;
if( i == 0 || (threadIdx.x == 0 && !(status & 9)) )
{
- xb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()];
- xf = solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] - value;
+ xb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()];
+ xf = solver->cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] - value;
}
else if( i == solver->Mesh.getDimensions().x() - 1 || (threadIdx.x == blockDim.x - 1 && !(status & 17)) )
{
- xb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
- xf = solver->cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()] - value;
+ xb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
+ xf = solver->cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()] - value;
}
else
{
- xb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
- xf = solver-> cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] - value;
+ xb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
+ xf = solver-> cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] - value;
}
if( j == 0 || (threadIdx.y == 0 && !(status & 3)) )
{
- yb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] ;
- yf = solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] - value;
+ yb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] ;
+ yf = solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] - value;
}
else if( j == solver->Mesh.getDimensions().y() - 1 || (threadIdx.y == blockDim.y - 1 && !(status & 5)) )
{
- yb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
- yf = solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()] - value;
+ yb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
+ yf = solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()] - value;
}
else
{
- yb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
- yf = solver-> cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] - value;
+ yb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
+ yf = solver-> cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] - value;
}
__syncthreads();
@@ -837,14 +837,14 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1111( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(INT_MAX,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -856,14 +856,14 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0000( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(-INT_MAX,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -875,18 +875,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1110( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -895,9 +895,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -908,18 +908,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1101( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -928,9 +928,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -941,18 +941,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1011( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -961,9 +961,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -974,17 +974,17 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0111( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
be=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
a = be/al;
@@ -994,9 +994,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1008,18 +1008,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0001( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -1028,9 +1028,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1041,18 +1041,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0010( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -1061,9 +1061,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1074,18 +1074,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0100( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -1094,9 +1094,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1107,17 +1107,17 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1000( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
be=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
a = be/al;
@@ -1127,9 +1127,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1144,18 +1144,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1100( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -1164,9 +1164,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1177,18 +1177,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1010( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -1197,9 +1197,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1210,14 +1210,14 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1001( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(cudaDofVector[Entity.getIndex()],cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1234,18 +1234,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0011( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -1254,9 +1254,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1267,18 +1267,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0101( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -1287,9 +1287,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1300,14 +1300,14 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0110( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(cudaDofVector[Entity.getIndex()],cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
#endif
diff --git a/examples/narrow-band/tnlNarrowBand2D_CUDA_v5_impl.h b/examples/narrow-band/tnlNarrowBand2D_CUDA_v5_impl.h
index 2d1296058391a0685d9385047469acb41a40b765..cb41d572674bd468cfa0286e274e630dd56fdfa6 100644
--- a/examples/narrow-band/tnlNarrowBand2D_CUDA_v5_impl.h
+++ b/examples/narrow-band/tnlNarrowBand2D_CUDA_v5_impl.h
@@ -158,7 +158,7 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
int n = Mesh.getDimensions().x();
@@ -167,17 +167,17 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
dim3 numBlocks2(statusGridSize ,statusGridSize);
initSetupGridCUDA<<<numBlocks2,threadsPerBlock2>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initSetupGrid2CUDA<<<numBlocks2,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
/*dim3 threadsPerBlock(16, 16);
dim3 numBlocks(n/16 + 1 ,n/16 +1);*/
initCUDA<<<numBlocks2,threadsPerBlock2>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cout << "Solver initialized." << endl;
@@ -208,7 +208,7 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: r
cout << "Hi!" << endl;
runCUDA<<<numBlocksFS,threadsPerBlockFS>>>(this->cudaSolver,0,0);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cout << "Hi2!" << endl;
while(time < finalTime)
{
@@ -217,30 +217,30 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: r
runNarrowBandCUDA<<<numBlocksNB,threadsPerBlockNB>>>(this->cudaSolver,tau);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
time += tau;
cudaMemcpy(&reinit, this->reinitialize, sizeof(int), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
if(reinit != 0 /*&& time != finalTime */)
{
cout << time << endl;
initSetupGridCUDA<<<numBlocksNB,threadsPerBlockNB>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initSetupGrid2CUDA<<<numBlocksNB,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initCUDA<<<numBlocksNB,threadsPerBlockNB>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
runCUDA<<<numBlocksFS,threadsPerBlockFS>>>(this->cudaSolver,0,0);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
}
@@ -276,31 +276,31 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: u
int subgridID = i/NARROWBAND_SUBGRID_SIZE + (j/NARROWBAND_SUBGRID_SIZE) * ((Mesh.getDimensions().x() + NARROWBAND_SUBGRID_SIZE-1 ) / NARROWBAND_SUBGRID_SIZE);
if(/*cudaStatusVector[subgridID] != 0 &&*/ i<Mesh.getDimensions().x() && Mesh.getDimensions().y())
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real value = cudaDofVector2[Entity.getIndex()];
Real a,b, tmp;
if( i == 0 /*|| (i/NARROWBAND_SUBGRID_SIZE == 0 && !(cudaStatusVector[subgridID] & 9)) */)
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()];
else if( i == Mesh.getDimensions().x() - 1 /*|| (i/NARROWBAND_SUBGRID_SIZE == NARROWBAND_SUBGRID_SIZE - 1 && !(cudaStatusVector[subgridID] & 17)) */)
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
else
{
- a = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] );
+ a = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] );
}
if( j == 0/* || (j/NARROWBAND_SUBGRID_SIZE == 0 && !(cudaStatusVector[subgridID] & 3)) */)
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()];
else if( j == Mesh.getDimensions().y() - 1 /* || (j/NARROWBAND_SUBGRID_SIZE == NARROWBAND_SUBGRID_SIZE - 1 && !(cudaStatusVector[subgridID] & 5))*/ )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
else
{
- b = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] );
+ b = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] );
}
@@ -316,7 +316,7 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: u
}
-__global__ void initCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
@@ -345,11 +345,11 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
int i = threadIdx.x + blockDim.x*blockIdx.x;
int j = blockDim.y*blockIdx.y + threadIdx.y;
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
int gid = Entity.getIndex();
@@ -392,18 +392,18 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
// {
// if(cudaDofVector[Entity.getIndex()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1111(i,j);
// else
// setupSquare1110(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1101(i,j);
// else
// setupSquare1100(i,j);
@@ -411,16 +411,16 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1011(i,j);
// else
// setupSquare1010(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1001(i,j);
// else
// setupSquare1000(i,j);
@@ -429,18 +429,18 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0111(i,j);
// else
// setupSquare0110(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0101(i,j);
// else
// setupSquare0100(i,j);
@@ -448,16 +448,16 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0011(i,j);
// else
// setupSquare0010(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0001(i,j);
// else
// setupSquare0000(i,j);
@@ -495,7 +495,7 @@ Real tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: f
-__global__ void runCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int i)
+__global__ void runCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i)
{
@@ -601,7 +601,7 @@ __global__ void runCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double
-__global__ void initSetupGridCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initSetupGridCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
__shared__ double u0;
int gx = threadIdx.x + blockDim.x*blockIdx.x;
@@ -639,7 +639,7 @@ __global__ void initSetupGridCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int
// run this with one thread per block
-__global__ void initSetupGrid2CUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initSetupGrid2CUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
// printf("Hello\n");
if(solver->cudaStatusVector[blockIdx.x + gridDim.x*blockIdx.y] == 1)
@@ -666,7 +666,7 @@ __global__ void initSetupGrid2CUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, in
-__global__ void runNarrowBandCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, double tau)
+__global__ void runNarrowBandCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, double tau)
{
int gid = (blockDim.y*blockIdx.y + threadIdx.y)*solver->Mesh.getDimensions().x()+ threadIdx.x;
int i = threadIdx.x + blockIdx.x*blockDim.x;
@@ -684,44 +684,44 @@ __global__ void runNarrowBandCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int
// if(status != 0)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(solver->Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(solver->Mesh);
Entity.setCoordinates(Containers::StaticVector<2,double>(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
double value = solver->cudaDofVector2[Entity.getIndex()];
double xf,xb,yf,yb, grad, fu, a,b;
a = b = 0.0;
if( i == 0 /*|| (threadIdx.x == 0 && !(status & 9)) */)
{
- xb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()];
- xf = solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] - value;
+ xb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()];
+ xf = solver->cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] - value;
}
else if( i == solver->Mesh.getDimensions().x() - 1 /*|| (threadIdx.x == blockDim.x - 1 && !(status & 17)) */)
{
- xb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
- xf = solver->cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()] - value;
+ xb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
+ xf = solver->cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()] - value;
}
else
{
- xb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
- xf = solver-> cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] - value;
+ xb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
+ xf = solver-> cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] - value;
}
if( j == 0/* || (threadIdx.y == 0 && !(status & 3))*/ )
{
- yb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] ;
- yf = solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] - value;
+ yb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] ;
+ yf = solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] - value;
}
else if( j == solver->Mesh.getDimensions().y() - 1 /*|| (threadIdx.y == blockDim.y - 1 && !(status & 5)) */)
{
- yb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
- yf = solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()] - value;
+ yb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
+ yf = solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()] - value;
}
else
{
- yb = value - solver->cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
- yf = solver-> cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] - value;
+ yb = value - solver->cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
+ yf = solver-> cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] - value;
}
__syncthreads();
@@ -833,14 +833,14 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1111( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(INT_MAX,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -852,14 +852,14 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0000( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(-INT_MAX,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -871,18 +871,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1110( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -891,9 +891,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -904,18 +904,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1101( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -924,9 +924,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -937,18 +937,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1011( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -957,9 +957,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -970,17 +970,17 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0111( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
be=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
a = be/al;
@@ -990,9 +990,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1004,18 +1004,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0001( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -1024,9 +1024,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1037,18 +1037,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0010( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -1057,9 +1057,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1070,18 +1070,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0100( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -1090,9 +1090,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1103,17 +1103,17 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1000( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
be=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
a = be/al;
@@ -1123,9 +1123,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1140,18 +1140,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1100( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -1160,9 +1160,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1173,18 +1173,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1010( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -1193,9 +1193,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1206,14 +1206,14 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1001( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(cudaDofVector[Entity.getIndex()],cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1230,18 +1230,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0011( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -1250,9 +1250,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1263,18 +1263,18 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0101( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -1283,9 +1283,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1296,14 +1296,14 @@ template< typename MeshReal,
typename Index >
void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0110( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(cudaDofVector[Entity.getIndex()],cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
#endif
diff --git a/examples/narrow-band/tnlNarrowBand2D_impl.h b/examples/narrow-band/tnlNarrowBand2D_impl.h
index 8248baa0949862de6ebcee797e6916c9634de72e..f1afb9ab8f7d12f1639d5a08d64edf8143f47083 100644
--- a/examples/narrow-band/tnlNarrowBand2D_impl.h
+++ b/examples/narrow-band/tnlNarrowBand2D_impl.h
@@ -93,7 +93,7 @@ template< typename MeshReal,
bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: initGrid()
{
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
for(int i=0; i< Mesh.getDimensions().x()*Mesh.getDimensions().x();i++)
{
dofVector2[i]=INT_MAX*sign(dofVector[i]);
@@ -105,22 +105,22 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- neighbourEntities.refresh(Mesh,Entity.getIndex());
+ neighborEntities.refresh(Mesh,Entity.getIndex());
if(dofVector[this->Entity.getIndex()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1111(i,j);
else
setupSquare1110(i,j);
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1101(i,j);
else
setupSquare1100(i,j);
@@ -128,16 +128,16 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1011(i,j);
else
setupSquare1010(i,j);
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1001(i,j);
else
setupSquare1000(i,j);
@@ -146,18 +146,18 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0111(i,j);
else
setupSquare0110(i,j);
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0101(i,j);
else
setupSquare0100(i,j);
@@ -165,16 +165,16 @@ bool tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: i
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0011(i,j);
else
setupSquare0010(i,j);
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0001(i,j);
else
setupSquare0000(i,j);
@@ -397,29 +397,29 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: u
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real value = dofVector2[Entity.getIndex()];
Real a,b, tmp;
if( i == 0 )
- a = dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ a = dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()];
else if( i == Mesh.getDimensions().x() - 1 )
- a = dofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ a = dofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
else
{
- a = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()],
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] );
+ a = fabsMin( dofVector2[neighborEntities.template getEntityIndex< -1, 0 >()],
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] );
}
if( j == 0 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()];
else if( j == Mesh.getDimensions().y() - 1 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
else
{
- b = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()],
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] );
+ b = fabsMin( dofVector2[neighborEntities.template getEntityIndex< 0, -1 >()],
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] );
}
@@ -466,11 +466,11 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
// this->Entity.setCoordinates(CoordinatesType(i,j));
// this->Entity.refresh();
-// auto neighbourEntities = Entity.getNeighbourEntities();
+// auto neighborEntities = Entity.getNeighborEntities();
// dofVector2[Entity.getIndex()]=fabsMin(INT_MAX,dofVector2[Entity.getIndex()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -484,11 +484,11 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
// this->Entity.setCoordinates(CoordinatesType(i,j));
// this->Entity.refresh();
-// auto neighbourEntities = Entity.getNeighbourEntities();
+// auto neighborEntities = Entity.getNeighborEntities();
// dofVector2[Entity.getIndex()]=fabsMin(-INT_MAX,dofVector2[(Entity.getIndex())]);
-// dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -502,15 +502,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -519,9 +519,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -534,15 +534,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -551,9 +551,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -566,15 +566,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -583,9 +583,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -598,14 +598,14 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
be=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
a = be/al;
@@ -615,9 +615,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -631,15 +631,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -648,9 +648,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -663,15 +663,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -680,9 +680,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -695,15 +695,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -712,9 +712,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -727,14 +727,14 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
be=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
a = be/al;
@@ -744,9 +744,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -763,15 +763,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -780,9 +780,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -795,15 +795,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -812,9 +812,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -827,11 +827,11 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
dofVector2[Entity.getIndex()]=fabsMin(dofVector[Entity.getIndex()],dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -850,15 +850,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -867,9 +867,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -882,15 +882,15 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -899,9 +899,9 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -914,11 +914,11 @@ void tnlNarrowBand< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: s
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
dofVector2[Entity.getIndex()]=fabsMin(dofVector[Entity.getIndex()],dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
diff --git a/examples/narrow-band/tnlNarrowBand3D_CUDA_impl.h b/examples/narrow-band/tnlNarrowBand3D_CUDA_impl.h
index ac8a023d7d93060f63858892c0541f5eda0ee706..2b96ad58205d0578776b2b2d645eb71292c0fad4 100644
--- a/examples/narrow-band/tnlNarrowBand3D_CUDA_impl.h
+++ b/examples/narrow-band/tnlNarrowBand3D_CUDA_impl.h
@@ -111,10 +111,10 @@ bool tnlNarrowBand< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > :: i
dim3 numBlocks(n/8 + 1, n/8 +1, n/8 +1);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
return true;
}
@@ -139,7 +139,7 @@ bool tnlNarrowBand< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > :: r
runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,0,0);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cudaMemcpy(this->dofVector.getData().getData(), cudaDofVector2, this->dofVector.getData().getSize()*sizeof(double), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
@@ -165,41 +165,41 @@ template< typename MeshReal,
__device__
void tnlNarrowBand< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > :: updateValue( Index i, Index j, Index k)
{
- tnlGridEntity< tnlGrid< 3,double, tnlHost, int >, 3, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 3,double, TNL::Devices::Host, int >, 3, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j,k));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities(Entity);
Real value = cudaDofVector2[Entity.getIndex()];
Real a,b,c, tmp;
if( i == 0 )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0, 0 >()];
else if( i == Mesh.getDimensions().x() - 1 )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0, 0 >()];
else
{
- a = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0, 0 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0, 0 >()] );
+ a = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0, 0 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0, 0 >()] );
}
if( j == 0 )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1, 0 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1, 0 >()];
else if( j == Mesh.getDimensions().y() - 1 )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1, 0 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1, 0 >()];
else
{
- b = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1, 0 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1, 0 >()] );
+ b = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1, 0 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1, 0 >()] );
}
if( k == 0 )
- c = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 0, 1 >()];
+ c = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 0, 1 >()];
else if( k == Mesh.getDimensions().z() - 1 )
- c = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 0, -1 >()];
+ c = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 0, -1 >()];
else
{
- c = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 0, -1 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 0, 1 >()] );
+ c = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< 0, 0, -1 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 0, 1 >()] );
}
Real hD = 3.0*h*h - 2.0*(a*a + b*b + c*c - a*b - a*c - b*c);
@@ -222,7 +222,7 @@ template< typename MeshReal,
__device__
bool tnlNarrowBand< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > :: initGrid(int i, int j, int k)
{
- tnlGridEntity< tnlGrid< 3,double, tnlHost, int >, 3, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 3,double, TNL::Devices::Host, int >, 3, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j,k));
Entity.refresh();
int gid = Entity.getIndex();
@@ -255,7 +255,7 @@ Real tnlNarrowBand< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > :: f
-__global__ void runCUDA(tnlNarrowBand< tnlGrid< 3,double, tnlHost, int >, double, int >* solver, int sweep, int i)
+__global__ void runCUDA(tnlNarrowBand< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i)
{
int gx = 0;
@@ -474,7 +474,7 @@ __global__ void runCUDA(tnlNarrowBand< tnlGrid< 3,double, tnlHost, int >, double
}
-__global__ void initCUDA(tnlNarrowBand< tnlGrid< 3,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlNarrowBand< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver)
{
int gx = threadIdx.x + blockDim.x*blockIdx.x;
int gy = blockDim.y*blockIdx.y + threadIdx.y;
diff --git a/examples/narrow-band/tnlNarrowBand3D_impl.h b/examples/narrow-band/tnlNarrowBand3D_impl.h
index eb446d9d2831967e9016b3eaaf326baf6751ab7c..33d7ef8cd2c3a19992760ed15dc732b6da7f42d8 100644
--- a/examples/narrow-band/tnlNarrowBand3D_impl.h
+++ b/examples/narrow-band/tnlNarrowBand3D_impl.h
@@ -237,38 +237,38 @@ void tnlNarrowBand< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > :: u
{
this->Entity.setCoordinates(CoordinatesType(i,j,k));
this->Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities(Entity);
Real value = dofVector2[Entity.getIndex()];
Real a,b,c, tmp;
if( i == 0 )
- a = dofVector2[neighbourEntities.template getEntityIndex< 1, 0, 0>()];
+ a = dofVector2[neighborEntities.template getEntityIndex< 1, 0, 0>()];
else if( i == Mesh.getDimensions().x() - 1 )
- a = dofVector2[neighbourEntities.template getEntityIndex< -1, 0, 0 >()];
+ a = dofVector2[neighborEntities.template getEntityIndex< -1, 0, 0 >()];
else
{
- a = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< -1, 0, 0>()],
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0, 0>()] );
+ a = fabsMin( dofVector2[neighborEntities.template getEntityIndex< -1, 0, 0>()],
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0, 0>()] );
}
if( j == 0 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, 1, 0>()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, 1, 0>()];
else if( j == Mesh.getDimensions().y() - 1 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, -1, 0>()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, -1, 0>()];
else
{
- b = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< 0, -1, 0>()],
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1, 0>()] );
+ b = fabsMin( dofVector2[neighborEntities.template getEntityIndex< 0, -1, 0>()],
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1, 0>()] );
}
if( k == 0 )
- c = dofVector2[neighbourEntities.template getEntityIndex< 0, 0, 1>()];
+ c = dofVector2[neighborEntities.template getEntityIndex< 0, 0, 1>()];
else if( k == Mesh.getDimensions().z() - 1 )
- c = dofVector2[neighbourEntities.template getEntityIndex< 0, 0, -1>()];
+ c = dofVector2[neighborEntities.template getEntityIndex< 0, 0, -1>()];
else
{
- c = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< 0, 0, -1>()],
- dofVector2[neighbourEntities.template getEntityIndex< 0, 0, 1>()] );
+ c = fabsMin( dofVector2[neighborEntities.template getEntityIndex< 0, 0, -1>()],
+ dofVector2[neighborEntities.template getEntityIndex< 0, 0, 1>()] );
}
Real hD = 3.0*h*h - 2.0*(a*a+b*b+c*c-a*b-a*c-b*c);
diff --git a/examples/narrow-band/tnlNarrowBand_CUDA.h b/examples/narrow-band/tnlNarrowBand_CUDA.h
index 8da92f5fc570275c85ebba69113f0e59a4be7988..ca9b1da2cc6e26b14bc003532b6eea75e89d907d 100644
--- a/examples/narrow-band/tnlNarrowBand_CUDA.h
+++ b/examples/narrow-band/tnlNarrowBand_CUDA.h
@@ -17,9 +17,9 @@
#define TNLNARROWBAND_H_
#include <TNL/Config/ParameterContainer.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
#include <TNL/Containers/StaticVector.h>
-#include <core/tnlHost.h>
+#include <TNL/Devices/Host.h>
#include <mesh/tnlGrid.h>
#include <mesh/grids/tnlGridEntity.h>
@@ -54,7 +54,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
tnlNarrowBand();
@@ -138,7 +138,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -183,16 +183,16 @@ protected:
#ifdef HAVE_CUDA
//template<int sweep_t>
-__global__ void runCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int i);
-//__global__ void runCUDA(tnlNarrowBand< tnlGrid< 3,double, tnlHost, int >, double, int >* solver, int sweep, int i);
+__global__ void runCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i);
+//__global__ void runCUDA(tnlNarrowBand< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i);
-__global__ void initCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver);
+__global__ void initCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver);
-__global__ void initSetupGridCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver);
-__global__ void initSetupGrid2CUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver);
-__global__ void initSetupGrid1_2CUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver);
-__global__ void runNarrowBandCUDA(tnlNarrowBand< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, double tau);
-//__global__ void initCUDA(tnlNarrowBand< tnlGrid< 3,double, tnlHost, int >, double, int >* solver);
+__global__ void initSetupGridCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver);
+__global__ void initSetupGrid2CUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver);
+__global__ void initSetupGrid1_2CUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver);
+__global__ void runNarrowBandCUDA(tnlNarrowBand< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, double tau);
+//__global__ void initCUDA(tnlNarrowBand< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver);
#endif
diff --git a/examples/navier-stokes/CMakeLists.txt b/examples/navier-stokes/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/examples/navier-stokes/share/CMakeLists.txt b/examples/navier-stokes/share/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/examples/navier-stokes/share/examples/CMakeLists.txt b/examples/navier-stokes/share/examples/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/examples/quad-test/main.cpp b/examples/quad-test/main.cpp
index dbaf0114448391b9c80744882e9fe7cb86ab4cca..583d3c4ca8470297d320ec305a99e96f6ef547d7 100644
--- a/examples/quad-test/main.cpp
+++ b/examples/quad-test/main.cpp
@@ -34,7 +34,7 @@ int main(int argc, char* argv[]) {
String inputFile = parameters.getParameter <String> ("input-file");
File binaryFile;
- if(! binaryFile.open(inputFile, tnlReadMode)) {
+ if(! binaryFile.open(inputFile, IOMode::read)) {
cerr << "I am not able to open the file " << inputFile << "." << std::endl;
return 1;
}
@@ -48,4 +48,4 @@ int main(int argc, char* argv[]) {
CSR <QuadDouble> quadMatrix("quad");
quadMatrix = doubleMatrix;
return EXIT_SUCCESS;
-}
\ No newline at end of file
+}
diff --git a/examples/transport-equation/transportEquationProblem.h b/examples/transport-equation/transportEquationProblem.h
index a9fc1482075afd51b4e5490dae2c67c3572e451e..6b0dd73f6353c826f5d1c6aee598dacb228b1c71 100644
--- a/examples/transport-equation/transportEquationProblem.h
+++ b/examples/transport-equation/transportEquationProblem.h
@@ -96,6 +96,11 @@ public PDEProblem< Mesh,
DofVectorPointer& rightHandSide,
MeshDependentDataPointer& meshDependentData );
+ template< typename Matrix >
+ void saveFailedLinearSystem( const Matrix& matrix,
+ const DofVectorType& dofs,
+ const DofVectorType& rightHandSide ) const;
+
protected:
MeshFunctionPointer uPointer;
diff --git a/examples/transport-equation/transportEquationProblem_impl.h b/examples/transport-equation/transportEquationProblem_impl.h
index 698a7bc687dc4c3b4df49a48fef2f5a73566ecbc..f647802eef18fc9073fdc9b439aa345048d54ea2 100644
--- a/examples/transport-equation/transportEquationProblem_impl.h
+++ b/examples/transport-equation/transportEquationProblem_impl.h
@@ -223,4 +223,17 @@ assemblyLinearSystem( const RealType& time,
{
}
+template< typename Mesh,
+ typename BoundaryCondition,
+ typename RightHandSide,
+ typename DifferentialOperator >
+ template< typename Matrix >
+void
+transportEquationProblem< Mesh, BoundaryCondition, RightHandSide, DifferentialOperator >::
+saveFailedLinearSystem( const Matrix& matrix,
+ const DofVectorType& dofs,
+ const DofVectorType& rightHandSide ) const
+{
+}
+
} // namespace TNL
diff --git a/install b/install
index 139fe67d3b2cd3b5bec3d973a1d336e071ea5182..6ce0b8a60e9d230b73947c65f2a8ef1e2e767fe1 100755
--- a/install
+++ b/install
@@ -35,10 +35,8 @@ then
mkdir Debug
fi
cd Debug
- if ../build --root-dir=.. --build=Debug ${OPTIONS};
+ if ! ../build --root-dir=.. --build=Debug --install=yes ${OPTIONS}
then
- make install
- else
exit 1
fi
cd ..
@@ -51,10 +49,8 @@ then
mkdir Release
fi
cd Release
- if ../build --root-dir=.. --build=Release ${OPTIONS};
+ if ! ../build --root-dir=.. --build=Release --install=yes ${OPTIONS};
then
- make install
- else
exit 1
fi
cd ..
diff --git a/share/CMakeLists.txt b/share/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/share/Tools/CMakeLists.txt b/share/Tools/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 9499c8dfd96ca922535eacf1358fb863762147da..75215d9ec394fea0d12748a53d9bec7bb9e227a9 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -1,4 +1,4 @@
ADD_SUBDIRECTORY( Python )
ADD_SUBDIRECTORY( TNL )
ADD_SUBDIRECTORY( Tools )
-ADD_SUBDIRECTORY( UnitTests )
\ No newline at end of file
+ADD_SUBDIRECTORY( UnitTests )
diff --git a/src/TNL/Assert.h b/src/TNL/Assert.h
index f8f60200d0bdfdc4910334174bb7b9ecd3a63977..ce1f7d10bb05d6f192fbee4dd280d63f9f425537 100644
--- a/src/TNL/Assert.h
+++ b/src/TNL/Assert.h
@@ -2,7 +2,7 @@
Assert.h - description
-------------------
begin : Jan 12, 2010
- copyright : (C) 2013 by Tomas Oberhuber
+ copyright : (C) 2013 by Tomas Oberhuber et al.
email : tomas.oberhuber@fjfi.cvut.cz
***************************************************************************/
@@ -11,51 +11,370 @@
#pragma once
/****
- * Debugging assert
+ * The purpose of this file is to define the TNL_ASSERT_* debugging macros as
+ * shown below.
+ *
+ * If the 'NDEBUG' macro is defined, the build is considered to be optimized
+ * and all assert macros are empty. Otherwise, the conditions are checked and
+ * failures lead to the diagnostics message being printed to std::cerr and
+ * program abortion (via 'throw EXIT_FAILURE' statement).
+ *
+ * For the purpose of providing Python bindings it is possible to change the
+ * reporting behaviour by defining the TNL_THROW_ASSERTION_ERROR macro, which
+ * leads to throwing the ::TNL::Assert::AssertionError holding the error
+ * message (which is not printed in this case). The AssertionError class does
+ * not inherit from std::exception to avoid being caught by normal exception
+ * handlers, but the code for Python bindings can use it to translate it to the
+ * Python's AssertionError exception.
+ *
+ * Implemented by: Jakub Klinkovsky
*/
-#ifndef NDEBUG
+#if defined(NDEBUG) || defined(HAVE_MIC)
+// empty macros for optimized build
+#define TNL_ASSERT_TRUE( val, msg )
+#define TNL_ASSERT_FALSE( val, msg )
+#define TNL_ASSERT_EQ( val1, val2, msg )
+#define TNL_ASSERT_NE( val1, val2, msg )
+#define TNL_ASSERT_LE( val1, val2, msg )
+#define TNL_ASSERT_LT( val1, val2, msg )
+#define TNL_ASSERT_GE( val1, val2, msg )
+#define TNL_ASSERT_GT( val1, val2, msg )
+#define TNL_ASSERT( ___tnl__assert_condition, ___tnl__assert_command )
+
+#else /* #ifdef NDEBUG */
+
+#include <sstream>
#include <iostream>
-#include <stdlib.h>
-#include <assert.h>
+#include <stdio.h>
+
+#include <TNL/Devices/CudaCallable.h>
+
+namespace TNL {
+namespace Assert {
+
+#ifdef TNL_THROW_ASSERTION_ERROR
+// This will be used by the code for Python bindings to translate assertion
+// failures to the Python's AssertionError exception.
+class AssertionError
+{
+public:
+ AssertionError( const std::string& msg )
+ : msg( msg )
+ {}
+
+ const char* what() const
+ {
+ return msg.c_str();
+ }
+
+private:
+ std::string msg;
+};
+
+inline void
+printDiagnosticsHost( const char* assertion,
+ const char* message,
+ const char* file,
+ const char* function,
+ int line,
+ const char* diagnostics )
+{
+ std::stringstream str;
+ str << "Assertion '" << assertion << "' failed !!!\n"
+ << "Message: " << message << "\n"
+ << "File: " << file << "\n"
+ << "Function: " << function << "\n"
+ << "Line: " << line << "\n"
+ << "Diagnostics:\n" << diagnostics << std::endl;
+ throw AssertionError( str.str() );
+}
+
+#else // TNL_THROW_ASSERTION_ERROR
+
+// This will be used in regular C++ code
+inline void
+printDiagnosticsHost( const char* assertion,
+ const char* message,
+ const char* file,
+ const char* function,
+ int line,
+ const char* diagnostics )
+{
+ std::cerr << "Assertion '" << assertion << "' failed !!!\n"
+ << "Message: " << message << "\n"
+ << "File: " << file << "\n"
+ << "Function: " << function << "\n"
+ << "Line: " << line << "\n"
+ << "Diagnostics:\n" << diagnostics << std::endl;
+}
+#endif // TNL_THROW_ASSERTION_ERROR
+__cuda_callable__
+inline void
+printDiagnosticsCuda( const char* assertion,
+ const char* message,
+ const char* file,
+ const char* function,
+ int line,
+ const char* diagnostics )
+{
+ printf( "Assertion '%s' failed !!!\n"
+ "Message: %s\n"
+ "File: %s\n"
+ "Function: %s\n"
+ "Line: %d\n"
+ "Diagnostics: %s\n",
+ assertion, message, file, function, line, diagnostics );
+}
+
+__cuda_callable__
+inline void
+fatalFailure()
+{
+#ifdef __CUDA_ARCH__
+ // https://devtalk.nvidia.com/default/topic/509584/how-to-cancel-a-running-cuda-kernel-/
+ // TODO: it is reported as "illegal instruction", but that leads to an abort as well...
+ asm("trap;");
+#else
+ throw EXIT_FAILURE;
#endif
+}
-#ifndef NDEBUG
-
+template< typename T >
+struct Formatter
+{
+ static std::string
+ printToString( const T& value )
+ {
+ ::std::stringstream ss;
+ ss << value;
+ return ss.str();
+ }
+};
+
+template<>
+struct Formatter< bool >
+{
+ static std::string
+ printToString( const bool& value )
+ {
+ if( value ) return "true";
+ else return "false";
+ }
+};
+
+template< typename T1, typename T2 >
+__cuda_callable__ void
+cmpHelperOpFailure( const char* assertion,
+ const char* message,
+ const char* file,
+ const char* function,
+ int line,
+ const char* lhs_expression,
+ const char* rhs_expression,
+ const T1& lhs_value,
+ const T2& rhs_value,
+ const char* op )
+{
+#ifdef __CUDA_ARCH__
+ // diagnostics is not supported - we don't have the machinery
+ // to construct the dynamic error message
+ printDiagnosticsCuda( assertion, message, file, function, line,
+ "Not supported in CUDA kernels." );
+#else
+ const std::string formatted_lhs_value = Formatter< T1 >::printToString( lhs_value );
+ const std::string formatted_rhs_value = Formatter< T2 >::printToString( rhs_value );
+ std::stringstream str;
+ if( std::string(op) == "==" ) {
+ str << " Expected: " << lhs_expression;
+ if( formatted_lhs_value != lhs_expression ) {
+ str << "\n Which is: " << formatted_lhs_value;
+ }
+ str << "\nTo be equal to: " << rhs_expression;
+ if( formatted_rhs_value != rhs_expression ) {
+ str << "\n Which is: " << formatted_rhs_value;
+ }
+ str << std::endl;
+ }
+ else {
+ str << "Expected: (" << lhs_expression << ") " << op << " (" << rhs_expression << "), "
+ << "actual: " << formatted_lhs_value << " vs " << formatted_rhs_value << std::endl;
+ }
+ printDiagnosticsHost( assertion, message, file, function, line,
+ str.str().c_str() );
+#endif
+ fatalFailure();
+}
+
+template< typename T1, typename T2 >
+__cuda_callable__ void
+cmpHelperTrue( const char* assertion,
+ const char* message,
+ const char* file,
+ const char* function,
+ int line,
+ const char* expr1,
+ const char* expr2,
+ const T1& val1,
+ const T2& val2 )
+{
+ // explicit cast is necessary, because T1::operator! might not be defined
+ if( ! (bool) val1 )
+ ::TNL::Assert::cmpHelperOpFailure( assertion, message, file, function, line,
+ expr1, "true", val1, true, "==" );
+}
+
+template< typename T1, typename T2 >
+__cuda_callable__ void
+cmpHelperFalse( const char* assertion,
+ const char* message,
+ const char* file,
+ const char* function,
+ int line,
+ const char* expr1,
+ const char* expr2,
+ const T1& val1,
+ const T2& val2 )
+{
+ if( val1 )
+ ::TNL::Assert::cmpHelperOpFailure( assertion, message, file, function, line,
+ expr1, "false", val1, false, "==" );
+}
+
+// A macro for implementing the helper functions needed to implement
+// TNL_ASSERT_??. It is here just to avoid copy-and-paste of similar code.
+#define TNL_IMPL_CMP_HELPER_( op_name, op ) \
+template< typename T1, typename T2 > \
+__cuda_callable__ void \
+cmpHelper##op_name( const char* assertion, \
+ const char* message, \
+ const char* file, \
+ const char* function, \
+ int line, \
+ const char* expr1, \
+ const char* expr2, \
+ const T1& val1, \
+ const T2& val2 ) \
+{\
+ if( ! ( (val1) op (val2) ) ) \
+ ::TNL::Assert::cmpHelperOpFailure( assertion, message, file, function, line, \
+ expr1, expr2, val1, val2, #op );\
+}
+
+// Implements the helper function for TNL_ASSERT_EQ
+TNL_IMPL_CMP_HELPER_( EQ, == );
+// Implements the helper function for TNL_ASSERT_NE
+TNL_IMPL_CMP_HELPER_( NE, != );
+// Implements the helper function for TNL_ASSERT_LE
+TNL_IMPL_CMP_HELPER_( LE, <= );
+// Implements the helper function for TNL_ASSERT_LT
+TNL_IMPL_CMP_HELPER_( LT, < );
+// Implements the helper function for TNL_ASSERT_GE
+TNL_IMPL_CMP_HELPER_( GE, >= );
+// Implements the helper function for TNL_ASSERT_GT
+TNL_IMPL_CMP_HELPER_( GT, > );
+
+#undef TNL_IMPL_CMP_HELPER_
+
+} // namespace Assert
+} // namespace TNL
+
+// Internal macro wrapping the __PRETTY_FUNCTION__ "magic".
#if defined( __NVCC__ ) && ( __CUDACC_VER__ < 80000 )
- #define TNL_PRETTY_FUNCTION "(not known in CUDA 7.5 or older)"
+ #define __TNL_PRETTY_FUNCTION "(not known in CUDA 7.5 or older)"
#else
- #define TNL_PRETTY_FUNCTION __PRETTY_FUNCTION__
+ #define __TNL_PRETTY_FUNCTION __PRETTY_FUNCTION__
#endif
+// Internal macro to compose the string representing the assertion.
+// We can't do it easily at runtime, because we have to support assertions
+// in CUDA kernels, which can't use std::string objects. Instead, we do it
+// at compile time - adjacent strings are joined at the language level.
+#define __TNL_JOIN_STRINGS( val1, op, val2 ) \
+ __STRING( val1 ) " " __STRING( op ) " " __STRING( val2 )
+
+// Internal macro to pass all the arguments to the specified cmpHelperOP
+#define __TNL_ASSERT_PRED2( pred, op, val1, val2, msg ) \
+ pred( __TNL_JOIN_STRINGS( val1, op, val2 ), \
+ msg, __FILE__, __TNL_PRETTY_FUNCTION, __LINE__, \
+ #val1, #val2, val1, val2 )
+
+// Main definitions of the TNL_ASSERT_* macros
+// unary
+#define TNL_ASSERT_TRUE( val, msg ) \
+ __TNL_ASSERT_PRED2( ::TNL::Assert::cmpHelperTrue, ==, val, true, msg )
+#define TNL_ASSERT_FALSE( val, msg ) \
+ __TNL_ASSERT_PRED2( ::TNL::Assert::cmpHelperFalse, ==, val, false, msg )
+// binary
+#define TNL_ASSERT_EQ( val1, val2, msg ) \
+ __TNL_ASSERT_PRED2( ::TNL::Assert::cmpHelperEQ, ==, val1, val2, msg )
+#define TNL_ASSERT_NE( val1, val2, msg ) \
+ __TNL_ASSERT_PRED2( ::TNL::Assert::cmpHelperNE, !=, val1, val2, msg )
+#define TNL_ASSERT_LE( val1, val2, msg ) \
+ __TNL_ASSERT_PRED2( ::TNL::Assert::cmpHelperLE, <=, val1, val2, msg )
+#define TNL_ASSERT_LT( val1, val2, msg ) \
+ __TNL_ASSERT_PRED2( ::TNL::Assert::cmpHelperLT, <, val1, val2, msg )
+#define TNL_ASSERT_GE( val1, val2, msg ) \
+ __TNL_ASSERT_PRED2( ::TNL::Assert::cmpHelperGE, >=, val1, val2, msg )
+#define TNL_ASSERT_GT( val1, val2, msg ) \
+ __TNL_ASSERT_PRED2( ::TNL::Assert::cmpHelperGT, >, val1, val2, msg )
+
+
+
+
+/****
+ * Original assert macro with custom command for diagnostics.
+ */
+
// __CUDA_ARCH__ is defined by the compiler only for code executed on GPU
#ifdef __CUDA_ARCH__
-#define TNL_ASSERT( ___tnl__assert_condition, ___tnl__assert_command ) \
- if( ! ( ___tnl__assert_condition ) ) \
- { \
- printf( "Assertion '%s' failed !!! \n File: %s \n Line: %d \n Diagnostics: Not supported with CUDA.\n", \
- __STRING( ___tnl__assert_condition ), \
- __FILE__, \
- __LINE__ ); \
- \
+#define TNL_ASSERT( ___tnl__assert_condition, ___tnl__assert_command ) \
+ if( ! ( ___tnl__assert_condition ) ) \
+ { \
+ printf( "Assertion '%s' failed !!! \n File: %s \n Line: %d \n Diagnostics: Not supported with CUDA.\n", \
+ __STRING( ___tnl__assert_condition ), \
+ __FILE__, \
+ __LINE__ ); \
+ asm("trap;"); \
}
-#else // __CUDA_ARCH__
-#define TNL_ASSERT( ___tnl__assert_condition, ___tnl__assert_command ) \
- if( ! ( ___tnl__assert_condition ) ) \
- { \
- std::cerr << "Assertion '" << __STRING( ___tnl__assert_condition ) << "' failed !!!" << std::endl \
- << "File: " << __FILE__ << std::endl \
- << "Function: " << TNL_PRETTY_FUNCTION << std::endl \
- << "Line: " << __LINE__ << std::endl \
- << "Diagnostics: "; \
- ___tnl__assert_command; \
- throw EXIT_FAILURE; \
+#else // #ifdef __CUDA_ARCH__
+#ifdef TNL_THROW_ASSERTION_ERROR
+// This will be used by the code for Python bindings to translate assertion
+// failures to the Python's AssertionError exception.
+#define TNL_ASSERT( ___tnl__assert_condition, ___tnl__assert_command ) \
+ if( ! ( ___tnl__assert_condition ) ) \
+ { \
+ std::stringstream buffer; \
+ auto old = std::cerr.rdbuf( buffer.rdbuf() ); \
+ \
+ std::cerr << "Assertion '" << __STRING( ___tnl__assert_condition ) << "' failed !!!" << std::endl \
+ << "File: " << __FILE__ << std::endl \
+ << "Function: " << __PRETTY_FUNCTION__ << std::endl \
+ << "Line: " << __LINE__ << std::endl \
+ << "Diagnostics: "; \
+ ___tnl__assert_command; \
+ \
+ std::string msg = buffer.str(); \
+ std::cerr.rdbuf( old ); \
+ throw ::TNL::Assert::AssertionError( msg ); \
}
-#endif // __CUDA_ARCH__
+#else // #ifdef TNL_THROW_ASSERTION_ERROR
+// This will be used in regular C++ code
+#define TNL_ASSERT( ___tnl__assert_condition, ___tnl__assert_command ) \
+ if( ! ( ___tnl__assert_condition ) ) \
+ { \
+ std::cerr << "Assertion '" << __STRING( ___tnl__assert_condition ) << "' failed !!!" << std::endl \
+ << "File: " << __FILE__ << std::endl \
+ << "Function: " << __TNL_PRETTY_FUNCTION << std::endl \
+ << "Line: " << __LINE__ << std::endl \
+ << "Diagnostics: "; \
+ ___tnl__assert_command; \
+ throw EXIT_FAILURE; \
+ }
+#endif // #ifdef TNL_THROW_ASSERTION_ERROR
+#endif // #ifdef __CUDA_ARCH__
-#else /* #ifndef NDEBUG */
-#define TNL_ASSERT( ___tnl__assert_condition, ___tnl__assert_command )
-#endif /* #ifndef NDEBUG */
+#endif // #ifdef NDEBUG
\ No newline at end of file
diff --git a/src/TNL/CMakeLists.txt b/src/TNL/CMakeLists.txt
old mode 100755
new mode 100644
index 957120fdac7d916dad1309131e8fb714224ee7a3..3bdfaca909298fc323cbcf493c5aabd986e60af0
--- a/src/TNL/CMakeLists.txt
+++ b/src/TNL/CMakeLists.txt
@@ -2,6 +2,7 @@ ADD_SUBDIRECTORY( Config )
ADD_SUBDIRECTORY( Containers )
ADD_SUBDIRECTORY( Debugging )
ADD_SUBDIRECTORY( Devices )
+ADD_SUBDIRECTORY( Exceptions )
ADD_SUBDIRECTORY( Experimental )
ADD_SUBDIRECTORY( Functions )
ADD_SUBDIRECTORY( Images )
@@ -30,6 +31,7 @@ set( headers
Logger_impl.h
Math.h
mpi-supp.h
+ ParallelFor.h
param-types.h
SharedPointer.h
SmartPointer.h
@@ -90,6 +92,8 @@ if( BUILD_CUDA )
else( BUILD_CUDA )
ADD_LIBRARY( tnl${debugExt}-${tnlVersion} SHARED
${tnl_SOURCES} )
+#ifMIC
+ #TARGET_COMPILE_DEFINITIONS( tnl${debugExt}-${tnlVersion} PUBLIC -DHAVE_MIC )
endif( BUILD_CUDA )
SET_TARGET_PROPERTIES( tnl${debugExt}-${tnlVersion} PROPERTIES
diff --git a/src/TNL/Config/CMakeLists.txt b/src/TNL/Config/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Config/ParameterContainer.h b/src/TNL/Config/ParameterContainer.h
index af57fa14a4bb07d248750f69c4927b4e357aec06..5daeb07861d0844ebaf6c4ead8efa5e47b02e53e 100644
--- a/src/TNL/Config/ParameterContainer.h
+++ b/src/TNL/Config/ParameterContainer.h
@@ -128,7 +128,7 @@ setParameter( const String& name,
{
if( parameters[ i ] -> name == name )
{
- if( parameters[ i ] -> type == getType( value ) )
+ if( parameters[ i ] -> type == TNL::getType< T >() )
{
( ( tnlParameter< T > * ) parameters[ i ] ) -> value = value;
return true;
@@ -137,7 +137,7 @@ setParameter( const String& name,
{
std::cerr << "Parameter " << name << " already exists with different type "
<< parameters[ i ] -> type << " not "
- << getType( value ) << std::endl;
+ << TNL::getType< T >() << std::endl;
abort( );
return false;
}
diff --git a/src/TNL/Containers/Algorithms/ArrayOperations.h b/src/TNL/Containers/Algorithms/ArrayOperations.h
index e32c7fd288c098ae2e648c1ad2fdbc7a6ccfc324..4861cc460262ec74cad2a35400eb472cf2e551dc 100644
--- a/src/TNL/Containers/Algorithms/ArrayOperations.h
+++ b/src/TNL/Containers/Algorithms/ArrayOperations.h
@@ -12,6 +12,7 @@
#include <TNL/Devices/Host.h>
#include <TNL/Devices/Cuda.h>
+#include <TNL/Devices/MIC.h>
namespace TNL {
namespace Containers {
@@ -27,11 +28,11 @@ class ArrayOperations< Devices::Host >
public:
template< typename Element, typename Index >
- static bool allocateMemory( Element*& data,
+ static void allocateMemory( Element*& data,
const Index size );
template< typename Element >
- static bool freeMemory( Element* data );
+ static void freeMemory( Element* data );
template< typename Element >
static void setMemoryElement( Element* data,
@@ -73,11 +74,11 @@ class ArrayOperations< Devices::Cuda >
public:
template< typename Element, typename Index >
- static bool allocateMemory( Element*& data,
+ static void allocateMemory( Element*& data,
const Index size );
template< typename Element >
- static bool freeMemory( Element* data );
+ static void freeMemory( Element* data );
template< typename Element >
static void setMemoryElement( Element* data,
@@ -86,6 +87,7 @@ class ArrayOperations< Devices::Cuda >
template< typename Element >
static Element getMemoryElement( const Element* data );
+ // TODO: does not make sense for CUDA - remove?
template< typename Element, typename Index >
static Element& getArrayElementReference( Element* data, const Index i );
@@ -152,9 +154,96 @@ class ArrayOperations< Devices::Host, Devices::Cuda >
const Index size );
};
+
+template<>
+class ArrayOperations< Devices::MIC >
+{
+ public:
+
+ template< typename Element, typename Index >
+ static void allocateMemory( Element*& data,
+ const Index size );
+
+ template< typename Element >
+ static void freeMemory( Element* data );
+
+ template< typename Element >
+ static void setMemoryElement( Element* data,
+ const Element& value );
+
+ template< typename Element >
+ static Element getMemoryElement( const Element* data );
+
+ template< typename Element, typename Index >
+ static Element& getArrayElementReference( Element* data, const Index i );
+
+ template< typename Element, typename Index >
+ static const Element& getArrayElementReference( const Element* data, const Index i );
+
+ template< typename Element, typename Index >
+ static bool setMemory( Element* data,
+ const Element& value,
+ const Index size );
+
+ template< typename DestinationElement,
+ typename SourceElement,
+ typename Index >
+ static bool copyMemory( DestinationElement* destination,
+ const SourceElement* source,
+ const Index size );
+
+ template< typename Element1,
+ typename Element2,
+ typename Index >
+ static bool compareMemory( const Element1* destination,
+ const Element2* source,
+ const Index size );
+};
+
+template<>
+class ArrayOperations< Devices::MIC, Devices::Host >
+{
+ public:
+
+ template< typename DestinationElement,
+ typename SourceElement,
+ typename Index >
+ static bool copyMemory( DestinationElement* destination,
+ const SourceElement* source,
+ const Index size );
+
+ template< typename DestinationElement,
+ typename SourceElement,
+ typename Index >
+ static bool compareMemory( const DestinationElement* destination,
+ const SourceElement* source,
+ const Index size );
+};
+
+template<>
+class ArrayOperations< Devices::Host, Devices::MIC >
+{
+ public:
+
+ template< typename DestinationElement,
+ typename SourceElement,
+ typename Index >
+ static bool copyMemory( DestinationElement* destination,
+ const SourceElement* source,
+ const Index size );
+
+ template< typename DestinationElement,
+ typename SourceElement,
+ typename Index >
+ static bool compareMemory( const DestinationElement* destination,
+ const SourceElement* source,
+ const Index size );
+};
+
} // namespace Algorithms
} // namespace Containers
} // namespace TNL
#include <TNL/Containers/Algorithms/ArrayOperationsHost_impl.h>
#include <TNL/Containers/Algorithms/ArrayOperationsCuda_impl.h>
+#include <TNL/Containers/Algorithms/ArrayOperationsMIC_impl.h>
diff --git a/src/TNL/Containers/Algorithms/ArrayOperationsCuda_impl.h b/src/TNL/Containers/Algorithms/ArrayOperationsCuda_impl.h
index 74bf0a73005e86015b443d4c99a6274d39ab2077..1465c6250bbcd2510b8ceb50a86f7bd77fc02c6c 100644
--- a/src/TNL/Containers/Algorithms/ArrayOperationsCuda_impl.h
+++ b/src/TNL/Containers/Algorithms/ArrayOperationsCuda_impl.h
@@ -14,6 +14,8 @@
#include <TNL/tnlConfig.h>
#include <TNL/Math.h>
+#include <TNL/Exceptions/CudaSupportMissing.h>
+#include <TNL/Exceptions/CudaBadAlloc.h>
#include <TNL/Containers/Algorithms/ArrayOperations.h>
#include <TNL/Containers/Algorithms/Reduction.h>
#include <TNL/Containers/Algorithms/reduction-operations.h>
@@ -23,36 +25,37 @@ namespace Containers {
namespace Algorithms {
template< typename Element, typename Index >
-bool
+void
ArrayOperations< Devices::Cuda >::
allocateMemory( Element*& data,
const Index size )
{
#ifdef HAVE_CUDA
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
if( cudaMalloc( ( void** ) &data,
( size_t ) size * sizeof( Element ) ) != cudaSuccess )
+ {
data = 0;
- return checkCudaDevice;
+ throw Exceptions::CudaBadAlloc();
+ }
+ TNL_CHECK_CUDA_DEVICE;
#else
- CudaSupportMissingMessage;
- return false;
+ throw Exceptions::CudaSupportMissing();
#endif
}
template< typename Element >
-bool
+void
ArrayOperations< Devices::Cuda >::
freeMemory( Element* data )
{
- TNL_ASSERT( data, );
+ TNL_ASSERT_TRUE( data, "Attempted to free a nullptr." );
#ifdef HAVE_CUDA
- checkCudaDevice;
- cudaFree( data );
- return checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
+ cudaFree( data );
+ TNL_CHECK_CUDA_DEVICE;
#else
- CudaSupportMissingMessage;;
- return true;
+ throw Exceptions::CudaSupportMissing();
#endif
}
@@ -62,7 +65,7 @@ ArrayOperations< Devices::Cuda >::
setMemoryElement( Element* data,
const Element& value )
{
- TNL_ASSERT( data, );
+ TNL_ASSERT_TRUE( data, "Attempted to set data through a nullptr." );
ArrayOperations< Devices::Cuda >::setMemory( data, value, 1 );
}
@@ -71,7 +74,7 @@ Element
ArrayOperations< Devices::Cuda >::
getMemoryElement( const Element* data )
{
- TNL_ASSERT( data, );
+ TNL_ASSERT_TRUE( data, "Attempted to get data through a nullptr." );
Element result;
ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< Element, Element, int >( &result, data, 1 );
return result;
@@ -82,7 +85,7 @@ Element&
ArrayOperations< Devices::Cuda >::
getArrayElementReference( Element* data, const Index i )
{
- TNL_ASSERT( data, );
+ TNL_ASSERT_TRUE( data, "Attempted to access data through a nullptr." );
return data[ i ];
}
@@ -91,7 +94,7 @@ const
Element& ArrayOperations< Devices::Cuda >::
getArrayElementReference( const Element* data, const Index i )
{
- TNL_ASSERT( data, );
+ TNL_ASSERT_TRUE( data, "Attempted to access data through a nullptr." );
return data[ i ];
}
@@ -120,17 +123,16 @@ setMemory( Element* data,
const Element& value,
const Index size )
{
- TNL_ASSERT( data, );
+ TNL_ASSERT_TRUE( data, "Attempted to set data through a nullptr." );
#ifdef HAVE_CUDA
dim3 blockSize( 0 ), gridSize( 0 );
blockSize. x = 256;
Index blocksNumber = ceil( ( double ) size / ( double ) blockSize. x );
gridSize. x = min( blocksNumber, Devices::Cuda::getMaxGridSize() );
setArrayValueCudaKernel<<< gridSize, blockSize >>>( data, size, value );
- return checkCudaDevice;
+ return TNL_CHECK_CUDA_DEVICE;
#else
- CudaSupportMissingMessage;;
- return false;
+ throw Exceptions::CudaSupportMissing();
#endif
}
@@ -162,30 +164,29 @@ copyMemory( DestinationElement* destination,
const SourceElement* source,
const Index size )
{
- TNL_ASSERT( destination, );
- TNL_ASSERT( source, );
- #ifdef HAVE_CUDA
- if( std::is_same< DestinationElement, SourceElement >::value )
- {
- if( cudaMemcpy( destination,
- source,
- size * sizeof( DestinationElement ),
- cudaMemcpyDeviceToDevice ) != cudaSuccess )
- return checkCudaDevice;
- }
- else
- {
- dim3 blockSize( 0 ), gridSize( 0 );
- blockSize. x = 256;
- Index blocksNumber = ceil( ( double ) size / ( double ) blockSize. x );
- gridSize. x = min( blocksNumber, Devices::Cuda::getMaxGridSize() );
- copyMemoryCudaToCudaKernel<<< gridSize, blockSize >>>( destination, source, size );
- return checkCudaDevice;
- }
- #else
- CudaSupportMissingMessage;;
- #endif
- return false;
+ TNL_ASSERT_TRUE( destination, "Attempted to copy data to a nullptr." );
+ TNL_ASSERT_TRUE( source, "Attempted to copy data from a nullptr." );
+#ifdef HAVE_CUDA
+ if( std::is_same< DestinationElement, SourceElement >::value )
+ {
+ cudaMemcpy( destination,
+ source,
+ size * sizeof( DestinationElement ),
+ cudaMemcpyDeviceToDevice );
+ return TNL_CHECK_CUDA_DEVICE;
+ }
+ else
+ {
+ dim3 blockSize( 0 ), gridSize( 0 );
+ blockSize. x = 256;
+ Index blocksNumber = ceil( ( double ) size / ( double ) blockSize. x );
+ gridSize. x = min( blocksNumber, Devices::Cuda::getMaxGridSize() );
+ copyMemoryCudaToCudaKernel<<< gridSize, blockSize >>>( destination, source, size );
+ return TNL_CHECK_CUDA_DEVICE;
+ }
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
}
template< typename Element1,
@@ -197,8 +198,8 @@ compareMemory( const Element1* destination,
const Element2* source,
const Index size )
{
- TNL_ASSERT( destination, );
- TNL_ASSERT( source, );
+ TNL_ASSERT_TRUE( destination, "Attempted to compare data through a nullptr." );
+ TNL_ASSERT_TRUE( source, "Attempted to compare data through a nullptr." );
//TODO: The parallel reduction on the CUDA device with different element types is needed.
bool result;
Algorithms::tnlParallelReductionEqualities< Element1, Index > reductionEqualities;
@@ -219,30 +220,21 @@ copyMemory( DestinationElement* destination,
const SourceElement* source,
const Index size )
{
- TNL_ASSERT( destination, );
- TNL_ASSERT( source, );
- #ifdef HAVE_CUDA
+ TNL_ASSERT_TRUE( destination, "Attempted to copy data to a nullptr." );
+ TNL_ASSERT_TRUE( source, "Attempted to copy data from a nullptr." );
+#ifdef HAVE_CUDA
if( std::is_same< DestinationElement, SourceElement >::value )
{
- cudaMemcpy( destination,
- source,
- size * sizeof( DestinationElement ),
- cudaMemcpyDeviceToHost );
- if( ! checkCudaDevice )
- {
+ if( cudaMemcpy( destination,
+ source,
+ size * sizeof( DestinationElement ),
+ cudaMemcpyDeviceToHost ) != cudaSuccess )
std::cerr << "Transfer of data from CUDA device to host failed." << std::endl;
- return false;
- }
- return true;
+ return TNL_CHECK_CUDA_DEVICE;
}
else
{
SourceElement* buffer = new SourceElement[ Devices::Cuda::getGPUTransferBufferSize() ];
- if( ! buffer )
- {
- std::cerr << "Unable to allocate supporting buffer to transfer data between the CUDA device and the host." << std::endl;
- return false;
- }
Index i( 0 );
while( i < size )
{
@@ -251,9 +243,9 @@ copyMemory( DestinationElement* destination,
min( size - i, Devices::Cuda::getGPUTransferBufferSize() ) * sizeof( SourceElement ),
cudaMemcpyDeviceToHost ) != cudaSuccess )
{
- checkCudaDevice;
delete[] buffer;
- return false;
+ std::cerr << "Transfer of data from CUDA device to host failed." << std::endl;
+ return TNL_CHECK_CUDA_DEVICE;
}
Index j( 0 );
while( j < Devices::Cuda::getGPUTransferBufferSize() && i + j < size )
@@ -265,11 +257,10 @@ copyMemory( DestinationElement* destination,
}
delete[] buffer;
}
- #else
- CudaSupportMissingMessage;;
- return false;
- #endif
return true;
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
}
@@ -285,16 +276,11 @@ compareMemory( const Element1* destination,
/***
* Here, destination is on host and source is on CUDA device.
*/
- TNL_ASSERT( destination, );
- TNL_ASSERT( source, );
- TNL_ASSERT( size >= 0, std::cerr << "size = " << size );
- #ifdef HAVE_CUDA
+ TNL_ASSERT_TRUE( destination, "Attempted to compare data through a nullptr." );
+ TNL_ASSERT_TRUE( source, "Attempted to compare data through a nullptr." );
+ TNL_ASSERT_GE( size, 0, "Array size must be non-negative." );
+#ifdef HAVE_CUDA
Element2* host_buffer = new Element2[ Devices::Cuda::getGPUTransferBufferSize() ];
- if( ! host_buffer )
- {
- std::cerr << "I am sorry but I cannot allocate supporting buffer on the host for comparing data between CUDA GPU and CPU." << std::endl;
- return false;
- }
Index compared( 0 );
while( compared < size )
{
@@ -304,10 +290,9 @@ compareMemory( const Element1* destination,
transfer * sizeof( Element2 ),
cudaMemcpyDeviceToHost ) != cudaSuccess )
{
- std::cerr << "Transfer of data from the device failed." << std::endl;
- checkCudaDevice;
delete[] host_buffer;
- return false;
+ std::cerr << "Transfer of data from CUDA device to host failed." << std::endl;
+ return TNL_CHECK_CUDA_DEVICE;
}
if( ! ArrayOperations< Devices::Host >::compareMemory( &destination[ compared ], host_buffer, transfer ) )
{
@@ -318,10 +303,9 @@ compareMemory( const Element1* destination,
}
delete[] host_buffer;
return true;
- #else
- CudaSupportMissingMessage;;
- return false;
- #endif
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
}
/****
@@ -336,31 +320,22 @@ copyMemory( DestinationElement* destination,
const SourceElement* source,
const Index size )
{
- TNL_ASSERT( destination, );
- TNL_ASSERT( source, );
- TNL_ASSERT( size >= 0, std::cerr << "size = " << size );
- #ifdef HAVE_CUDA
+ TNL_ASSERT_TRUE( destination, "Attempted to copy data to a nullptr." );
+ TNL_ASSERT_TRUE( source, "Attempted to copy data from a nullptr." );
+ TNL_ASSERT_GE( size, 0, "Array size must be non-negative." );
+#ifdef HAVE_CUDA
if( std::is_same< DestinationElement, SourceElement >::value )
{
- cudaMemcpy( destination,
- source,
- size * sizeof( DestinationElement ),
- cudaMemcpyHostToDevice );
- if( ! checkCudaDevice )
- {
+ if( cudaMemcpy( destination,
+ source,
+ size * sizeof( DestinationElement ),
+ cudaMemcpyHostToDevice ) != cudaSuccess )
std::cerr << "Transfer of data from host to CUDA device failed." << std::endl;
- return false;
- }
- return true;
+ return TNL_CHECK_CUDA_DEVICE;
}
else
{
DestinationElement* buffer = new DestinationElement[ Devices::Cuda::getGPUTransferBufferSize() ];
- if( ! buffer )
- {
- std::cerr << "Unable to allocate supporting buffer to transfer data between the CUDA device and the host." << std::endl;
- return false;
- }
Index i( 0 );
while( i < size )
{
@@ -375,19 +350,18 @@ copyMemory( DestinationElement* destination,
j * sizeof( DestinationElement ),
cudaMemcpyHostToDevice ) != cudaSuccess )
{
- checkCudaDevice;
delete[] buffer;
- return false;
+ std::cerr << "Transfer of data from host to CUDA device failed." << std::endl;
+ return TNL_CHECK_CUDA_DEVICE;
}
i += j;
}
delete[] buffer;
return true;
}
- #else
- CudaSupportMissingMessage;;
- return false;
- #endif
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
}
template< typename Element1,
@@ -399,9 +373,9 @@ compareMemory( const Element1* hostData,
const Element2* deviceData,
const Index size )
{
- TNL_ASSERT( hostData, );
- TNL_ASSERT( deviceData, );
- TNL_ASSERT( size >= 0, std::cerr << "size = " << size );
+ TNL_ASSERT_TRUE( hostData, "Attempted to compare data through a nullptr." );
+ TNL_ASSERT_TRUE( deviceData, "Attempted to compare data through a nullptr." );
+ TNL_ASSERT_GE( size, 0, "Array size must be non-negative." );
return ArrayOperations< Devices::Host, Devices::Cuda >::compareMemory( deviceData, hostData, size );
}
diff --git a/src/TNL/Containers/Algorithms/ArrayOperationsHost_impl.h b/src/TNL/Containers/Algorithms/ArrayOperationsHost_impl.h
index 8d6df347765c636f67bd6122801fa26971670501..7390cfdb0e317a036f45c1db7b58d09a18afac0f 100644
--- a/src/TNL/Containers/Algorithms/ArrayOperationsHost_impl.h
+++ b/src/TNL/Containers/Algorithms/ArrayOperationsHost_impl.h
@@ -21,24 +21,27 @@ namespace Containers {
namespace Algorithms {
template< typename Element, typename Index >
-bool
+void
ArrayOperations< Devices::Host >::
allocateMemory( Element*& data,
const Index size )
{
- if( ! ( data = new Element[ size ] ) )
- return false;
- return true;
+ data = new Element[ size ];
+ // According to the standard, new either throws, or returns non-nullptr.
+ // Some (old) compilers don't comply:
+ // https://stackoverflow.com/questions/550451/will-new-return-null-in-any-case
+ TNL_ASSERT_TRUE( data, "Operator 'new' returned a nullptr. This should never happen - there is "
+ "either a bug or the compiler does not comply to the standard." );
}
template< typename Element >
-bool
+void
ArrayOperations< Devices::Host >::
freeMemory( Element* data )
{
delete[] data;
- return true;
}
+
template< typename Element >
void
ArrayOperations< Devices::Host >::
@@ -95,7 +98,9 @@ copyMemory( DestinationElement* destination,
const SourceElement* source,
const Index size )
{
- if( std::is_same< DestinationElement, SourceElement >::value )
+ if( std::is_same< DestinationElement, SourceElement >::value &&
+ ( std::is_fundamental< DestinationElement >::value ||
+ std::is_pointer< DestinationElement >::value ) )
memcpy( destination, source, size * sizeof( DestinationElement ) );
else
for( Index i = 0; i < size; i ++ )
@@ -112,7 +117,9 @@ compareMemory( const DestinationElement* destination,
const SourceElement* source,
const Index size )
{
- if( std::is_same< DestinationElement, SourceElement >::value )
+ if( std::is_same< DestinationElement, SourceElement >::value &&
+ ( std::is_fundamental< DestinationElement >::value ||
+ std::is_pointer< DestinationElement >::value ) )
{
if( memcmp( destination, source, size * sizeof( DestinationElement ) ) != 0 )
return false;
diff --git a/src/TNL/Containers/Algorithms/ArrayOperationsMIC_impl.h b/src/TNL/Containers/Algorithms/ArrayOperationsMIC_impl.h
new file mode 100644
index 0000000000000000000000000000000000000000..754be26878ae64076ec52c61cd793ae132fd12b6
--- /dev/null
+++ b/src/TNL/Containers/Algorithms/ArrayOperationsMIC_impl.h
@@ -0,0 +1,444 @@
+/***************************************************************************
+ ArrayOperationsMIC_impl.h - description
+ -------------------
+ begin : Mar 4, 2017
+ copyright : (C) 2017 by Tomas Oberhuber
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Vit Hanousek
+
+#pragma once
+
+#include <iostream>
+
+#include <TNL/tnlConfig.h>
+#include <TNL/Math.h>
+#include <TNL/Exceptions/MICSupportMissing.h>
+#include <TNL/Exceptions/MICBadAlloc.h>
+#include <TNL/Containers/Algorithms/ArrayOperations.h>
+#include <TNL/Containers/Algorithms/Reduction.h>
+#include <TNL/Containers/Algorithms/reduction-operations.h>
+
+namespace TNL {
+namespace Containers {
+namespace Algorithms {
+
+static constexpr std::size_t MIC_STACK_VAR_LIM = 5*1024*1024;
+
+template< typename Element, typename Index >
+void
+ArrayOperations< Devices::MIC >::
+allocateMemory( Element*& data,
+ const Index size )
+{
+#ifdef HAVE_MIC
+ data = (Element*) Devices::MIC::AllocMIC( size * sizeof(Element) );
+ if( ! data )
+ throw Exceptions::MICBadAlloc();
+#else
+ throw Exceptions::MICSupportMissing();
+#endif
+}
+
+template< typename Element >
+void
+ArrayOperations< Devices::MIC >::
+freeMemory( Element* data )
+{
+ TNL_ASSERT( data, );
+#ifdef HAVE_MIC
+ Devices::MIC::FreeMIC( data );
+#else
+ throw Exceptions::MICSupportMissing();
+#endif
+}
+
+template< typename Element >
+void
+ArrayOperations< Devices::MIC >::
+setMemoryElement( Element* data,
+ const Element& value )
+{
+ TNL_ASSERT( data, );
+ ArrayOperations< Devices::MIC >::setMemory( data, value, 1 );
+}
+
+template< typename Element >
+Element
+ArrayOperations< Devices::MIC >::
+getMemoryElement( const Element* data )
+{
+ TNL_ASSERT( data, );
+ Element result;
+ ArrayOperations< Devices::Host, Devices::MIC >::copyMemory< Element, Element, int >( &result, data, 1 );
+ return result;
+}
+
+template< typename Element, typename Index >
+Element&
+ArrayOperations< Devices::MIC >::
+getArrayElementReference( Element* data, const Index i )
+{
+ TNL_ASSERT( data, );
+ return data[ i ];
+}
+
+template< typename Element, typename Index >
+const
+Element& ArrayOperations< Devices::MIC >::
+getArrayElementReference( const Element* data, const Index i )
+{
+ TNL_ASSERT( data, );
+ return data[ i ];
+}
+
+template< typename Element, typename Index >
+bool
+ArrayOperations< Devices::MIC >::
+setMemory( Element* data,
+ const Element& value,
+ const Index size )
+{
+ TNL_ASSERT( data, );
+#ifdef HAVE_MIC
+ Element tmp=value;
+ Devices::MICHider<Element> hide_ptr;
+ hide_ptr.pointer=data;
+ #pragma offload target(mic) in(hide_ptr,tmp,size)
+ {
+ Element * dst= hide_ptr.pointer;
+ for(int i=0;i<size;i++)
+ dst[i]=tmp;
+ }
+ return true;
+#else
+ throw Exceptions::MICSupportMissing();
+#endif
+}
+
+template< typename DestinationElement,
+ typename SourceElement,
+ typename Index >
+bool
+ArrayOperations< Devices::MIC >::
+copyMemory( DestinationElement* destination,
+ const SourceElement* source,
+ const Index size )
+{
+ TNL_ASSERT( destination, );
+ TNL_ASSERT( source, );
+ #ifdef HAVE_MIC
+ if( std::is_same< DestinationElement, SourceElement >::value )
+ {
+ Devices::MICHider<void> src_ptr;
+ src_ptr.pointer=(void*)source;
+ Devices::MICHider<void> dst_ptr;
+ dst_ptr.pointer=(void*)destination;
+ #pragma offload target(mic) in(src_ptr,dst_ptr,size)
+ {
+ memcpy(dst_ptr.pointer,src_ptr.pointer,size*sizeof(DestinationElement));
+ }
+ return true;
+ }
+ else
+ {
+ Devices::MICHider<const SourceElement> src_ptr;
+ src_ptr.pointer=source;
+ Devices::MICHider<DestinationElement> dst_ptr;
+ dst_ptr.pointer=destination;
+ #pragma offload target(mic) in(src_ptr,dst_ptr,size)
+ {
+ for(int i=0;i<size;i++)
+ dst_ptr.pointer[i]=src_ptr.pointer[i];
+ }
+ return true;
+
+ }
+ #else
+ throw Exceptions::MICSupportMissing();
+ #endif
+ return false;
+}
+
+template< typename Element1,
+ typename Element2,
+ typename Index >
+bool
+ArrayOperations< Devices::MIC >::
+compareMemory( const Element1* destination,
+ const Element2* source,
+ const Index size )
+{
+ TNL_ASSERT( destination, );
+ TNL_ASSERT( source, );
+#ifdef HAVE_MIC
+ if( std::is_same< Element1, Element2 >::value )
+ {
+ Devices::MICHider<void> src_ptr;
+ src_ptr.pointer=(void*)source;
+ Devices::MICHider<void> dst_ptr;
+ dst_ptr.pointer=(void*)destination;
+ int ret=0;
+ #pragma offload target(mic) in(src_ptr,dst_ptr,size) out(ret)
+ {
+ ret=memcmp(dst_ptr.pointer,src_ptr.pointer,size*sizeof(Element1));
+ }
+ if(ret==0)
+ return true;
+ }
+ else
+ {
+ Devices::MICHider<const Element1> src_ptr;
+ src_ptr.pointer=source;
+ Devices::MICHider<const Element2> dst_ptr;
+ dst_ptr.pointer=destination;
+ bool ret=false;
+ #pragma offload target(mic) in(src_ptr,dst_ptr,size) out(ret)
+ {
+ int i=0;
+ for(i=0;i<size;i++)
+ if(dst_ptr.pointer[i]!=src_ptr.pointer[i])
+ break;
+ if(i==size)
+ ret=true;
+ else
+ ret=false;
+ }
+ return ret;
+ }
+ return false;
+#else
+ throw Exceptions::MICSupportMissing();
+#endif
+}
+
+/****
+ * Operations MIC -> Host
+ */
+
+template< typename DestinationElement,
+ typename SourceElement,
+ typename Index >
+bool
+ArrayOperations< Devices::Host, Devices::MIC >::
+copyMemory( DestinationElement* destination,
+ const SourceElement* source,
+ const Index size )
+{
+ TNL_ASSERT( destination, );
+ TNL_ASSERT( source, );
+#ifdef HAVE_MIC
+ if( std::is_same< DestinationElement, SourceElement >::value )
+ {
+ Devices::MICHider<void> src_ptr;
+ src_ptr.pointer=(void*)source;
+
+ //JAKA KONSTANTA se vejde do stacku 5MB?
+ if(size<MIC_STACK_VAR_LIM)
+ {
+ uint8_t tmp[size*sizeof(SourceElement)];
+
+ #pragma offload target(mic) in(src_ptr,size) out(tmp)
+ {
+ memcpy((void*)&tmp,src_ptr.pointer,size*sizeof(SourceElement));
+ }
+
+ memcpy((void*)destination,(void*)&tmp,size*sizeof(SourceElement));
+ return true;
+ }
+ else
+ {
+ //direct -- pomalejÅ¡ÃÂ
+ uint8_t* tmp=(uint8_t*)destination;
+ #pragma offload target(mic) in(src_ptr,size) out(tmp:length(size))
+ {
+ memcpy((void*)tmp,src_ptr.pointer,size*sizeof(SourceElement));
+ }
+ return true;
+ }
+ }
+ else
+ {
+ Devices::MICHider<const SourceElement> src_ptr;
+ src_ptr.pointer=source;
+
+ if(size<MIC_STACK_VAR_LIM)
+ {
+ uint8_t tmp[size*sizeof(DestinationElement)];
+
+ #pragma offload target(mic) in(src_ptr,size) out(tmp)
+ {
+ DestinationElement *dst=(DestinationElement*)&tmp;
+ for(int i=0;i<size;i++)
+ dst[i]=src_ptr.pointer[i];
+ }
+
+ memcpy((void*)destination,(void*)&tmp,size*sizeof(DestinationElement));
+ return true;
+ }
+ else
+ {
+ //direct pseudo heap-- pomalejÅ¡ÃÂ
+ uint8_t* tmp=(uint8_t*)destination;
+ #pragma offload target(mic) in(src_ptr,size) out(tmp:length(size*sizeof(DestinationElement)))
+ {
+ DestinationElement *dst=(DestinationElement*)tmp;
+ for(int i=0;i<size;i++)
+ dst[i]=src_ptr.pointer[i];
+ }
+ return true;
+ }
+ }
+ return false;
+#else
+ throw Exceptions::MICSupportMissing();
+#endif
+}
+
+
+template< typename Element1,
+ typename Element2,
+ typename Index >
+bool
+ArrayOperations< Devices::Host, Devices::MIC >::
+compareMemory( const Element1* destination,
+ const Element2* source,
+ const Index size )
+{
+ /***
+ * Here, destination is on host and source is on MIC device.
+ */
+ TNL_ASSERT( destination, );
+ TNL_ASSERT( source, );
+ TNL_ASSERT( size >= 0, std::cerr << "size = " << size );
+#ifdef HAVE_MIC
+ Index compared( 0 );
+ Index transfer( 0 );
+ std::size_t max_transfer=MIC_STACK_VAR_LIM/sizeof(Element2);
+ uint8_t host_buffer[max_transfer*sizeof(Element2)];
+
+ Devices::MICHider<const Element2> src_ptr;
+
+ while( compared < size )
+ {
+ transfer=min(size-compared,max_transfer);
+ src_ptr.pointer=source+compared;
+ #pragma offload target(mic) out(host_buffer) in(src_ptr,transfer)
+ {
+ memcpy((void*)&host_buffer,(void*)src_ptr.pointer,transfer*sizeof(Element2));
+ }
+ if( ! ArrayOperations< Devices::Host >::compareMemory( &destination[ compared ], (Element2*)&host_buffer, transfer ) )
+ {
+ return false;
+ }
+ compared += transfer;
+ }
+ return true;
+#else
+ throw Exceptions::MICSupportMissing();
+#endif
+}
+
+/****
+ * Operations Host -> MIC
+ */
+template< typename DestinationElement,
+ typename SourceElement,
+ typename Index >
+bool
+ArrayOperations< Devices::MIC, Devices::Host >::
+copyMemory( DestinationElement* destination,
+ const SourceElement* source,
+ const Index size )
+{
+ TNL_ASSERT( destination, );
+ TNL_ASSERT( source, );
+ TNL_ASSERT( size >= 0, std::cerr << "size = " << size );
+#ifdef HAVE_MIC
+ if( std::is_same< DestinationElement, SourceElement >::value )
+ {
+ Devices::MICHider<void> dst_ptr;
+ dst_ptr.pointer=(void*)destination;
+
+ //JAKA KONSTANTA se vejde do stacku 5MB?
+ if(size<MIC_STACK_VAR_LIM)
+ {
+ uint8_t tmp[size*sizeof(SourceElement)];
+ memcpy((void*)&tmp,(void*)source,size*sizeof(SourceElement));
+
+ #pragma offload target(mic) in(dst_ptr,tmp,size)
+ {
+ memcpy(dst_ptr.pointer,(void*)&tmp,size*sizeof(SourceElement));
+ }
+
+ return true;
+ }
+ else
+ {
+ //direct pseudo heap-- pomalejÅ¡ÃÂ
+ uint8_t* tmp=(uint8_t*)source;
+ #pragma offload target(mic) in(dst_ptr,size) in(tmp:length(size))
+ {
+ memcpy(dst_ptr.pointer,(void*)tmp,size*sizeof(SourceElement));
+ }
+ return true;
+ }
+ }
+ else
+ {
+ Devices::MICHider<DestinationElement> dst_ptr;
+ dst_ptr.pointer=destination;
+
+ if(size<MIC_STACK_VAR_LIM)
+ {
+ uint8_t tmp[size*sizeof(SourceElement)];
+ memcpy((void*)&tmp,(void*)source,size*sizeof(SourceElement));
+
+ #pragma offload target(mic) in(dst_ptr,size,tmp)
+ {
+ SourceElement *src=(SourceElement*)&tmp;
+ for(int i=0;i<size;i++)
+ dst_ptr.pointer[i]=src[i];
+ }
+ return true;
+ }
+ else
+ {
+ //direct pseudo heap-- pomalejÅ¡ÃÂ
+ uint8_t* tmp=(uint8_t*)source;
+ #pragma offload target(mic) in(dst_ptr,size) in(tmp:length(size*sizeof(SourceElement)))
+ {
+ SourceElement *src=(SourceElement*)tmp;
+ for(int i=0;i<size;i++)
+ dst_ptr.pointer[i]=src[i];
+ }
+ return true;
+ }
+ }
+ return false;
+#else
+ throw Exceptions::MICSupportMissing();
+#endif
+}
+
+template< typename Element1,
+ typename Element2,
+ typename Index >
+bool
+ArrayOperations< Devices::MIC, Devices::Host >::
+compareMemory( const Element1* hostData,
+ const Element2* deviceData,
+ const Index size )
+{
+ TNL_ASSERT( hostData, );
+ TNL_ASSERT( deviceData, );
+ TNL_ASSERT( size >= 0, std::cerr << "size = " << size );
+ return ArrayOperations< Devices::Host, Devices::MIC >::compareMemory( deviceData, hostData, size );
+}
+
+} // namespace Algorithms
+} // namespace Containers
+} // namespace TNL
diff --git a/src/TNL/Containers/Algorithms/CMakeLists.txt b/src/TNL/Containers/Algorithms/CMakeLists.txt
old mode 100755
new mode 100644
index 4b2744aced7e565692ee3a2469580fff1f7186a8..89fbb0368766b45a0c9aca161c44ef29bce5db47
--- a/src/TNL/Containers/Algorithms/CMakeLists.txt
+++ b/src/TNL/Containers/Algorithms/CMakeLists.txt
@@ -3,6 +3,7 @@ ADD_SUBDIRECTORY( TemplateExplicitInstantiation )
set( headers ArrayOperations.h
ArrayOperationsHost_impl.h
ArrayOperationsCuda_impl.h
+ ArrayOperationsMIC_impl.h
cuda-prefix-sum.h
cuda-prefix-sum_impl.h
reduction-operations.h
@@ -17,6 +18,7 @@ set( headers ArrayOperations.h
VectorOperations.h
VectorOperationsHost_impl.h
VectorOperationsCuda_impl.h
+ VectorOperationsMIC_impl.h
)
INSTALL( FILES ${headers} DESTINATION include/tnl-${tnlVersion}/TNL/Containers/Algorithms )
diff --git a/src/TNL/Containers/Algorithms/CudaMultireductionKernel.h b/src/TNL/Containers/Algorithms/CudaMultireductionKernel.h
index 5ec18f3c13a5227fe081d8f0757dd4e62b8e994a..bc12d3030a17d95280c949a11f738c61152ed05c 100644
--- a/src/TNL/Containers/Algorithms/CudaMultireductionKernel.h
+++ b/src/TNL/Containers/Algorithms/CudaMultireductionKernel.h
@@ -32,6 +32,9 @@ namespace Algorithms {
* architecture so that there are no local memory spills.
*/
static constexpr int Multireduction_maxThreadsPerBlock = 256; // must be a power of 2
+static constexpr int Multireduction_registersPerThread = 38; // empirically determined optimal value
+
+// __CUDA_ARCH__ is defined only in device code!
#if (__CUDA_ARCH__ >= 300 )
static constexpr int Multireduction_minBlocksPerMultiprocessor = 6;
#else
@@ -187,12 +190,14 @@ CudaMultireductionKernelLauncher( Operation& operation,
// we run the kernel with a fixed number of blocks, so the amount of work per
// block increases with enlarging the problem, so even small imbalance can
// cost us dearly.
- // On Tesla K40c, desGridSizeX = 4 * 6 * 15 = 360.
-// const IndexType desGridSizeX = 4 * Multireduction_minBlocksPerMultiprocessor
-// * Devices::CudaDeviceInfo::getCudaMultiprocessors( Devices::CudaDeviceInfo::getActiveDevice() );
- // On Tesla K40c, desGridSizeX = 6 * 15 = 90.
- const IndexType desGridSizeX = Multireduction_minBlocksPerMultiprocessor
- * Devices::CudaDeviceInfo::getCudaMultiprocessors( Devices::CudaDeviceInfo::getActiveDevice() );
+ // Therefore, desGridSize = blocksPerMultiprocessor * numberOfMultiprocessors
+ // where blocksPerMultiprocessor is determined according to the number of
+ // available registers on the multiprocessor.
+ // On Tesla K40c, desGridSize = 8 * 15 = 120.
+ const int activeDevice = Devices::CudaDeviceInfo::getActiveDevice();
+ const int blocksdPerMultiprocessor = Devices::CudaDeviceInfo::getRegistersPerMultiprocessor( activeDevice )
+ / ( Multireduction_maxThreadsPerBlock * Multireduction_registersPerThread );
+ const int desGridSizeX = blocksdPerMultiprocessor * Devices::CudaDeviceInfo::getCudaMultiprocessors( activeDevice );
dim3 blockSize, gridSize;
// version A: max 16 rows of threads
@@ -230,8 +235,7 @@ CudaMultireductionKernelLauncher( Operation& operation,
// (make an overestimate to avoid reallocation on every call if n is increased by 1 each time)
const size_t buf_size = 8 * ( n / 8 + 1 ) * desGridSizeX * sizeof( ResultType );
CudaReductionBuffer& cudaReductionBuffer = CudaReductionBuffer::getInstance();
- if( ! cudaReductionBuffer.setSize( buf_size ) )
- throw 1;
+ cudaReductionBuffer.setSize( buf_size );
output = cudaReductionBuffer.template getData< ResultType >();
// when there is only one warp per blockSize.x, we need to allocate two warps
@@ -304,7 +308,7 @@ CudaMultireductionKernelLauncher( Operation& operation,
default:
TNL_ASSERT( false, std::cerr << "Block size is " << blockSize.x << " which is none of 1, 2, 4, 8, 16, 32, 64, 128, 256 or 512." << std::endl );
}
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
// return the size of the output array on the CUDA device
return gridSize.x;
diff --git a/src/TNL/Containers/Algorithms/CudaReductionBuffer.h b/src/TNL/Containers/Algorithms/CudaReductionBuffer.h
index 3b7a2150267b2a98c845221bd54d95a9ae509fff..2897c7280a6bc61f9b60a9cb3c7b44a94ad20de3 100644
--- a/src/TNL/Containers/Algorithms/CudaReductionBuffer.h
+++ b/src/TNL/Containers/Algorithms/CudaReductionBuffer.h
@@ -8,10 +8,15 @@
/* See Copyright Notice in tnl/Copyright */
+// Implemented by: Jakub Klinkovsky
+
#pragma once
#include <stdlib.h>
+
#include <TNL/Devices/Cuda.h>
+#include <TNL/Exceptions/CudaBadAlloc.h>
+#include <TNL/Exceptions/CudaSupportMissing.h>
namespace TNL {
namespace Containers {
@@ -20,35 +25,34 @@ namespace Algorithms {
class CudaReductionBuffer
{
public:
- inline static CudaReductionBuffer& getInstance( size_t size = 0 )
+ inline static CudaReductionBuffer& getInstance()
{
- static CudaReductionBuffer instance( size );
+ static CudaReductionBuffer instance;
return instance;
}
- inline bool setSize( size_t size )
+ inline void setSize( size_t size )
{
#ifdef HAVE_CUDA
if( size > this->size )
{
- if( data ) cudaFree( data );
- this->size = size;
- if( cudaMalloc( ( void** ) &this->data, size ) != cudaSuccess )
- {
- std::cerr << "I am not able to allocate reduction buffer on the GPU." << std::endl;
+ this->free();
+ if( cudaMalloc( ( void** ) &this->data, size ) != cudaSuccess ) {
this->data = 0;
+ throw Exceptions::CudaBadAlloc();
}
- return checkCudaDevice;
+ this->size = size;
}
- else
- return true;
#else
- return false;
+ throw Exceptions::CudaSupportMissing();
#endif
}
template< typename Type >
- Type* getData() { return ( Type* ) this->data; }
+ Type* getData()
+ {
+ return ( Type* ) this->data;
+ }
private:
// stop the compiler generating methods of copy the object
@@ -56,10 +60,10 @@ class CudaReductionBuffer
CudaReductionBuffer& operator=( CudaReductionBuffer const& copy ); // Not Implemented
// private constructor of the singleton
- inline CudaReductionBuffer( size_t size = 0 ): data( 0 ), size( 0 )
+ inline CudaReductionBuffer( size_t size = 0 )
{
#ifdef HAVE_CUDA
- if( size != 0 ) setSize( size );
+ setSize( size );
atexit( CudaReductionBuffer::free_atexit );
#endif
}
@@ -76,17 +80,17 @@ class CudaReductionBuffer
if( data )
{
cudaFree( data );
- data = 0;
+ data = nullptr;
+ TNL_CHECK_CUDA_DEVICE;
}
#endif
}
- void* data;
+ void* data = nullptr;
- size_t size;
+ size_t size = 0;
};
} // namespace Algorithms
} // namespace Containers
} // namespace TNL
-
diff --git a/src/TNL/Containers/Algorithms/CudaReductionKernel.h b/src/TNL/Containers/Algorithms/CudaReductionKernel.h
index 98a8841a36917cbb420dd54ef25c619a699853fb..a456dad4fa25dacc41f55b79d8fcf6791c5dfd35 100644
--- a/src/TNL/Containers/Algorithms/CudaReductionKernel.h
+++ b/src/TNL/Containers/Algorithms/CudaReductionKernel.h
@@ -30,6 +30,9 @@ namespace Algorithms {
* architecture so that there are no local memory spills.
*/
static constexpr int Reduction_maxThreadsPerBlock = 256; // must be a power of 2
+static constexpr int Reduction_registersPerThread = 32; // empirically determined optimal value
+
+// __CUDA_ARCH__ is defined only in device code!
#if (__CUDA_ARCH__ >= 300 )
static constexpr int Reduction_minBlocksPerMultiprocessor = 8;
#else
@@ -189,20 +192,22 @@ CudaReductionKernelLauncher( Operation& operation,
// we run the kernel with a fixed number of blocks, so the amount of work per
// block increases with enlarging the problem, so even small imbalance can
// cost us dearly.
- // On Tesla K40c, desGridSize = 4 * 6 * 15 = 360.
-// const IndexType desGridSize = 4 * Reduction_minBlocksPerMultiprocessor
-// * Devices::CudaDeviceInfo::getCudaMultiprocessors( Devices::CudaDeviceInfo::getActiveDevice() );
- // On Tesla K40c, desGridSize = 6 * 15 = 90.
- const IndexType desGridSize = Reduction_minBlocksPerMultiprocessor
- * Devices::CudaDeviceInfo::getCudaMultiprocessors( Devices::CudaDeviceInfo::getActiveDevice() );
- dim3 blockSize( 256 ), gridSize( 0 );
+ // Therefore, desGridSize = blocksPerMultiprocessor * numberOfMultiprocessors
+ // where blocksPerMultiprocessor is determined according to the number of
+ // available registers on the multiprocessor.
+ // On Tesla K40c, desGridSize = 8 * 15 = 120.
+ const int activeDevice = Devices::CudaDeviceInfo::getActiveDevice();
+ const int blocksdPerMultiprocessor = Devices::CudaDeviceInfo::getRegistersPerMultiprocessor( activeDevice )
+ / ( Reduction_maxThreadsPerBlock * Reduction_registersPerThread );
+ const int desGridSize = blocksdPerMultiprocessor * Devices::CudaDeviceInfo::getCudaMultiprocessors( activeDevice );
+ dim3 blockSize, gridSize;
+ blockSize.x = Reduction_maxThreadsPerBlock;
gridSize.x = min( Devices::Cuda::getNumberOfBlocks( size, blockSize.x ), desGridSize );
// create reference to the reduction buffer singleton and set size
const size_t buf_size = desGridSize * sizeof( ResultType );
CudaReductionBuffer& cudaReductionBuffer = CudaReductionBuffer::getInstance();
- if( ! cudaReductionBuffer.setSize( buf_size ) )
- throw 1;
+ cudaReductionBuffer.setSize( buf_size );
output = cudaReductionBuffer.template getData< ResultType >();
// when there is only one warp per blockSize.x, we need to allocate two warps
@@ -273,7 +278,7 @@ CudaReductionKernelLauncher( Operation& operation,
default:
TNL_ASSERT( false, std::cerr << "Block size is " << blockSize. x << " which is none of 1, 2, 4, 8, 16, 32, 64, 128, 256 or 512." );
}
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
// return the size of the output array on the CUDA device
return gridSize.x;
diff --git a/src/TNL/Containers/Algorithms/Multireduction.h b/src/TNL/Containers/Algorithms/Multireduction.h
index 9087db93ba6bfded63438c2e59175cc80794042c..8aa314e754d9657cef49d09c42d0bd544a2848b7 100644
--- a/src/TNL/Containers/Algorithms/Multireduction.h
+++ b/src/TNL/Containers/Algorithms/Multireduction.h
@@ -54,6 +54,21 @@ public:
typename Operation::ResultType* hostResult );
};
+template<>
+class Multireduction< Devices::MIC >
+{
+public:
+ template< typename Operation >
+ static bool
+ reduce( Operation& operation,
+ int n,
+ const typename Operation::IndexType size,
+ const typename Operation::RealType* deviceInput1,
+ const typename Operation::IndexType ldInput1,
+ const typename Operation::RealType* deviceInput2,
+ typename Operation::ResultType* hostResult );
+};
+
} // namespace Algorithms
} // namespace Containers
} // namespace TNL
diff --git a/src/TNL/Containers/Algorithms/Multireduction_impl.h b/src/TNL/Containers/Algorithms/Multireduction_impl.h
index 01db7844985447ac883020dde9478987c4d3b2a9..e5433f045e90c1ce5cd2f4d090b5a2e5a863464c 100644
--- a/src/TNL/Containers/Algorithms/Multireduction_impl.h
+++ b/src/TNL/Containers/Algorithms/Multireduction_impl.h
@@ -17,6 +17,7 @@
//#define CUDA_REDUCTION_PROFILING
#include <TNL/Assert.h>
+#include <TNL/Exceptions/CudaSupportMissing.h>
#include <TNL/Containers/Algorithms/reduction-operations.h>
#include <TNL/Containers/Algorithms/ArrayOperations.h>
#include <TNL/Containers/Algorithms/CudaMultireductionKernel.h>
@@ -59,8 +60,8 @@ reduce( Operation& operation,
typename Operation::ResultType* hostResult )
{
#ifdef HAVE_CUDA
- TNL_ASSERT( n > 0, );
- TNL_ASSERT( size <= ldInput1, );
+ TNL_ASSERT_GT( n, 0, "The number of datasets must be positive." );
+ TNL_ASSERT_LE( size, ldInput1, "The size of the input cannot exceed its leading dimension." );
typedef typename Operation::IndexType IndexType;
typedef typename Operation::RealType RealType;
@@ -143,10 +144,9 @@ reduce( Operation& operation,
std::cout << " Multireduction of small data set on CPU took " << timer.getRealTime() << " sec. " << std::endl;
#endif
- return checkCudaDevice;
+ return TNL_CHECK_CUDA_DEVICE;
#else
- CudaSupportMissingMessage;
- return false;
+ throw Exceptions::CudaSupportMissing();
#endif
};
@@ -171,8 +171,8 @@ reduce( Operation& operation,
const typename Operation::RealType* input2,
typename Operation::ResultType* result )
{
- TNL_ASSERT( n > 0, );
- TNL_ASSERT( size <= ldInput1, );
+ TNL_ASSERT_GT( n, 0, "The number of datasets must be positive." );
+ TNL_ASSERT_LE( size, ldInput1, "The size of the input cannot exceed its leading dimension." );
typedef typename Operation::IndexType IndexType;
typedef typename Operation::RealType RealType;
@@ -250,6 +250,30 @@ reduce( Operation& operation,
return true;
}
+template< typename Operation >
+bool
+Multireduction< Devices::MIC >::
+reduce( Operation& operation,
+ int n,
+ const typename Operation::IndexType size,
+ const typename Operation::RealType* input1,
+ const typename Operation::IndexType ldInput1,
+ const typename Operation::RealType* input2,
+ typename Operation::ResultType* result )
+{
+ TNL_ASSERT( n > 0, );
+ TNL_ASSERT( size <= ldInput1, );
+
+ typedef typename Operation::IndexType IndexType;
+ typedef typename Operation::RealType RealType;
+ typedef typename Operation::ResultType ResultType;
+
+
+ std::cout << "Not Implemented yet Multireduction< Devices::MIC >::reduce" << std::endl;
+ return true;
+}
+
+
} // namespace Algorithms
} // namespace Containers
} // namespace TNL
diff --git a/src/TNL/Containers/Algorithms/Reduction_impl.h b/src/TNL/Containers/Algorithms/Reduction_impl.h
index cd4d636c824481c56c56e30fb6f285045df0fe7c..e684b44ac5393adf2348b3a6fed45ea8bba57370 100644
--- a/src/TNL/Containers/Algorithms/Reduction_impl.h
+++ b/src/TNL/Containers/Algorithms/Reduction_impl.h
@@ -15,6 +15,7 @@
//#define CUDA_REDUCTION_PROFILING
#include <TNL/Assert.h>
+#include <TNL/Exceptions/CudaSupportMissing.h>
#include <TNL/Containers/Algorithms/reduction-operations.h>
#include <TNL/Containers/Algorithms/ArrayOperations.h>
#include <TNL/Containers/Algorithms/CudaReductionKernel.h>
@@ -51,11 +52,19 @@ reductionOnCudaDevice( Operation& operation,
typedef typename Operation::ResultType ResultType;
typedef typename Operation::LaterReductionOperation LaterReductionOperation;
+ /***
+ * Only fundamental and pointer types can be safely reduced on host. Complex
+ * objects stored on the device might contain pointers into the device memory,
+ * in which case reduction on host might fail.
+ */
+ constexpr bool can_reduce_all_on_host = std::is_fundamental< RealType >::value || std::is_pointer< RealType >::value;
+ constexpr bool can_reduce_later_on_host = std::is_fundamental< ResultType >::value || std::is_pointer< ResultType >::value;
+
/***
* First check if the input array(s) is/are large enough for the reduction on GPU.
* Otherwise copy it/them to host and reduce on CPU.
*/
- if( size <= minGPUReductionDataSize )
+ if( can_reduce_all_on_host && size <= minGPUReductionDataSize )
{
RealType hostArray1[ minGPUReductionDataSize ];
RealType hostArray2[ minGPUReductionDataSize ];
@@ -92,40 +101,68 @@ reductionOnCudaDevice( Operation& operation,
timer.start();
#endif
- /***
- * Transfer the reduced data from device to host.
- */
- ResultType resultArray[ reducedSize ];
- if( ! ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< ResultType, ResultType, IndexType >( resultArray, deviceAux1, reducedSize ) )
- return false;
-
- #ifdef CUDA_REDUCTION_PROFILING
- timer.stop();
- std::cout << " Transferring data to CPU took " << timer.getRealTime() << " sec. " << std::endl;
- #endif
+ if( can_reduce_later_on_host ) {
+ /***
+ * Transfer the reduced data from device to host.
+ */
+ ResultType resultArray[ reducedSize ];
+ if( ! ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< ResultType, ResultType, IndexType >( resultArray, deviceAux1, reducedSize ) )
+ return false;
+
+ #ifdef CUDA_REDUCTION_PROFILING
+ timer.stop();
+ std::cout << " Transferring data to CPU took " << timer.getRealTime() << " sec. " << std::endl;
+ timer.reset();
+ timer.start();
+ #endif
+
+ /***
+ * Reduce the data on the host system.
+ */
+ LaterReductionOperation laterReductionOperation;
+ result = laterReductionOperation. initialValue();
+ for( IndexType i = 0; i < reducedSize; i ++ )
+ result = laterReductionOperation.reduceOnHost( i, result, resultArray, ( ResultType*) 0 );
+
+ #ifdef CUDA_REDUCTION_PROFILING
+ timer.stop();
+ std::cout << " Reduction of small data set on CPU took " << timer.getRealTime() << " sec. " << std::endl;
+ #endif
+ }
+ else {
+ /***
+ * Data can't be safely reduced on host, so continue with the reduction on the CUDA device.
+ */
+ LaterReductionOperation laterReductionOperation;
+ while( reducedSize > 1 ) {
+ reducedSize = CudaReductionKernelLauncher( laterReductionOperation,
+ reducedSize,
+ deviceAux1,
+ (ResultType*) 0,
+ deviceAux1 );
+ }
+
+ #ifdef CUDA_REDUCTION_PROFILING
+ timer.stop();
+ std::cout << " Reduction of small data set on GPU took " << timer.getRealTime() << " sec. " << std::endl;
+ timer.reset();
+ timer.start();
+ #endif
+
+ ResultType resultArray[ 1 ];
+ if( ! ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< ResultType, ResultType, IndexType >( resultArray, deviceAux1, reducedSize ) )
+ return false;
+ result = resultArray[ 0 ];
- #ifdef CUDA_REDUCTION_PROFILING
- timer.reset();
- timer.start();
- #endif
-
- /***
- * Reduce the data on the host system.
- */
- LaterReductionOperation laterReductionOperation;
- result = laterReductionOperation. initialValue();
- for( IndexType i = 0; i < reducedSize; i ++ )
- result = laterReductionOperation.reduceOnHost( i, result, resultArray, ( ResultType*) 0 );
-
- #ifdef CUDA_REDUCTION_PROFILING
- timer.stop();
- std::cout << " Reduction of small data set on CPU took " << timer.getRealTime() << " sec. " << std::endl;
- #endif
+ #ifdef CUDA_REDUCTION_PROFILING
+ timer.stop();
+ std::cout << " Transferring the result to CPU took " << timer.getRealTime() << " sec. " << std::endl;
+ #endif
+ }
- return checkCudaDevice;
+ return TNL_CHECK_CUDA_DEVICE;
#else
- CudaSupportMissingMessage;;
- return false;
+ throw Exceptions::CudaSupportMissing();
#endif
};
diff --git a/src/TNL/Containers/Algorithms/TemplateExplicitInstantiation/CMakeLists.txt b/src/TNL/Containers/Algorithms/TemplateExplicitInstantiation/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Containers/Algorithms/VectorOperations.h b/src/TNL/Containers/Algorithms/VectorOperations.h
index fa490959f475fbe8a801fa33ef18767eb13d498d..fb24f97c178c8571d081b77496c91101b5a45e31 100644
--- a/src/TNL/Containers/Algorithms/VectorOperations.h
+++ b/src/TNL/Containers/Algorithms/VectorOperations.h
@@ -238,9 +238,123 @@ class VectorOperations< Devices::Cuda >
const typename Vector::IndexType end );
};
+#ifdef HAVE_MIC
+template<>
+class VectorOperations< Devices::MIC >
+{
+ public:
+
+ template< typename Vector >
+ static void addElement( Vector& v,
+ const typename Vector::IndexType i,
+ const typename Vector::RealType& value );
+
+ template< typename Vector >
+ static void addElement( Vector& v,
+ const typename Vector::IndexType i,
+ const typename Vector::RealType& value,
+ const typename Vector::RealType& thisElementMultiplicator );
+
+ template< typename Vector >
+ static typename Vector::RealType getVectorMax( const Vector& v );
+
+ template< typename Vector >
+ static typename Vector::RealType getVectorMin( const Vector& v );
+
+ template< typename Vector >
+ static typename Vector::RealType getVectorAbsMax( const Vector& v );
+
+ template< typename Vector >
+ static typename Vector::RealType getVectorAbsMin( const Vector& v );
+
+ template< typename Vector >
+ static typename Vector::RealType getVectorL1Norm( const Vector& v );
+
+ template< typename Vector >
+ static typename Vector::RealType getVectorL2Norm( const Vector& v );
+
+ template< typename Vector >
+ static typename Vector::RealType getVectorLpNorm( const Vector& v,
+ const typename Vector::RealType& p );
+
+ template< typename Vector >
+ static typename Vector::RealType getVectorSum( const Vector& v );
+
+ template< typename Vector1, typename Vector2 >
+ static typename Vector1::RealType getVectorDifferenceMax( const Vector1& v1,
+ const Vector2& v2 );
+
+ template< typename Vector1, typename Vector2 >
+ static typename Vector1::RealType getVectorDifferenceMin( const Vector1& v1,
+ const Vector2& v2 );
+
+ template< typename Vector1, typename Vector2 >
+ static typename Vector1::RealType getVectorDifferenceAbsMax( const Vector1& v1,
+ const Vector2& v2 );
+
+ template< typename Vector1, typename Vector2 >
+ static typename Vector1::RealType getVectorDifferenceAbsMin( const Vector1& v1,
+ const Vector2& v2 );
+
+ template< typename Vector1, typename Vector2 >
+ static typename Vector1::RealType getVectorDifferenceL1Norm( const Vector1& v1,
+ const Vector2& v2 );
+
+ template< typename Vector1, typename Vector2 >
+ static typename Vector1::RealType getVectorDifferenceL2Norm( const Vector1& v1,
+ const Vector2& v2 );
+
+ template< typename Vector1, typename Vector2 >
+ static typename Vector1::RealType getVectorDifferenceLpNorm( const Vector1& v1,
+ const Vector2& v2,
+ const typename Vector1::RealType& p );
+
+ template< typename Vector1, typename Vector2 >
+ static typename Vector1::RealType getVectorDifferenceSum( const Vector1& v1,
+ const Vector2& v2 );
+
+ template< typename Vector >
+ static void vectorScalarMultiplication( Vector& v,
+ const typename Vector::RealType& alpha );
+
+ template< typename Vector1, typename Vector2 >
+ static typename Vector1::RealType getScalarProduct( const Vector1& v1,
+ const Vector2& v2 );
+
+ template< typename Vector1, typename Vector2 >
+ static void addVector( Vector1& y,
+ const Vector2& x,
+ const typename Vector2::RealType& alpha,
+ const typename Vector1::RealType& thisMultiplicator = 1.0 );
+
+ template< typename Vector1, typename Vector2, typename Vector3 >
+ static void addVectors( Vector1& v,
+ const Vector2& v1,
+ const typename Vector2::RealType& multiplicator1,
+ const Vector3& v2,
+ const typename Vector3::RealType& multiplicator2,
+ const typename Vector1::RealType& thisMultiplicator = 1.0 );
+
+
+ template< typename Vector >
+ static void computePrefixSum( Vector& v,
+ const typename Vector::IndexType begin,
+ const typename Vector::IndexType end );
+
+ template< typename Vector >
+ static void computeExclusivePrefixSum( Vector& v,
+ const typename Vector::IndexType begin,
+ const typename Vector::IndexType end );
+};
+#endif
+
} // namespace Algorithms
} // namespace Containers
} // namespace TNL
+#ifdef HAVE_MIC
+#include <TNL/Containers/Algorithms/VectorOperationsMIC_impl.h>
+#endif
#include <TNL/Containers/Algorithms/VectorOperationsHost_impl.h>
#include <TNL/Containers/Algorithms/VectorOperationsCuda_impl.h>
+
diff --git a/src/TNL/Containers/Algorithms/VectorOperationsCuda_impl.h b/src/TNL/Containers/Algorithms/VectorOperationsCuda_impl.h
index 5d4b29379d62a609752d1d2cdedbdaf0b1ba662b..c32f44bbaf3a318a8d43b9116baac563b26cdf5e 100644
--- a/src/TNL/Containers/Algorithms/VectorOperationsCuda_impl.h
+++ b/src/TNL/Containers/Algorithms/VectorOperationsCuda_impl.h
@@ -11,6 +11,7 @@
#pragma once
#include <TNL/tnlConfig.h>
+#include <TNL/Exceptions/CudaSupportMissing.h>
#include <TNL/Containers/Algorithms/VectorOperations.h>
#include <TNL/Containers/Algorithms/cuda-prefix-sum.h>
#include <TNL/Containers/Algorithms/CublasWrapper.h>
@@ -48,7 +49,7 @@ getVectorMax( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result( 0 );
Algorithms::tnlParallelReductionMax< Real, Index > operation;
@@ -68,7 +69,7 @@ getVectorMin( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result( 0 );
Algorithms::tnlParallelReductionMin< Real, Index > operation;
@@ -88,7 +89,7 @@ getVectorAbsMax( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result( 0 );
Algorithms::tnlParallelReductionAbsMax< Real, Index > operation;
@@ -108,7 +109,7 @@ getVectorAbsMin( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result( 0 );
Algorithms::tnlParallelReductionAbsMin< Real, Index > operation;
@@ -128,7 +129,7 @@ getVectorL1Norm( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result( 0 );
Algorithms::tnlParallelReductionAbsSum< Real, Index > operation;
@@ -148,7 +149,7 @@ getVectorL2Norm( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result( 0 );
Algorithms::tnlParallelReductionL2Norm< Real, Index > operation;
@@ -170,9 +171,8 @@ getVectorLpNorm( const Vector& v,
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
- TNL_ASSERT( p > 0.0,
- std::cerr << " p = " << p );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_GE( p, 1.0, "Parameter of the L^p norm must be at least 1.0." );
if( p == 1 )
return getVectorL1Norm( v );
@@ -197,7 +197,7 @@ getVectorSum( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result( 0 );
Algorithms::tnlParallelReductionSum< Real, Index > operation;
@@ -218,8 +218,8 @@ getVectorDifferenceMax( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0 );
Algorithms::tnlParallelReductionDiffMax< Real, Index > operation;
@@ -240,8 +240,8 @@ getVectorDifferenceMin( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0 );
Algorithms::tnlParallelReductionDiffMin< Real, Index > operation;
@@ -263,8 +263,8 @@ getVectorDifferenceAbsMax( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0 );
Algorithms::tnlParallelReductionDiffAbsMax< Real, Index > operation;
@@ -285,8 +285,8 @@ getVectorDifferenceAbsMin( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0 );
Algorithms::tnlParallelReductionDiffAbsMin< Real, Index > operation;
@@ -307,8 +307,8 @@ getVectorDifferenceL1Norm( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0 );
Algorithms::tnlParallelReductionDiffAbsSum< Real, Index > operation;
@@ -329,8 +329,8 @@ getVectorDifferenceL2Norm( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0 );
Algorithms::tnlParallelReductionDiffL2Norm< Real, Index > operation;
@@ -353,10 +353,9 @@ getVectorDifferenceLpNorm( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( p > 0.0,
- std::cerr << " p = " << p );
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
+ TNL_ASSERT_GE( p, 1.0, "Parameter of the L^p norm must be at least 1.0." );
Real result( 0 );
Algorithms::tnlParallelReductionDiffLpNorm< Real, Index > operation;
@@ -378,8 +377,8 @@ getVectorDifferenceSum( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0 );
Algorithms::tnlParallelReductionDiffSum< Real, Index > operation;
@@ -414,24 +413,22 @@ VectorOperations< Devices::Cuda >::
vectorScalarMultiplication( Vector& v,
const typename Vector::RealType& alpha )
{
- typedef typename Vector::RealType Real;
- typedef typename Vector::IndexType Index;
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
- TNL_ASSERT( v.getSize() > 0, );
-
- #ifdef HAVE_CUDA
- dim3 blockSize( 0 ), gridSize( 0 );
- const Index& size = v.getSize();
- blockSize.x = 256;
- Index blocksNumber = ceil( ( double ) size / ( double ) blockSize.x );
- gridSize.x = min( blocksNumber, Devices::Cuda::getMaxGridSize() );
- vectorScalarMultiplicationCudaKernel<<< gridSize, blockSize >>>( v.getData(),
- size,
- alpha );
- checkCudaDevice;
- #else
- CudaSupportMissingMessage;;
- #endif
+#ifdef HAVE_CUDA
+ typedef typename Vector::IndexType Index;
+ dim3 blockSize( 0 ), gridSize( 0 );
+ const Index& size = v.getSize();
+ blockSize.x = 256;
+ Index blocksNumber = ceil( ( double ) size / ( double ) blockSize.x );
+ gridSize.x = min( blocksNumber, Devices::Cuda::getMaxGridSize() );
+ vectorScalarMultiplicationCudaKernel<<< gridSize, blockSize >>>( v.getData(),
+ size,
+ alpha );
+ TNL_CHECK_CUDA_DEVICE;
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
}
@@ -444,8 +441,8 @@ getScalarProduct( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0 );
/*#if defined HAVE_CUBLAS && defined HAVE_CUDA
@@ -498,31 +495,28 @@ addVector( Vector1& y,
const typename Vector2::RealType& alpha,
const typename Vector1::RealType& thisMultiplicator )
{
- typedef typename Vector1::RealType Real;
- typedef typename Vector1::IndexType Index;
+ TNL_ASSERT_GT( x.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( x.getSize(), y.getSize(), "The vector sizes must be the same." );
- TNL_ASSERT( y.getSize() > 0, );
- TNL_ASSERT( y.getSize() == x.getSize(), );
- TNL_ASSERT( y.getData() != 0, );
- TNL_ASSERT( x.getData() != 0, );
-
- #ifdef HAVE_CUDA
- dim3 blockSize( 0 ), gridSize( 0 );
-
- const Index& size = x.getSize();
- dim3 cudaBlockSize( 256 );
- dim3 cudaBlocks;
- cudaBlocks.x = min( Devices::Cuda::getMaxGridSize(), Devices::Cuda::getNumberOfBlocks( size, cudaBlockSize.x ) );
-
- vectorAddVectorCudaKernel<<< cudaBlocks, cudaBlockSize >>>( y.getData(),
- x.getData(),
- size,
- alpha,
- thisMultiplicator);
- checkCudaDevice;
- #else
- CudaSupportMissingMessage;;
- #endif
+#ifdef HAVE_CUDA
+ typedef typename Vector1::IndexType Index;
+
+ dim3 blockSize( 0 ), gridSize( 0 );
+
+ const Index& size = x.getSize();
+ dim3 cudaBlockSize( 256 );
+ dim3 cudaBlocks;
+ cudaBlocks.x = min( Devices::Cuda::getMaxGridSize(), Devices::Cuda::getNumberOfBlocks( size, cudaBlockSize.x ) );
+
+ vectorAddVectorCudaKernel<<< cudaBlocks, cudaBlockSize >>>( y.getData(),
+ x.getData(),
+ size,
+ alpha,
+ thisMultiplicator);
+ TNL_CHECK_CUDA_DEVICE;
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
}
#ifdef HAVE_CUDA
@@ -569,37 +563,30 @@ addVectors( Vector1& v,
const typename Vector3::RealType& multiplicator2,
const typename Vector1::RealType& thisMultiplicator )
{
- typedef typename Vector1::RealType Real;
- typedef typename Vector1::IndexType Index;
-
- TNL_ASSERT( v.getSize() > 0, );
- TNL_ASSERT( v.getSize() == v1.getSize(), );
- TNL_ASSERT( v.getSize() == v2.getSize(), );
- TNL_ASSERT( v.getData() != 0, );
- TNL_ASSERT( v1.getData() != 0, );
- TNL_ASSERT( v2.getData() != 0, );
-
- #ifdef HAVE_CUDA
- dim3 blockSize( 0 ), gridSize( 0 );
-
- const Index& size = v.getSize();
- dim3 cudaBlockSize( 256 );
- dim3 cudaBlocks;
- cudaBlocks.x = min( Devices::Cuda::getMaxGridSize(), Devices::Cuda::getNumberOfBlocks( size, cudaBlockSize.x ) );
-
- vectorAddVectorsCudaKernel<<< cudaBlocks, cudaBlockSize >>>( v.getData(),
- v1.getData(),
- v2.getData(),
- size,
- multiplicator1,
- multiplicator2,
- thisMultiplicator);
- checkCudaDevice;
- #else
- CudaSupportMissingMessage;;
- #endif
-
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v.getSize(), v1.getSize(), "The vector sizes must be the same." );
+ TNL_ASSERT_EQ( v.getSize(), v2.getSize(), "The vector sizes must be the same." );
+#ifdef HAVE_CUDA
+ typedef typename Vector1::IndexType Index;
+ dim3 blockSize( 0 ), gridSize( 0 );
+
+ const Index& size = v.getSize();
+ dim3 cudaBlockSize( 256 );
+ dim3 cudaBlocks;
+ cudaBlocks.x = min( Devices::Cuda::getMaxGridSize(), Devices::Cuda::getNumberOfBlocks( size, cudaBlockSize.x ) );
+
+ vectorAddVectorsCudaKernel<<< cudaBlocks, cudaBlockSize >>>( v.getData(),
+ v1.getData(),
+ v2.getData(),
+ size,
+ multiplicator1,
+ multiplicator2,
+ thisMultiplicator);
+ TNL_CHECK_CUDA_DEVICE;
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
}
template< typename Vector >
@@ -609,7 +596,7 @@ computePrefixSum( Vector& v,
typename Vector::IndexType begin,
typename Vector::IndexType end )
{
- #ifdef HAVE_CUDA
+#ifdef HAVE_CUDA
typedef Algorithms::tnlParallelReductionSum< typename Vector::RealType, typename Vector::IndexType > OperationType;
OperationType operation;
@@ -621,10 +608,10 @@ computePrefixSum( Vector& v,
&v.getData()[ begin ],
&v.getData()[ begin ],
operation,
- Algorithms::inclusivePrefixSum );
- #else
- CudaSupportMissingMessage;;
- #endif
+ Algorithms::PrefixSumType::inclusive );
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
}
template< typename Vector >
@@ -646,7 +633,7 @@ computeExclusivePrefixSum( Vector& v,
&v.getData()[ begin ],
&v.getData()[ begin ],
operation,
- Algorithms::exclusivePrefixSum );
+ Algorithms::PrefixSumType::exclusive );
#endif
}
diff --git a/src/TNL/Containers/Algorithms/VectorOperationsHost_impl.h b/src/TNL/Containers/Algorithms/VectorOperationsHost_impl.h
index aab2a84769c1bf745351ec1e141df1eb740f04ce..d8cbca17eae6398235a165455872d7c9284dacd9 100644
--- a/src/TNL/Containers/Algorithms/VectorOperationsHost_impl.h
+++ b/src/TNL/Containers/Algorithms/VectorOperationsHost_impl.h
@@ -48,7 +48,7 @@ getVectorMax( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result = v.getElement( 0 );
const Index n = v.getSize();
@@ -68,7 +68,7 @@ getVectorMin( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result = v.getElement( 0 );
const Index n = v.getSize();
@@ -88,7 +88,7 @@ getVectorAbsMax( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result = std::fabs( v.getElement( 0 ) );
const Index n = v.getSize();
@@ -109,7 +109,7 @@ getVectorAbsMin( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result = std::fabs( v.getElement( 0 ) );
const Index n = v.getSize();
@@ -129,7 +129,7 @@ getVectorL1Norm( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result( 0.0 );
const Index n = v.getSize();
@@ -149,7 +149,7 @@ getVectorL2Norm( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
const Index n = v.getSize();
@@ -212,9 +212,8 @@ getVectorLpNorm( const Vector& v,
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
- TNL_ASSERT( p > 0.0,
- std::cerr << " p = " << p );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_GE( p, 1.0, "Parameter of the L^p norm must be at least 1.0." );
if( p == 1.0 )
return getVectorL1Norm( v );
@@ -239,7 +238,7 @@ getVectorSum( const Vector& v )
typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
Real result( 0.0 );
const Index n = v.getSize();
@@ -260,8 +259,8 @@ getVectorDifferenceMax( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result = v1.getElement( 0 ) - v2.getElement( 0 );
const Index n = v1.getSize();
@@ -282,8 +281,8 @@ getVectorDifferenceMin( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result = v1.getElement( 0 ) - v2.getElement( 0 );
const Index n = v1.getSize();
@@ -304,8 +303,8 @@ getVectorDifferenceAbsMax( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result = std::fabs( v1.getElement( 0 ) - v2.getElement( 0 ) );
const Index n = v1.getSize();
@@ -326,8 +325,8 @@ getVectorDifferenceAbsMin( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result = std::fabs( v1[ 0 ] - v2[ 0 ] );
const Index n = v1.getSize();
@@ -348,8 +347,8 @@ getVectorDifferenceL1Norm( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0.0 );
const Index n = v1.getSize();
@@ -370,8 +369,8 @@ getVectorDifferenceL2Norm( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0.0 );
const Index n = v1.getSize();
@@ -397,10 +396,9 @@ getVectorDifferenceLpNorm( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( p > 0.0,
- std::cerr << " p = " << p );
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
+ TNL_ASSERT_GE( p, 1.0, "Parameter of the L^p norm must be at least 1.0." );
if( p == 1.0 )
return getVectorDifferenceL1Norm( v1, v2 );
@@ -426,8 +424,8 @@ getVectorDifferenceSum( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
Real result( 0.0 );
const Index n = v1.getSize();
@@ -446,10 +444,9 @@ VectorOperations< Devices::Host >::
vectorScalarMultiplication( Vector& v,
const typename Vector::RealType& alpha )
{
- typedef typename Vector::RealType Real;
typedef typename Vector::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
const Index n = v.getSize();
#ifdef HAVE_OPENMP
@@ -469,8 +466,8 @@ getScalarProduct( const Vector1& v1,
typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v1.getSize() > 0, );
- TNL_ASSERT( v1.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v1.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v1.getSize(), v2.getSize(), "The vector sizes must be the same." );
const Index n = v1.getSize();
#ifdef OPTIMIZED_VECTOR_HOST_OPERATIONS
@@ -530,11 +527,10 @@ addVector( Vector1& y,
const typename Vector2::RealType& alpha,
const typename Vector1::RealType& thisMultiplicator )
{
- typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( x.getSize() > 0, );
- TNL_ASSERT( x.getSize() == y.getSize(), );
+ TNL_ASSERT_GT( x.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( x.getSize(), y.getSize(), "The vector sizes must be the same." );
const Index n = y.getSize();
@@ -542,6 +538,7 @@ addVector( Vector1& y,
#ifdef __GNUC__
// We need to get the address of the first element to avoid
// bounds checking in TNL::Array::operator[]
+ typedef typename Vector1::RealType Real;
Real* Y = y.getData();
const Real* X = x.getData();
#endif
@@ -600,12 +597,11 @@ addVectors( Vector1& v,
const typename Vector3::RealType& multiplicator2,
const typename Vector1::RealType& thisMultiplicator )
{
- typedef typename Vector1::RealType Real;
typedef typename Vector1::IndexType Index;
- TNL_ASSERT( v.getSize() > 0, );
- TNL_ASSERT( v.getSize() == v1.getSize(), );
- TNL_ASSERT( v.getSize() == v2.getSize(), );
+ TNL_ASSERT_GT( v.getSize(), 0, "Vector size must be positive." );
+ TNL_ASSERT_EQ( v.getSize(), v1.getSize(), "The vector sizes must be the same." );
+ TNL_ASSERT_EQ( v.getSize(), v2.getSize(), "The vector sizes must be the same." );
const Index n = v.getSize();
if( thisMultiplicator == 1.0 )
@@ -631,6 +627,7 @@ computePrefixSum( Vector& v,
{
typedef typename Vector::IndexType Index;
+ // TODO: parallelize with OpenMP
for( Index i = begin + 1; i < end; i++ )
v[ i ] += v[ i - 1 ];
}
@@ -645,6 +642,7 @@ computeExclusivePrefixSum( Vector& v,
typedef typename Vector::IndexType Index;
typedef typename Vector::RealType Real;
+ // TODO: parallelize with OpenMP
Real aux( v[ begin ] );
v[ begin ] = 0.0;
for( Index i = begin + 1; i < end; i++ )
diff --git a/src/TNL/Containers/Algorithms/VectorOperationsMIC_impl.h b/src/TNL/Containers/Algorithms/VectorOperationsMIC_impl.h
new file mode 100644
index 0000000000000000000000000000000000000000..ae462d7a8b7d5c7a425e243543f90b8c4444d03b
--- /dev/null
+++ b/src/TNL/Containers/Algorithms/VectorOperationsMIC_impl.h
@@ -0,0 +1,669 @@
+/***************************************************************************
+ VectorOperationsMIC_impl.h - description
+ by hanouvit
+ -------------------
+ begin : Nov 7, 2012
+ copyright : (C) 2012 by Tomas Oberhuber
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+
+#pragma once
+
+#include <TNL/Devices/MIC.h>
+#include <TNL/Math.h>
+
+namespace TNL {
+namespace Containers {
+namespace Algorithms {
+
+//static const int OpenMPVectorOperationsThreshold = 65536; // TODO: check this threshold
+
+template< typename Vector >
+void
+VectorOperations< Devices::MIC >::
+addElement( Vector& v,
+ const typename Vector::IndexType i,
+ const typename Vector::RealType& value )
+{
+ // v[ i ] += value;
+ //cout << "Errorous function, not clear wher should be called (device or Host)" << endl;
+ v.setElement(i,v.getElemet(i)+value);
+}
+
+template< typename Vector >
+void
+VectorOperations< Devices::MIC >::
+addElement( Vector& v,
+ const typename Vector::IndexType i,
+ const typename Vector::RealType& value,
+ const typename Vector::RealType& thisElementMultiplicator )
+{
+ //v[ i ] = thisElementMultiplicator * v[ i ] + value;
+ //cout << "Errorous function, not clear wher should be called (device or Host)" << endl;
+ v.setElement(i,thisElementMultiplicator*v.getElemet(i)+value);
+}
+
+template< typename Vector >
+typename Vector::RealType
+VectorOperations< Devices::MIC >::
+getVectorMax( const Vector& v )
+{
+ //tady je možnost paralelizace
+ typename Vector::RealType result;
+ typename Vector::IndexType size=v.getSize();
+ Devices::MICHider<const typename Vector::RealType > vct;
+ vct.pointer=v.getData();
+
+ #pragma offload target(mic) in(vct,size) out(result)
+ {
+ result=vct.pointer[0];
+ for(typename Vector::IndexType i=1;i<size;i++)
+ {
+ if(result<vct.pointer[i])
+ result=vct.pointer[i];
+ }
+ }
+ return result;
+}
+
+template< typename Vector >
+typename Vector::RealType
+VectorOperations< Devices::MIC >::
+getVectorMin( const Vector& v )
+{
+ //tady je možnost paralelizace
+ typename Vector::RealType result;
+ typename Vector::IndexType size=v.getSize();
+ Devices::MICHider<const typename Vector::RealType > vct;
+ vct.pointer=v.getData();
+
+ #pragma offload target(mic) in(vct,size) out(result)
+ {
+ result=vct.pointer[0];
+ for(typename Vector::IndexType i=1;i<size;i++)
+ {
+ if(result>vct.pointer[i])
+ result=vct.pointer[i];
+ }
+ }
+ return result;
+}
+
+template< typename Vector >
+typename Vector::RealType
+VectorOperations< Devices::MIC >::
+getVectorAbsMax( const Vector& v )
+{
+ //tady je možnost paralelizace
+ typename Vector::RealType result;
+ typename Vector::IndexType size=v.getSize();
+ Devices::MICHider<const typename Vector::RealType > vct;
+ vct.pointer=v.getData();
+
+ #pragma offload target(mic) in(vct,size) out(result)
+ {
+ result=TNL::abs(vct.pointer[0]);
+ for(typename Vector::IndexType i=1;i<size;i++)
+ {
+ if(result<TNL::abs(vct.pointer[i]))
+ result=TNL::abs(vct.pointer[i]);
+ }
+ }
+ return result;
+}
+
+template< typename Vector >
+typename Vector::RealType
+VectorOperations< Devices::MIC >::
+getVectorAbsMin( const Vector& v )
+{
+ //tady je možnost paralelizace
+ typename Vector::RealType result;
+ typename Vector::IndexType size=v.getSize();
+ Devices::MICHider<const typename Vector::RealType > vct;
+ vct.pointer=v.getData();
+
+ #pragma offload target(mic) in(vct,size) out(result)
+ {
+ result=TNL::abs(vct.pointer[0]);
+ for(typename Vector::IndexType i=1;i<size;i++)
+ {
+ if(result>TNL::abs(vct.pointer[i]))
+ result=TNL::abs(vct.pointer[i]);
+ }
+ }
+ return result;
+}
+
+template< typename Vector >
+typename Vector::RealType
+VectorOperations< Devices::MIC >::
+getVectorL1Norm( const Vector& v )
+{
+ typedef typename Vector::RealType Real;
+ typedef typename Vector::IndexType Index;
+ TNL_ASSERT( v. getSize() > 0, );
+
+ Real result( 0.0 );
+ const Index n = v. getSize();
+ Devices::MICHider<const Real > vct;
+ vct.pointer=v.getData();
+
+ #pragma offload target(mic) in(vct,n) inout(result)
+ {
+ #pragma omp parallel for reduction(+:result)// if( n > OpenMPVectorOperationsThreshold ) // TODO: check this threshold
+ for( Index i = 0; i < n; i ++ )
+ result += TNL::abs( vct.pointer[ i ] );
+ }
+ return result;
+}
+
+template< typename Vector >
+typename Vector::RealType
+VectorOperations< Devices::MIC >::
+getVectorL2Norm( const Vector& v )
+{
+ typedef typename Vector::RealType Real;
+ typedef typename Vector::IndexType Index;
+ TNL_ASSERT( v. getSize() > 0, );
+ Real result( 0.0 );
+ const Index n = v. getSize();
+ Devices::MICHider<const Real > vct;
+ vct.pointer=v.getData();
+
+ #pragma offload target(mic) in(vct,n) inout(result)
+ {
+ #pragma omp parallel for reduction(+:result) //if( n > OpenMPVectorOperationsThreshold ) // TODO: check this threshold
+ for( Index i = 0; i < n; i ++ )
+ {
+ const Real& aux = vct.pointer[ i ];
+ result += aux * aux;
+ }
+ }
+ return TNL::sqrt( result );
+}
+
+template< typename Vector >
+typename Vector::RealType
+VectorOperations< Devices::MIC >::
+getVectorLpNorm( const Vector& v,
+ const typename Vector::RealType& p )
+{
+ typedef typename Vector::RealType Real;
+ typedef typename Vector::IndexType Index;
+ TNL_ASSERT( v. getSize() > 0, );
+ TNL_ASSERT( p > 0.0,
+ std::cerr << " p = " << p );
+ if( p == 1.0 )
+ return getVectorL1Norm( v );
+ if( p == 2.0 )
+ return getVectorL2Norm( v );
+
+ Real result( 0.0 );
+ const Index n = v. getSize();
+ Devices::MICHider<const Real > vct;
+ vct.pointer=v.getData();
+
+ #pragma offload target(mic) in(vct,n) inout(result)
+ {
+ #pragma omp parallel for reduction(+:result) //if( n > OpenMPVectorOperationsThreshold ) // TODO: check this threshold
+ for( Index i = 0; i < n; i ++ )
+ {
+ result += TNL::pow( TNL::abs( vct.pointer[ i ] ), p );
+ }
+ }
+ return TNL::pow( result, 1.0 / p );
+}
+
+template< typename Vector >
+typename Vector::RealType
+VectorOperations< Devices::MIC >::
+getVectorSum( const Vector& v )
+{
+ typedef typename Vector::RealType Real;
+ typedef typename Vector::IndexType Index;
+ TNL_ASSERT( v. getSize() > 0, );
+
+ Real result( 0.0 );
+ const Index n = v. getSize();
+ Devices::MICHider<const Real > vct;
+ vct.pointer=v.getData();
+
+ #pragma offload target(mic) in(vct,n) inout(result)
+ {
+ #pragma omp parallel for reduction(+:result)// if( n > OpenMPVectorOperationsThreshold ) // TODO: check this threshold
+ for( Index i = 0; i < n; i ++ )
+ result += vct.pointer[ i ] ;
+ }
+ return result;
+}
+
+template< typename Vector1, typename Vector2 >
+typename Vector1::RealType
+VectorOperations< Devices::MIC>::
+getVectorDifferenceMax( const Vector1& v1,
+ const Vector2& v2 )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+ TNL_ASSERT( v1. getSize() > 0, );
+ TNL_ASSERT( v1. getSize() == v2. getSize(), );
+
+ Real result( 0.0 );
+ const Index n = v1. getSize();
+ Devices::MICHider<const Real > vct1;
+ Devices::MICHider<const Real > vct2;
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+
+ #pragma offload target(mic) in(n,vct1,vct2) out(result)
+ {
+ result = vct1.pointer[0] - vct2.pointer[0];
+ for( Index i = 1; i < n; i ++ )
+ result = TNL::max( result, vct1.pointer[ i ] - vct2.pointer[ i ] );
+ }
+ return result;
+}
+
+template< typename Vector1, typename Vector2 >
+typename Vector1::RealType
+VectorOperations< Devices::MIC >::
+getVectorDifferenceMin( const Vector1& v1,
+ const Vector2& v2 )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+ TNL_ASSERT( v1. getSize() > 0, );
+ TNL_ASSERT( v1. getSize() == v2. getSize(), );
+
+ Real result( 0.0 );
+ const Index n = v1. getSize();
+ Devices::MICHider<const Real > vct1;
+ Devices::MICHider<const Real > vct2;
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+
+ #pragma offload target(mic) in(n,vct1,vct2) out(result)
+ {
+ result = vct1.pointer[0] - vct2.pointer[0];
+ for( Index i = 1; i < n; i ++ )
+ result = TNL::min( result, vct1.pointer[ i ] - vct2.pointer[ i ] );
+ }
+ return result;
+}
+
+template< typename Vector1, typename Vector2 >
+typename Vector1::RealType
+VectorOperations< Devices::MIC >::
+getVectorDifferenceAbsMax( const Vector1& v1,
+ const Vector2& v2 )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+ TNL_ASSERT( v1. getSize() > 0, );
+ TNL_ASSERT( v1. getSize() == v2. getSize(), );
+
+ Real result( 0.0 );
+ const Index n = v1. getSize();
+ Devices::MICHider<const Real > vct1;
+ Devices::MICHider<const Real > vct2;
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+
+ #pragma offload target(mic) in(n,vct1,vct2) out(result)
+ {
+ result = TNL::abs(vct1.pointer[0] - vct2.pointer[0]);
+ for( Index i = 1; i < n; i ++ )
+ result = TNL::max( result, TNL::abs(vct1.pointer[ i ] - vct2.pointer[ i ]) );
+ }
+ return result;
+}
+
+template< typename Vector1, typename Vector2 >
+typename Vector1::RealType
+VectorOperations< Devices::MIC >::
+getVectorDifferenceAbsMin( const Vector1& v1,
+ const Vector2& v2 )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+ TNL_ASSERT( v1. getSize() > 0, );
+ TNL_ASSERT( v1. getSize() == v2. getSize(), );
+
+ Real result( 0.0 );
+ const Index n = v1. getSize();
+ Devices::MICHider<const Real > vct1;
+ Devices::MICHider<const Real > vct2;
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+
+ #pragma offload target(mic) in(n,vct1,vct2) out(result)
+ {
+ result = TNL::abs(vct1.pointer[0] - vct2.pointer[0]);
+ for( Index i = 1; i < n; i ++ )
+ result = TNL::min( result, TNL::abs(vct1.pointer[ i ] - vct2.pointer[ i ]) );
+ }
+ return result;
+}
+
+template< typename Vector1, typename Vector2 >
+typename Vector1::RealType
+VectorOperations< Devices::MIC >::
+getVectorDifferenceL1Norm( const Vector1& v1,
+ const Vector2& v2 )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+
+ TNL_ASSERT( v1. getSize() > 0, );
+ TNL_ASSERT( v1. getSize() == v2. getSize(), );
+
+ Real result( 0.0 );
+ const Index n = v1. getSize();
+ Devices::MICHider<const Real> vct1;
+ Devices::MICHider<const Real > vct2;
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+
+ #pragma offload target(mic) in(n,vct1,vct2) inout(result)
+ {
+ for( Index i = 0; i < n; i ++ )
+ result += TNL::abs( vct1.pointer[ i ] - vct2.pointer[ i ] );
+ }
+ return result;
+}
+
+template< typename Vector1, typename Vector2 >
+typename Vector1::RealType
+VectorOperations< Devices::MIC >::
+getVectorDifferenceL2Norm( const Vector1& v1,
+ const Vector2& v2 )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+
+ TNL_ASSERT( v1. getSize() > 0, );
+ TNL_ASSERT( v1. getSize() == v2. getSize(), );
+
+ Real result( 0.0 );
+ const Index n = v1. getSize();
+
+ Devices::MICHider<const Real > vct1;
+ Devices::MICHider<const Real > vct2;
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+
+ #pragma offload target(mic) in(n,vct1,vct2) inout(result)
+ {
+ for( Index i = 0; i < n; i ++ )
+ {
+ Real aux = TNL::abs( vct1.pointer[ i ] - vct2.pointer[ i ] );
+ result += aux * aux;
+ }
+ }
+
+ return TNL::sqrt( result );
+}
+
+
+template< typename Vector1, typename Vector2 >
+typename Vector1::RealType
+VectorOperations< Devices::MIC >::
+getVectorDifferenceLpNorm( const Vector1& v1,
+ const Vector2& v2,
+ const typename Vector1::RealType& p )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+
+ TNL_ASSERT( p > 0.0,
+ std::cerr << " p = " << p );
+ TNL_ASSERT( v1. getSize() > 0, );
+ TNL_ASSERT( v1. getSize() == v2. getSize(), );
+
+ if( p == 1.0 )
+ return getVectorDifferenceL1Norm( v1, v2 );
+ if( p == 2.0 )
+ return getVectorDifferenceL2Norm( v1, v2 );
+
+ Real result( 0.0 );
+ const Index n = v1. getSize();
+
+ Devices::MICHider<const Real > vct1;
+ Devices::MICHider<const Real > vct2;
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+
+ #pragma offload target(mic) in(n,vct1,vct2) inout(result)
+ {
+ for( Index i = 0; i < n; i ++ )
+ {
+ result += TNL::pow( TNL::abs( vct1.pointer[ i ] - vct2.pointer[ i ] ), p );
+ }
+ }
+ return TNL::pow( result, 1.0 / p );
+}
+
+template< typename Vector1, typename Vector2 >
+typename Vector1::RealType
+VectorOperations< Devices::MIC >::
+getVectorDifferenceSum( const Vector1& v1,
+ const Vector2& v2 )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+
+ TNL_ASSERT( v1. getSize() > 0, );
+ TNL_ASSERT( v1. getSize() == v2. getSize(), );
+
+ Real result( 0.0 );
+ const Index n = v1. getSize();
+ Devices::MICHider<const Real > vct1;
+ Devices::MICHider<const Real > vct2;
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+
+ #pragma offload target(mic) in(n,vct1,vct2) inout(result)
+ {
+ for( Index i = 0; i < n; i ++ )
+ result += vct1.pointer[ i ] - vct2.pointer[ i ];
+ }
+ return result;
+}
+
+
+template< typename Vector >
+void
+VectorOperations< Devices::MIC >::
+vectorScalarMultiplication( Vector& v,
+ const typename Vector::RealType& alpha )
+{
+ typedef typename Vector::RealType Real;
+ typedef typename Vector::IndexType Index;
+
+ TNL_ASSERT( v. getSize() > 0, );
+
+ const Index n = v. getSize();
+ Devices::MICHider<Real > vct;
+ vct.pointer=v.getData();
+ Real a=alpha;
+
+ #pragma offload target(mic) in(vct,a,n)
+ {
+ for( Index i = 0; i < n; i ++ )
+ vct.pointer[ i ] *= a;
+ }
+}
+
+
+template< typename Vector1, typename Vector2 >
+typename Vector1::RealType
+VectorOperations< Devices::MIC >::
+getScalarProduct( const Vector1& v1,
+ const Vector2& v2 )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+ TNL_ASSERT( v1. getSize() > 0, );
+ TNL_ASSERT( v1. getSize() == v2. getSize(), );
+
+ Real result( 0.0 );
+ const Index n = v1. getSize();
+ Devices::MICHider<const Real > vct1;
+ Devices::MICHider<const Real > vct2;
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+
+ #pragma offload target(mic) in(vct1,vct2,n) inout(result)
+ {
+ #pragma omp parallel for reduction(+:result)// if( n > OpenMPVectorOperationsThreshold ) // TODO: check this threshold
+ for( Index i = 0; i < n; i++ )
+ result += vct1.pointer[ i ] * vct2.pointer[ i ];
+ }
+ /*Real result1( 0.0 ), result2( 0.0 ), result3( 0.0 ), result4( 0.0 ),
+ result5( 0.0 ), result6( 0.0 ), result7( 0.0 ), result8( 0.0 );
+ Index i( 0 );
+ while( i + 8 < n )
+ {
+ result1 += v1[ i ] * v2[ i ];
+ result2 += v1[ i + 1 ] * v2[ i + 1 ];
+ result3 += v1[ i + 2 ] * v2[ i + 2 ];
+ result4 += v1[ i + 3 ] * v2[ i + 3 ];
+ result5 += v1[ i + 4 ] * v2[ i + 4 ];
+ result6 += v1[ i + 5 ] * v2[ i + 5 ];
+ result7 += v1[ i + 6 ] * v2[ i + 6 ];
+ result8 += v1[ i + 7 ] * v2[ i + 7 ];
+ i += 8;
+ }
+ Real result = result1 + result2 + result3 + result4 + result5 +result6 +result7 +result8;
+ while( i < n )
+ result += v1[ i ] * v2[ i++ ];*/
+ return result;
+}
+
+template< typename Vector1, typename Vector2 >
+void
+VectorOperations< Devices::MIC >::
+addVector( Vector1& y,
+ const Vector2& x,
+ const typename Vector2::RealType& alpha,
+ const typename Vector1::RealType& thisMultiplicator )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+ TNL_ASSERT( x. getSize() > 0, );
+ TNL_ASSERT( x. getSize() == y. getSize(), );
+
+ const Index n = y. getSize();
+ Devices::MICHider<Real> vct;
+ Devices::MICHider<const Real> vct2;
+ vct.pointer=y.getData();
+ vct2.pointer=x.getData();
+ Real a=alpha;
+ Real t=thisMultiplicator;
+
+ #pragma offload target(mic) in(vct,vct2,n,a,t)
+ {
+ for( Index i = 0; i < n; i ++ )
+ vct.pointer[ i ] = t * vct.pointer[ i ] + a * vct2.pointer[ i ];
+ }
+}
+
+template< typename Vector1,
+ typename Vector2,
+ typename Vector3 >
+void
+VectorOperations< Devices::MIC >::
+addVectors( Vector1& v,
+ const Vector2& v1,
+ const typename Vector2::RealType& multiplicator1,
+ const Vector3& v2,
+ const typename Vector3::RealType& multiplicator2,
+ const typename Vector1::RealType& thisMultiplicator )
+{
+ typedef typename Vector1::RealType Real;
+ typedef typename Vector1::IndexType Index;
+ TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT( v.getSize() == v1.getSize(), );
+ TNL_ASSERT( v.getSize() == v2.getSize(), );
+
+ const Index n = v. getSize();
+ Devices::MICHider<Real> vct;
+ Devices::MICHider<const Real> vct1;
+ Devices::MICHider<const Real> vct2;
+ vct.pointer=v.getData();
+ vct1.pointer=v1.getData();
+ vct2.pointer=v2.getData();
+ Real t=thisMultiplicator;
+ Real m1=multiplicator1;
+ Real m2=multiplicator2;
+
+ #pragma offload target(mic) in(vct,vct1,vct2,n,t,m1,m2)
+ {
+ for( Index i = 0; i < n; i ++ )
+ vct.pointer[ i ] = t * vct.pointer[ i ] + m1 * vct1.pointer[ i ] + m2 * vct2.pointer[ i ];
+ }
+}
+
+template< typename Vector >
+void
+VectorOperations< Devices::MIC >::
+computePrefixSum( Vector& v,
+ typename Vector::IndexType begin,
+ typename Vector::IndexType end )
+{
+ typedef typename Vector::IndexType Index;
+
+ //std::cout << v.getSize()<< " " << end <<endl;
+
+ TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT( v.getSize() >= end, );
+ TNL_ASSERT( v.getSize() > begin, );
+ TNL_ASSERT( end > begin, );
+
+ Devices::MICHider<typename Vector::RealType> vct;
+ vct.pointer=v.getData();
+ #pragma offload target(mic) in(vct,begin,end)
+ {
+ for( Index i = begin + 1; i < end; i++ )
+ vct.pointer[ i ] += vct.pointer[ i - 1 ];
+ }
+}
+
+template< typename Vector >
+void
+VectorOperations< Devices::MIC >::
+computeExclusivePrefixSum( Vector& v,
+ typename Vector::IndexType begin,
+ typename Vector::IndexType end )
+{
+ typedef typename Vector::IndexType Index;
+ typedef typename Vector::RealType Real;
+ TNL_ASSERT( v.getSize() > 0, );
+ TNL_ASSERT( v.getSize() >= end, );
+ TNL_ASSERT( v.getSize() > begin, );
+ TNL_ASSERT( begin >= 0, );
+ TNL_ASSERT( end > begin, );
+
+ Devices::MICHider<Real> vct;
+ vct.pointer=v.getData();
+
+ #pragma offload target(mic) in(vct,begin,end)
+ {
+ Real aux( vct.pointer[ begin ] );
+ vct.pointer[ begin ] = 0.0;
+ for( Index i = begin + 1; i < end; i++ )
+ {
+ Real x = vct.pointer[ i ];
+ vct.pointer[ i ] = aux;
+ aux += x;
+ }
+ }
+}
+
+} // namespace Algorithms
+} // namespace Containers
+} // namespace TNL
diff --git a/src/TNL/Containers/Algorithms/cuda-prefix-sum.h b/src/TNL/Containers/Algorithms/cuda-prefix-sum.h
index 2c7ec97c5dee2adea84d8c67bafce27fda4fc3aa..37215a99570676c04a9681ec0540f752e15a26f4 100644
--- a/src/TNL/Containers/Algorithms/cuda-prefix-sum.h
+++ b/src/TNL/Containers/Algorithms/cuda-prefix-sum.h
@@ -14,8 +14,11 @@ namespace TNL {
namespace Containers {
namespace Algorithms {
-enum enumPrefixSumType { exclusivePrefixSum = 0,
- inclusivePrefixSum };
+enum class PrefixSumType
+{
+ exclusive,
+ inclusive
+};
template< typename DataType,
typename Operation,
@@ -25,7 +28,7 @@ bool cudaPrefixSum( const Index size,
const DataType *deviceInput,
DataType* deviceOutput,
const Operation& operation,
- const enumPrefixSumType prefixSumType = inclusivePrefixSum );
+ const PrefixSumType prefixSumType = PrefixSumType::inclusive );
} // namespace Algorithms
} // namespace Containers
diff --git a/src/TNL/Containers/Algorithms/cuda-prefix-sum_impl.h b/src/TNL/Containers/Algorithms/cuda-prefix-sum_impl.h
index 971544a024454ef488db269cbddd3b4c755d055f..85f8eace6944f27ce84e602b5333b153a0010c0d 100644
--- a/src/TNL/Containers/Algorithms/cuda-prefix-sum_impl.h
+++ b/src/TNL/Containers/Algorithms/cuda-prefix-sum_impl.h
@@ -11,7 +11,10 @@
#pragma once
#include <iostream>
+
+#include <TNL/Math.h>
#include <TNL/Devices/Cuda.h>
+#include <TNL/Exceptions/CudaBadAlloc.h>
#include <TNL/Containers/Algorithms/reduction-operations.h>
#ifdef HAVE_CUDA
@@ -23,56 +26,56 @@ namespace Algorithms {
template< typename DataType,
typename Operation,
typename Index >
-__global__ void cudaFirstPhaseBlockPrefixSum( const enumPrefixSumType prefixSumType,
- Operation operation,
- const Index size,
- const Index elementsInBlock,
- const DataType* input,
- DataType* output,
- DataType* auxArray )
+__global__ void
+cudaFirstPhaseBlockPrefixSum( const PrefixSumType prefixSumType,
+ Operation operation,
+ const Index size,
+ const Index elementsInBlock,
+ const DataType* input,
+ DataType* output,
+ DataType* auxArray )
{
DataType* sharedData = TNL::Devices::Cuda::getSharedMemory< DataType >();
DataType* auxData = &sharedData[ elementsInBlock + elementsInBlock / Devices::Cuda::getNumberOfSharedMemoryBanks() + 2 ];
- DataType* warpSums = &auxData[ blockDim. x ];
+ DataType* warpSums = &auxData[ blockDim.x ];
- const Index lastElementIdx = size - blockIdx. x * elementsInBlock;
+ const Index lastElementIdx = size - blockIdx.x * elementsInBlock;
Index lastElementInBlock = ( lastElementIdx < elementsInBlock ?
lastElementIdx : elementsInBlock );
/***
* Load data into the shared memory.
*/
- const Index blockOffset = blockIdx. x * elementsInBlock;
- Index idx = threadIdx. x;
- if( prefixSumType == exclusivePrefixSum )
+ const Index blockOffset = blockIdx.x * elementsInBlock;
+ Index idx = threadIdx.x;
+ if( prefixSumType == PrefixSumType::exclusive )
{
if( idx == 0 )
sharedData[ 0 ] = operation.initialValue();
while( idx < elementsInBlock && blockOffset + idx < size )
{
sharedData[ Devices::Cuda::getInterleaving( idx + 1 ) ] = input[ blockOffset + idx ];
- idx += blockDim. x;
+ idx += blockDim.x;
}
}
else
while( idx < elementsInBlock && blockOffset + idx < size )
{
sharedData[ Devices::Cuda::getInterleaving( idx ) ] = input[ blockOffset + idx ];
- idx += blockDim. x;
+ idx += blockDim.x;
}
/***
* Perform the sequential prefix-sum.
*/
__syncthreads();
- const int chunkSize = elementsInBlock / blockDim. x;
- const int chunkOffset = threadIdx. x * chunkSize;
- const int numberOfChunks = lastElementInBlock / chunkSize +
- ( lastElementInBlock % chunkSize != 0 );
+ const int chunkSize = elementsInBlock / blockDim.x;
+ const int chunkOffset = threadIdx.x * chunkSize;
+ const int numberOfChunks = roundUpDivision( lastElementInBlock, chunkSize );
if( chunkOffset < lastElementInBlock )
{
- auxData[ threadIdx. x ] =
+ auxData[ threadIdx.x ] =
sharedData[ Devices::Cuda::getInterleaving( chunkOffset ) ];
}
@@ -82,22 +85,22 @@ __global__ void cudaFirstPhaseBlockPrefixSum( const enumPrefixSumType prefixSumT
{
operation.commonReductionOnDevice( sharedData[ Devices::Cuda::getInterleaving( chunkOffset + chunkPointer ) ],
sharedData[ Devices::Cuda::getInterleaving( chunkOffset + chunkPointer - 1 ) ] );
- auxData[ threadIdx. x ] =
+ auxData[ threadIdx.x ] =
sharedData[ Devices::Cuda::getInterleaving( chunkOffset + chunkPointer ) ];
- chunkPointer ++;
+ chunkPointer++;
}
/***
* Perform the parallel prefix-sum inside warps.
*/
- const int threadInWarpIdx = threadIdx. x % Devices::Cuda::getWarpSize();
- const int warpIdx = threadIdx. x / Devices::Cuda::getWarpSize();
+ const int threadInWarpIdx = threadIdx.x % Devices::Cuda::getWarpSize();
+ const int warpIdx = threadIdx.x / Devices::Cuda::getWarpSize();
for( int stride = 1; stride < Devices::Cuda::getWarpSize(); stride *= 2 )
- if( threadInWarpIdx >= stride && threadIdx. x < numberOfChunks )
- operation.commonReductionOnDevice( auxData[ threadIdx. x ], auxData[ threadIdx. x - stride ] );
+ if( threadInWarpIdx >= stride && threadIdx.x < numberOfChunks )
+ operation.commonReductionOnDevice( auxData[ threadIdx.x ], auxData[ threadIdx.x - stride ] );
if( threadInWarpIdx == Devices::Cuda::getWarpSize() - 1 )
- warpSums[ warpIdx ] = auxData[ threadIdx. x ];
+ warpSums[ warpIdx ] = auxData[ threadIdx.x ];
__syncthreads();
/****
@@ -113,13 +116,13 @@ __global__ void cudaFirstPhaseBlockPrefixSum( const enumPrefixSumType prefixSumT
* Shift the warp prefix-sums.
*/
if( warpIdx > 0 )
- operation.commonReductionOnDevice( auxData[ threadIdx. x ], warpSums[ warpIdx - 1 ] );
+ operation.commonReductionOnDevice( auxData[ threadIdx.x ], warpSums[ warpIdx - 1 ] );
/***
* Store the result back in global memory.
*/
__syncthreads();
- idx = threadIdx. x;
+ idx = threadIdx.x;
while( idx < elementsInBlock && blockOffset + idx < size )
{
const Index chunkIdx = idx / chunkSize;
@@ -128,49 +131,49 @@ __global__ void cudaFirstPhaseBlockPrefixSum( const enumPrefixSumType prefixSumT
chunkShift = auxData[ chunkIdx - 1 ];
operation.commonReductionOnDevice( sharedData[ Devices::Cuda::getInterleaving( idx ) ], chunkShift );
output[ blockOffset + idx ] = sharedData[ Devices::Cuda::getInterleaving( idx ) ];
- idx += blockDim. x;
+ idx += blockDim.x;
}
__syncthreads();
- if( threadIdx. x == 0 )
+ if( threadIdx.x == 0 )
{
- if( prefixSumType == exclusivePrefixSum )
+ if( prefixSumType == PrefixSumType::exclusive )
{
- /*auxArray[ blockIdx. x ] = operation.commonReductionOnDevice( Devices::Cuda::getInterleaving( lastElementInBlock - 1 ),
+ /*auxArray[ blockIdx.x ] = operation.commonReductionOnDevice( Devices::Cuda::getInterleaving( lastElementInBlock - 1 ),
Devices::Cuda::getInterleaving( lastElementInBlock ),
sharedData );*/
DataType aux = operation.initialValue();
operation.commonReductionOnDevice( aux, sharedData[ Devices::Cuda::getInterleaving( lastElementInBlock - 1 ) ] );
operation.commonReductionOnDevice( aux, sharedData[ Devices::Cuda::getInterleaving( lastElementInBlock ) ] );
- auxArray[ blockIdx. x ] = aux;
+ auxArray[ blockIdx.x ] = aux;
}
else
- auxArray[ blockIdx. x ] = sharedData[ Devices::Cuda::getInterleaving( lastElementInBlock - 1 ) ];
+ auxArray[ blockIdx.x ] = sharedData[ Devices::Cuda::getInterleaving( lastElementInBlock - 1 ) ];
}
-
}
template< typename DataType,
typename Operation,
typename Index >
-__global__ void cudaSecondPhaseBlockPrefixSum( Operation operation,
- const Index size,
- const Index elementsInBlock,
- const Index gridShift,
- const DataType* auxArray,
- DataType* data )
+__global__ void
+cudaSecondPhaseBlockPrefixSum( Operation operation,
+ const Index size,
+ const Index elementsInBlock,
+ const Index gridShift,
+ const DataType* auxArray,
+ DataType* data )
{
- if( blockIdx. x > 0 )
+ if( blockIdx.x > 0 )
{
DataType shift( gridShift );
- operation.commonReductionOnDevice( shift, auxArray[ blockIdx. x - 1 ] );
+ operation.commonReductionOnDevice( shift, auxArray[ blockIdx.x - 1 ] );
- const Index readOffset = blockIdx. x * elementsInBlock;
- Index readIdx = threadIdx. x;
+ const Index readOffset = blockIdx.x * elementsInBlock;
+ Index readIdx = threadIdx.x;
while( readIdx < elementsInBlock && readOffset + readIdx < size )
{
operation.commonReductionOnDevice( data[ readIdx + readOffset ], shift );
- readIdx += blockDim. x;
+ readIdx += blockDim.x;
}
}
}
@@ -179,89 +182,69 @@ __global__ void cudaSecondPhaseBlockPrefixSum( Operation operation,
template< typename DataType,
typename Operation,
typename Index >
-bool cudaRecursivePrefixSum( const enumPrefixSumType prefixSumType,
- Operation& operation,
- const Index size,
- const Index blockSize,
- const Index elementsInBlock,
- const Index gridShift,
- const DataType* input,
- DataType *output )
+void
+cudaRecursivePrefixSum( const PrefixSumType prefixSumType,
+ Operation& operation,
+ const Index size,
+ const Index blockSize,
+ const Index elementsInBlock,
+ const Index gridShift,
+ const DataType* input,
+ DataType *output )
{
- const Index numberOfBlocks = ceil( ( double ) size / ( double ) elementsInBlock );
+ const Index numberOfBlocks = roundUpDivision( size, elementsInBlock );
const Index auxArraySize = numberOfBlocks * sizeof( DataType );
- DataType *auxArray1, *auxArray2;
- if( cudaMalloc( ( void** ) &auxArray1, auxArraySize ) != cudaSuccess ||
- cudaMalloc( ( void** ) &auxArray2, auxArraySize ) != cudaSuccess )
- {
- {
- std::cerr << "Not enough memory on device to allocate auxilliary arrays." << std::endl;
- return false;
- }
- }
+ Array< DataType, Devices::Cuda > auxArray1, auxArray2;
+ auxArray1.setSize( auxArraySize );
+ auxArray2.setSize( auxArraySize );
/****
* Setup block and grid size.
*/
dim3 cudaBlockSize( 0 ), cudaGridSize( 0 );
- cudaBlockSize. x = blockSize;
- cudaGridSize. x = size / elementsInBlock +
- ( size % elementsInBlock != 0 );
+ cudaBlockSize.x = blockSize;
+ cudaGridSize.x = roundUpDivision( size, elementsInBlock );
/****
* Run the kernel.
*/
- size_t sharedDataSize = elementsInBlock +
- elementsInBlock / Devices::Cuda::getNumberOfSharedMemoryBanks() + 2;
- size_t sharedMemory = ( sharedDataSize + blockSize + Devices::Cuda::getWarpSize() ) * sizeof( DataType );
- cudaFirstPhaseBlockPrefixSum< DataType, Operation, Index >
- <<< cudaGridSize, cudaBlockSize, sharedMemory >>>
- ( prefixSumType,
- operation,
- size,
- elementsInBlock,
- input,
- output,
- auxArray1 );
- if( ! checkCudaDevice )
- {
- std::cerr << "The CUDA kernel 'cudaFirstPhaseBlockPrefixSum' ended with error." << std::endl;
- cudaFree( auxArray1 );
- cudaFree( auxArray2 );
- return false;
- }
+ const std::size_t sharedDataSize = elementsInBlock +
+ elementsInBlock / Devices::Cuda::getNumberOfSharedMemoryBanks() + 2;
+ const std::size_t sharedMemory = ( sharedDataSize + blockSize + Devices::Cuda::getWarpSize() ) * sizeof( DataType );
+ cudaFirstPhaseBlockPrefixSum<<< cudaGridSize, cudaBlockSize, sharedMemory >>>
+ ( prefixSumType,
+ operation,
+ size,
+ elementsInBlock,
+ input,
+ output,
+ auxArray1.getData() );
+ TNL_CHECK_CUDA_DEVICE;
/***
* In auxArray1 there is now a sum of numbers in each block.
* We must compute prefix-sum of auxArray1 and then shift
* each block.
*/
- if( numberOfBlocks > 1 &&
- ! cudaRecursivePrefixSum< DataType, Operation, Index >
- ( inclusivePrefixSum,
- operation,
- numberOfBlocks,
- blockSize,
- elementsInBlock,
- 0,
- auxArray1,
- auxArray2 ) )
- return false;
- cudaSecondPhaseBlockPrefixSum< DataType, Operation, Index >
- <<< cudaGridSize, cudaBlockSize >>>
- ( operation, size, elementsInBlock, gridShift, auxArray2, output );
-
- if( ! checkCudaDevice )
- {
- std::cerr << "The CUDA kernel 'cudaSecondPhaseBlockPrefixSum' ended with error." << std::endl;
- cudaFree( auxArray1 );
- cudaFree( auxArray2 );
- return false;
- }
- cudaFree( auxArray1 );
- cudaFree( auxArray2 );
- return true;
+ if( numberOfBlocks > 1 )
+ cudaRecursivePrefixSum( PrefixSumType::inclusive,
+ operation,
+ numberOfBlocks,
+ blockSize,
+ elementsInBlock,
+ (Index) 0,
+ auxArray1.getData(),
+ auxArray2.getData() );
+
+ cudaSecondPhaseBlockPrefixSum<<< cudaGridSize, cudaBlockSize >>>
+ ( operation,
+ size,
+ elementsInBlock,
+ gridShift,
+ auxArray2.getData(),
+ output );
+ TNL_CHECK_CUDA_DEVICE;
}
@@ -269,85 +252,74 @@ bool cudaRecursivePrefixSum( const enumPrefixSumType prefixSumType,
template< typename DataType,
typename Operation,
typename Index >
-bool cudaGridPrefixSum( enumPrefixSumType prefixSumType,
- Operation& operation,
- const Index size,
- const Index blockSize,
- const Index elementsInBlock,
- const DataType *deviceInput,
- DataType *deviceOutput,
- Index& gridShift )
+void
+cudaGridPrefixSum( PrefixSumType prefixSumType,
+ Operation& operation,
+ const Index size,
+ const Index blockSize,
+ const Index elementsInBlock,
+ const DataType *deviceInput,
+ DataType *deviceOutput,
+ Index& gridShift )
{
-
- if( ! cudaRecursivePrefixSum< DataType, Operation, Index >
- ( prefixSumType,
- operation,
- size,
- blockSize,
- elementsInBlock,
- gridShift,
- deviceInput,
- deviceOutput ) )
- return false;
- if( cudaMemcpy( &gridShift,
- &deviceOutput[ size - 1 ],
- sizeof( DataType ),
- cudaMemcpyDeviceToHost ) != cudaSuccess )
- {
- std::cerr << "I am not able to copy data from device to host." << std::endl;
- return false;
- }
- return true;
+ cudaRecursivePrefixSum( prefixSumType,
+ operation,
+ size,
+ blockSize,
+ elementsInBlock,
+ gridShift,
+ deviceInput,
+ deviceOutput );
+
+ cudaMemcpy( &gridShift,
+ &deviceOutput[ size - 1 ],
+ sizeof( DataType ),
+ cudaMemcpyDeviceToHost );
+ TNL_CHECK_CUDA_DEVICE;
}
template< typename DataType,
typename Operation,
typename Index >
-bool cudaPrefixSum( const Index size,
- const Index blockSize,
- const DataType *deviceInput,
- DataType* deviceOutput,
- Operation& operation,
- const enumPrefixSumType prefixSumType )
+void
+cudaPrefixSum( const Index size,
+ const Index blockSize,
+ const DataType *deviceInput,
+ DataType* deviceOutput,
+ Operation& operation,
+ const PrefixSumType prefixSumType )
{
/****
* Compute the number of grids
*/
const Index elementsInBlock = 8 * blockSize;
- const Index gridSize = size / elementsInBlock + ( size % elementsInBlock != 0 );
- const Index maxGridSize = 65536;
- const Index gridsNumber = gridSize / maxGridSize + ( gridSize % maxGridSize != 0 );
+ const Index numberOfBlocks = roundUpDivision( size, elementsInBlock );
+ const auto maxGridSize = Devices::Cuda::getMaxGridSize();
+ const Index numberOfGrids = Devices::Cuda::getNumberOfGrids( numberOfBlocks, maxGridSize );
/****
* Loop over all grids.
*/
- Index gridShift( 0 );
- for( Index gridIdx = 0; gridIdx < gridsNumber; gridIdx ++ )
+ Index gridShift = 0;
+ for( Index gridIdx = 0; gridIdx < numberOfGrids; gridIdx++ )
{
/****
* Compute current grid size and size of data to be scanned
*/
- Index gridSize = ( size - gridIdx * maxGridSize * elementsInBlock ) /
- elementsInBlock;
Index currentSize = size - gridIdx * maxGridSize * elementsInBlock;
- if( gridSize > maxGridSize )
- {
- gridSize = maxGridSize;
+ if( currentSize / elementsInBlock > maxGridSize )
currentSize = maxGridSize * elementsInBlock;
- }
- Index gridOffset = gridIdx * maxGridSize * elementsInBlock;
- if( ! cudaGridPrefixSum< DataType, Operation, Index >
- ( prefixSumType,
- operation,
- currentSize,
- blockSize,
- elementsInBlock,
- &deviceInput[ gridOffset ],
- &deviceOutput[ gridOffset ],
- gridShift ) )
- return false;
+ const Index gridOffset = gridIdx * maxGridSize * elementsInBlock;
+
+ cudaGridPrefixSum( prefixSumType,
+ operation,
+ currentSize,
+ blockSize,
+ elementsInBlock,
+ &deviceInput[ gridOffset ],
+ &deviceOutput[ gridOffset ],
+ gridShift );
}
- return true;
}
#ifdef TEMPLATE_EXPLICIT_INSTANTIATION
@@ -356,7 +328,7 @@ extern template bool cudaPrefixSum( const int size,
const int *deviceInput,
int* deviceOutput,
tnlParallelReductionSum< int, int >& operation,
- const enumPrefixSumType prefixSumType );
+ const PrefixSumType prefixSumType );
extern template bool cudaPrefixSum( const int size,
@@ -364,14 +336,14 @@ extern template bool cudaPrefixSum( const int size,
const float *deviceInput,
float* deviceOutput,
tnlParallelReductionSum< float, int >& operation,
- const enumPrefixSumType prefixSumType );
+ const PrefixSumType prefixSumType );
extern template bool cudaPrefixSum( const int size,
const int blockSize,
const double *deviceInput,
double* deviceOutput,
tnlParallelReductionSum< double, int >& operation,
- const enumPrefixSumType prefixSumType );
+ const PrefixSumType prefixSumType );
#ifdef INSTANTIATE_LONG_DOUBLE
extern template bool cudaPrefixSum( const int size,
@@ -379,7 +351,7 @@ extern template bool cudaPrefixSum( const int size,
const long double *deviceInput,
long double* deviceOutput,
tnlParallelReductionSum< long double, int >& operation,
- const enumPrefixSumType prefixSumType );
+ const PrefixSumType prefixSumType );
#endif
#ifdef INSTANTIATE_LONG_INT
@@ -388,7 +360,7 @@ extern template bool cudaPrefixSum( const long int size,
const int *deviceInput,
int* deviceOutput,
tnlParallelReductionSum< int, long int >& operation,
- const enumPrefixSumType prefixSumType );
+ const PrefixSumType prefixSumType );
extern template bool cudaPrefixSum( const long int size,
@@ -396,14 +368,14 @@ extern template bool cudaPrefixSum( const long int size,
const float *deviceInput,
float* deviceOutput,
tnlParallelReductionSum< float, long int >& operation,
- const enumPrefixSumType prefixSumType );
+ const PrefixSumType prefixSumType );
extern template bool cudaPrefixSum( const long int size,
const long int blockSize,
const double *deviceInput,
double* deviceOutput,
tnlParallelReductionSum< double, long int >& operation,
- const enumPrefixSumType prefixSumType );
+ const PrefixSumType prefixSumType );
#ifdef INSTANTIATE_LONG_DOUBLE
extern template bool cudaPrefixSum( const long int size,
@@ -411,7 +383,7 @@ extern template bool cudaPrefixSum( const long int size,
const long double *deviceInput,
long double* deviceOutput,
tnlParallelReductionSum< long double, long int >& operation,
- const enumPrefixSumType prefixSumType );
+ const PrefixSumType prefixSumType );
#endif
#endif
diff --git a/src/TNL/Containers/Array.h b/src/TNL/Containers/Array.h
index 92a3ccd6fd519b50435751e3753218b7060778a4..89e72cb8b4f87731f2b31065901bde59c4954510 100644
--- a/src/TNL/Containers/Array.h
+++ b/src/TNL/Containers/Array.h
@@ -8,7 +8,7 @@
/* See Copyright Notice in tnl/Copyright */
-#pragma once
+#pragma once
#include <TNL/Object.h>
#include <TNL/File.h>
@@ -35,18 +35,17 @@ class Array : public virtual Object
typedef Index IndexType;
typedef Containers::Array< Element, Devices::Host, Index > HostType;
typedef Containers::Array< Element, Devices::Cuda, Index > CudaType;
- typedef Containers::Array< Element, Device, Index > ThisType;
-
+
Array();
-
+
Array( const IndexType& size );
-
+
Array( Element* data,
- const IndexType& size );
+ const IndexType& size );
- Array( Array< Element, Device, Index >& array,
- const IndexType& begin = 0,
- const IndexType& size = 0 );
+ Array( Array& array,
+ const IndexType& begin = 0,
+ const IndexType& size = 0 );
static String getType();
@@ -61,28 +60,28 @@ class Array : public virtual Object
* these data are released. If the current data are not shared and the current
* size is the same as the new one, nothing happens.
*/
- bool setSize( Index size );
+ void setSize( Index size );
- template< typename Array >
- bool setLike( const Array& array );
+ __cuda_callable__ Index getSize() const;
+
+ template< typename ArrayT >
+ void setLike( const ArrayT& array );
void bind( Element* _data,
const Index _size );
- template< typename Array >
- void bind( const Array& array,
+ template< typename ArrayT >
+ void bind( const ArrayT& array,
const IndexType& begin = 0,
const IndexType& size = 0 );
template< int Size >
void bind( StaticArray< Size, Element >& array );
- void swap( Array< Element, Device, Index >& array );
+ void swap( Array& array );
void reset();
- __cuda_callable__ Index getSize() const;
-
void setElement( const Index& i, const Element& x );
Element getElement( const Index& i ) const;
@@ -91,10 +90,10 @@ class Array : public virtual Object
__cuda_callable__ inline const Element& operator[] ( const Index& i ) const;
- Array< Element, Device, Index >& operator = ( const Array< Element, Device, Index >& array );
+ Array& operator = ( const Array& array );
template< typename ArrayT >
- Array< Element, Device, Index >& operator = ( const ArrayT& array );
+ Array& operator = ( const ArrayT& array );
template< typename ArrayT >
bool operator == ( const ArrayT& array ) const;
@@ -118,20 +117,15 @@ class Array : public virtual Object
* Every time one touches this grid touches * size * sizeof( Real ) bytes are added
* to transfered bytes in tnlStatistics.
*/
- #ifdef HAVE_NOT_CXX11
- template< typename IndexType2 >
- void touch( IndexType2 touches = 1 ) const;
- #else
template< typename IndexType2 = Index >
void touch( IndexType2 touches = 1 ) const;
- #endif
//! Method for saving the object to a file as a binary data.
bool save( File& file ) const;
//! Method for loading the object from a file as a binary data.
bool load( File& file );
-
+
//! This method loads data without reallocation.
/****
* This is useful for loading data into shared arrays.
@@ -140,17 +134,17 @@ class Array : public virtual Object
* the size of array being loaded.
*/
bool boundLoad( File& file );
-
- bool boundLoad( const String& fileName );
-
- using Object::load;
using Object::save;
+ using Object::load;
+
+ using Object::boundLoad;
+
~Array();
protected:
-
+
void releaseData() const;
//!Number of elements in array
@@ -183,4 +177,3 @@ std::ostream& operator << ( std::ostream& str, const Array< Element, Device, Ind
} // namespace TNL
#include <TNL/Containers/Array_impl.h>
-
diff --git a/src/TNL/Containers/Array_impl.cpp b/src/TNL/Containers/Array_impl.cpp
index 966b0e5f35292a6a0a000b5b030108d67769b4be..c951e0526badefc0806979821e2b0decfbdc2cdf 100644
--- a/src/TNL/Containers/Array_impl.cpp
+++ b/src/TNL/Containers/Array_impl.cpp
@@ -8,13 +8,13 @@
/* See Copyright Notice in tnl/Copyright */
+#ifdef TEMPLATE_EXPLICIT_INSTANTIATION
+
#include <TNL/Containers/Array.h>
namespace TNL {
namespace Containers {
-#ifdef TEMPLATE_EXPLICIT_INSTANTIATION
-
#ifdef INSTANTIATE_FLOAT
template class Array< float, Devices::Host, int >;
#endif
@@ -55,7 +55,7 @@ template class Array< long double, Devices::Cuda, long int >;
#endif
-#endif
-
} // namespace Containers
-} // namespace TNL
\ No newline at end of file
+} // namespace TNL
+
+#endif // #ifdef TEMPLATE_EXPLICIT_INSTANTIATION
diff --git a/src/TNL/Containers/Array_impl.cu b/src/TNL/Containers/Array_impl.cu
index b14e5c30d96f7987b87595afff37d0da2ae00ac7..5c43ec36e69ccb02b2b9950d85e27f9271192b3d 100644
--- a/src/TNL/Containers/Array_impl.cu
+++ b/src/TNL/Containers/Array_impl.cu
@@ -8,13 +8,13 @@
/* See Copyright Notice in tnl/Copyright */
+#ifdef TEMPLATE_EXPLICIT_INSTANTIATION
+
#include <TNL/Containers/Array.h>
namespace TNL {
namespace Containers {
-#ifdef TEMPLATE_EXPLICIT_INSTANTIATION
-
#ifdef HAVE_CUDA
#ifdef INSTANTIATE_FLOAT
template class Array< float, Devices::Cuda, int >;
@@ -37,7 +37,7 @@ template class Array< long double, Devices::Cuda, long int >;
#endif
-#endif
-
} // namespace Containers
} // namespace TNL
+
+#endif // #ifdef TEMPLATE_EXPLICIT_INSTANTIATION
diff --git a/src/TNL/Containers/Array_impl.h b/src/TNL/Containers/Array_impl.h
index e9d34f860fa6c0dcff23e9847c311a3fe4d85e45..52504e47a31f057adf878c86e32891a1210daa7c 100644
--- a/src/TNL/Containers/Array_impl.h
+++ b/src/TNL/Containers/Array_impl.h
@@ -20,9 +20,7 @@
#include <TNL/Containers/Array.h>
namespace TNL {
-namespace Containers {
-
-using namespace std;
+namespace Containers {
template< typename Element,
typename Device,
@@ -34,7 +32,7 @@ Array()
allocationPointer( 0 ),
referenceCounter( 0 )
{
-};
+}
template< typename Element,
typename Device,
@@ -74,10 +72,10 @@ Array( Array< Element, Device, Index >& array,
allocationPointer( array.allocationPointer ),
referenceCounter( 0 )
{
- TNL_ASSERT( begin < array.getSize(),
- std::cerr << " begin = " << begin << " array.getSize() = " << array.getSize() );
- TNL_ASSERT( begin + size < array.getSize(),
- std::cerr << " begin = " << begin << " size = " << size << " array.getSize() = " << array.getSize() );
+ TNL_ASSERT_TRUE( array.getData(), "Empty arrays cannot be bound." );
+ TNL_ASSERT_LT( begin, array.getSize(), "Begin of array is out of bounds." );
+ TNL_ASSERT_LE( begin + size, array.getSize(), "End of array is out of bounds." );
+
if( ! this->size )
this->size = array.getSize() - begin;
if( array.allocationPointer )
@@ -105,7 +103,7 @@ getType()
TNL::getType< Element >() + ", " +
Device::getDeviceType() + ", " +
TNL::getType< Index >() + " >";
-};
+}
template< typename Element,
typename Device,
@@ -115,7 +113,7 @@ Array< Element, Device, Index >::
getTypeVirtual() const
{
return this->getType();
-};
+}
template< typename Element,
typename Device,
@@ -125,7 +123,7 @@ Array< Element, Device, Index >::
getSerializationType()
{
return HostType::getType();
-};
+}
template< typename Element,
typename Device,
@@ -135,7 +133,7 @@ Array< Element, Device, Index >::
getSerializationTypeVirtual() const
{
return this->getSerializationType();
-};
+}
template< typename Element,
typename Device,
@@ -165,41 +163,48 @@ releaseData() const
template< typename Element,
typename Device,
typename Index >
-bool
+void
Array< Element, Device, Index >::
setSize( const Index size )
{
- TNL_ASSERT( size >= 0,
- std::cerr << "You try to set size of Array to negative value."
- << "New size: " << size << std::endl );
- if( this->size == size && allocationPointer && ! referenceCounter ) return true;
+ TNL_ASSERT_GE( size, 0, "Array size must be non-negative." );
+
+ if( this->size == size && allocationPointer && ! referenceCounter )
+ return;
this->releaseData();
- Algorithms::ArrayOperations< Device >::allocateMemory( this->allocationPointer, size );
- this->data = this->allocationPointer;
- this->size = size;
- if( size > 0 && ! this->allocationPointer )
- {
- cerr << "I am not able to allocate new array with size "
- << ( double ) this->size * sizeof( ElementType ) / 1.0e9 << " GB." << endl;
- this -> size = 0;
- return false;
+
+ // Allocating zero bytes is useless. Moreover, the allocators don't behave the same way:
+ // "operator new" returns some non-zero address, the latter returns a null pointer.
+ if( size > 0 ) {
+ Algorithms::ArrayOperations< Device >::allocateMemory( this->allocationPointer, size );
+ this->data = this->allocationPointer;
+ this->size = size;
+ TNL_ASSERT_TRUE( this->allocationPointer,
+ "This should never happen - allocator did not throw on an error." );
}
- return true;
-};
+}
+
+template< typename Element,
+ typename Device,
+ typename Index >
+__cuda_callable__
+Index
+Array< Element, Device, Index >::
+getSize() const
+{
+ return this -> size;
+}
template< typename Element,
typename Device,
typename Index >
template< typename ArrayT >
-bool
+void
Array< Element, Device, Index >::
setLike( const ArrayT& array )
{
- TNL_ASSERT( array. getSize() >= 0,
- std::cerr << "You try to set size of Array to negative value."
- << "Array size: " << array. getSize() << std::endl );
- return setSize( array.getSize() );
-};
+ setSize( array.getSize() );
+}
template< typename Element,
typename Device,
@@ -209,6 +214,7 @@ Array< Element, Device, Index >::
bind( Element* data,
const Index size )
{
+ TNL_ASSERT_TRUE( data, "Null pointer cannot be bound." );
this->releaseData();
this->data = data;
this->size = size;
@@ -228,11 +234,10 @@ bind( const ArrayT& array,
static_assert( std::is_same< Element, typename ArrayT::ElementType >::value, "ElementType of both arrays must be the same." );
static_assert( std::is_same< Device, typename ArrayT::DeviceType >::value, "DeviceType of both arrays must be the same." );
static_assert( std::is_same< Index, typename ArrayT::IndexType >::value, "IndexType of both arrays must be the same." );
- TNL_ASSERT( begin <= array.getSize(),
- std::cerr << " begin = " << begin << " array.getSize() = " << array.getSize() );
- TNL_ASSERT( begin + size <= array.getSize(),
- std::cerr << " begin = " << begin << " size = " << size << " array.getSize() = " << array.getSize() );
-
+ TNL_ASSERT_TRUE( array.getData(), "Empty array cannot be bound." );
+ TNL_ASSERT_LT( begin, array.getSize(), "Begin of array is out of bounds." );
+ TNL_ASSERT_LE( begin + size, array.getSize(), "End of array is out of bounds." );
+
this->releaseData();
if( size )
this->size = size;
@@ -249,8 +254,7 @@ bind( const ArrayT& array,
}
else
{
- this->referenceCounter = array.referenceCounter = new int;
- *this->referenceCounter = 2;
+ this->referenceCounter = array.referenceCounter = new int( 2 );
//std::cerr << "Allocating reference counter " << this->referenceCounter << std::endl;
}
}
@@ -281,7 +285,7 @@ swap( Array< Element, Device, Index >& array )
TNL::swap( this->data, array.data );
TNL::swap( this->allocationPointer, array.allocationPointer );
TNL::swap( this->referenceCounter, array.referenceCounter );
-};
+}
template< typename Element,
typename Device,
@@ -291,17 +295,6 @@ Array< Element, Device, Index >::
reset()
{
this->releaseData();
-};
-
-template< typename Element,
- typename Device,
- typename Index >
-__cuda_callable__
-Index
-Array< Element, Device, Index >::
-getSize() const
-{
- return this -> size;
}
template< typename Element,
@@ -311,12 +304,10 @@ void
Array< Element, Device, Index >::
setElement( const Index& i, const Element& x )
{
- TNL_ASSERT( 0 <= i && i < this->getSize(),
- std::cerr << "Wrong index for setElement method in Array "
- << " index is " << i
- << " and array size is " << this->getSize() );
- return Algorithms::ArrayOperations< Device > :: setMemoryElement( &( this->data[ i ] ), x );
-};
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, this->getSize(), "Element index is out of bounds." );
+ return Algorithms::ArrayOperations< Device >::setMemoryElement( &( this->data[ i ] ), x );
+}
template< typename Element,
typename Device,
@@ -325,12 +316,10 @@ Element
Array< Element, Device, Index >::
getElement( const Index& i ) const
{
- TNL_ASSERT( 0 <= i && i < this->getSize(),
- std::cerr << "Wrong index for getElement method in Array "
- << " index is " << i
- << " and array size is " << this->getSize() );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, this->getSize(), "Element index is out of bounds." );
return Algorithms::ArrayOperations< Device >::getMemoryElement( & ( this->data[ i ] ) );
-};
+}
template< typename Element,
typename Device,
@@ -340,12 +329,10 @@ inline Element&
Array< Element, Device, Index >::
operator[] ( const Index& i )
{
- TNL_ASSERT( 0 <= i && i < this->getSize(),
- std::cerr << "Wrong index for operator[] in Array "
- << " index is " << i
- << " and array size is " << this->getSize() );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, this->getSize(), "Element index is out of bounds." );
return this->data[ i ];
-};
+}
template< typename Element,
typename Device,
@@ -355,12 +342,10 @@ inline const Element&
Array< Element, Device, Index >::
operator[] ( const Index& i ) const
{
- TNL_ASSERT( 0 <= i && i < this->getSize(),
- std::cerr << "Wrong index for operator[] in Array "
- << " index is " << i
- << " and array size is " << this->getSize() );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, this->getSize(), "Element index is out of bounds." );
return this->data[ i ];
-};
+}
template< typename Element,
typename Device,
@@ -369,19 +354,19 @@ Array< Element, Device, Index >&
Array< Element, Device, Index >::
operator = ( const Array< Element, Device, Index >& array )
{
- TNL_ASSERT( array. getSize() == this->getSize(),
- std::cerr << "Source size: " << array. getSize() << std::endl
- << "Target size: " << this->getSize() << std::endl );
+ //TNL_ASSERT_EQ( array.getSize(), this->getSize(), "Array sizes must be the same." );
+ if( this->getSize() != array.getSize() )
+ this->setLike( array );
if( this->getSize() > 0 )
Algorithms::ArrayOperations< Device >::
template copyMemory< Element,
Element,
Index >
( this->getData(),
- array. getData(),
- array. getSize() );
+ array.getData(),
+ array.getSize() );
return ( *this );
-};
+}
template< typename Element,
typename Device,
@@ -391,19 +376,19 @@ Array< Element, Device, Index >&
Array< Element, Device, Index >::
operator = ( const ArrayT& array )
{
- TNL_ASSERT( array. getSize() == this->getSize(),
- std::cerr << "Source size: " << array. getSize() << std::endl
- << "Target size: " << this->getSize() << std::endl );
+ //TNL_ASSERT_EQ( array.getSize(), this->getSize(), "Array sizes must be the same." );
+ if( this->getSize() != array.getSize() )
+ this->setLike( array );
if( this->getSize() > 0 )
Algorithms::ArrayOperations< Device, typename ArrayT::DeviceType >::
template copyMemory< Element,
typename ArrayT::ElementType,
typename ArrayT::IndexType >
( this->getData(),
- array. getData(),
- array. getSize() );
+ array.getData(),
+ array.getSize() );
return ( *this );
-};
+}
template< typename Element,
typename Device,
@@ -413,7 +398,7 @@ bool
Array< Element, Device, Index >::
operator == ( const ArrayT& array ) const
{
- if( array. getSize() != this -> getSize() )
+ if( array.getSize() != this->getSize() )
return false;
if( this->getSize() == 0 )
return true;
@@ -430,7 +415,7 @@ template< typename Element,
typename Device,
typename Index >
template< typename ArrayT >
-bool Array< Element, Device, Index > :: operator != ( const ArrayT& array ) const
+bool Array< Element, Device, Index >::operator != ( const ArrayT& array ) const
{
return ! ( ( *this ) == array );
}
@@ -439,9 +424,9 @@ bool Array< Element, Device, Index > :: operator != ( const ArrayT& array ) cons
template< typename Element,
typename Device,
typename Index >
-void Array< Element, Device, Index > :: setValue( const Element& e )
+void Array< Element, Device, Index >::setValue( const Element& e )
{
- TNL_ASSERT( this->getData(),);
+ TNL_ASSERT_TRUE( this->getData(), "Attempted to set a value of an empty array." );
Algorithms::ArrayOperations< Device >::setMemory( this->getData(), e, this->getSize() );
}
@@ -449,7 +434,7 @@ template< typename Element,
typename Device,
typename Index >
__cuda_callable__
-const Element* Array< Element, Device, Index > :: getData() const
+const Element* Array< Element, Device, Index >::getData() const
{
return this -> data;
}
@@ -458,7 +443,7 @@ template< typename Element,
typename Device,
typename Index >
__cuda_callable__
-Element* Array< Element, Device, Index > :: getData()
+Element* Array< Element, Device, Index >::getData()
{
return this -> data;
}
@@ -466,43 +451,38 @@ Element* Array< Element, Device, Index > :: getData()
template< typename Element,
typename Device,
typename Index >
-Array< Element, Device, Index > :: operator bool() const
+Array< Element, Device, Index >::operator bool() const
{
return data != 0;
-};
+}
template< typename Element,
typename Device,
typename Index >
template< typename IndexType2 >
-void Array< Element, Device, Index > :: touch( IndexType2 touches ) const
+void Array< Element, Device, Index >::touch( IndexType2 touches ) const
{
//TODO: implement
-};
+}
template< typename Element,
typename Device,
typename Index >
-bool Array< Element, Device, Index > :: save( File& file ) const
+bool Array< Element, Device, Index >::save( File& file ) const
{
- if( ! Object :: save( file ) )
- return false;
-#ifdef HAVE_NOT_CXX11
- if( ! file. write< const Index, Devices::Host >( &this->size ) )
+ if( ! Object::save( file ) )
return false;
-#else
- if( ! file. write( &this->size ) )
+ if( ! file.write( &this->size ) )
return false;
-#endif
if( this->size != 0 && ! ArrayIO< Element, Device, Index >::save( file, this->data, this->size ) )
{
- cerr << "I was not able to save " << this->getType()
- << " with size " << this -> getSize() << endl;
+ std::cerr << "I was not able to save " << this->getType()
+ << " with size " << this -> getSize() << std::endl;
return false;
}
return true;
-};
+}
template< typename Element,
typename Device,
@@ -511,19 +491,17 @@ bool
Array< Element, Device, Index >::
load( File& file )
{
- if( ! Object :: load( file ) )
+ if( ! Object::load( file ) )
return false;
Index _size;
-#ifdef HAVE_NOT_CXX11
- if( ! file. read< Index, Devices::Host >( &_size ) )
- return false;
-#else
- if( ! file. read( &_size ) )
+ if( ! file.read( &_size ) )
+ {
+ std::cerr << "Unable to read the array size." << std::endl;
return false;
-#endif
+ }
if( _size < 0 )
{
- cerr << "Error: The size " << _size << " of the file is not a positive number or zero." << endl;
+ std::cerr << "Error: The size " << _size << " of the file is not a positive number or zero." << std::endl;
return false;
}
setSize( _size );
@@ -531,8 +509,8 @@ load( File& file )
{
if( ! ArrayIO< Element, Device, Index >::load( file, this->data, this->size ) )
{
- cerr << "I was not able to load " << this->getType()
- << " with size " << this -> getSize() << endl;
+ std::cerr << "I was not able to load " << this->getType()
+ << " with size " << this -> getSize() << std::endl;
return false;
}
}
@@ -546,19 +524,14 @@ bool
Array< Element, Device, Index >::
boundLoad( File& file )
{
- if( ! Object :: load( file ) )
+ if( ! Object::load( file ) )
return false;
Index _size;
-#ifdef HAVE_NOT_CXX11
- if( ! file. read< Index, Devices::Host >( &_size ) )
- return false;
-#else
- if( ! file. read( &_size ) )
+ if( ! file.read( &_size ) )
return false;
-#endif
if( _size < 0 )
{
- cerr << "Error: The size " << _size << " of the file is not a positive number or zero." << endl;
+ std::cerr << "Error: The size " << _size << " of the file is not a positive number or zero." << std::endl;
return false;
}
if( this->getSize() != 0 )
@@ -575,37 +548,14 @@ boundLoad( File& file )
{
if( ! ArrayIO< Element, Device, Index >::load( file, this->data, this->size ) )
{
- cerr << "I was not able to load " << this->getType()
- << " with size " << this -> getSize() << endl;
+ std::cerr << "I was not able to load " << this->getType()
+ << " with size " << this -> getSize() << std::endl;
return false;
}
}
return true;
}
-template< typename Element,
- typename Device,
- typename Index >
-bool
-Array< Element, Device, Index >::
-boundLoad( const String& fileName )
-{
- File file;
- if( ! file. open( fileName, tnlReadMode ) )
- {
- cerr << "I am not bale to open the file " << fileName << " for reading." << endl;
- return false;
- }
- if( ! this->boundLoad( file ) )
- return false;
- if( ! file. close() )
- {
- cerr << "An error occurred when I was closing the file " << fileName << "." << endl;
- return false;
- }
- return true;
-}
-
template< typename Element,
typename Device,
typename Index >
@@ -623,7 +573,7 @@ std::ostream& operator << ( std::ostream& str, const Array< Element, Device, Ind
{
str << v.getElement( 0 );
for( Index i = 1; i < v.getSize(); i++ )
- str << ", " << v. getElement( i );
+ str << ", " << v.getElement( i );
}
str << " ]";
return str;
diff --git a/src/TNL/Containers/CMakeLists.txt b/src/TNL/Containers/CMakeLists.txt
old mode 100755
new mode 100644
index b35fa285d9623457ddd5c45a887428a55161fde1..5ec11514d37db3dbbfbf00a952a909985f8de0c2
--- a/src/TNL/Containers/CMakeLists.txt
+++ b/src/TNL/Containers/CMakeLists.txt
@@ -4,6 +4,8 @@ set( headers Array.h
Array_impl.h
ArrayIO.h
DynamicTypeTag.h
+ IndexedMap.h
+ IndexedMap_impl.h
IndexedSet.h
IndexedSet_impl.h
List.h
diff --git a/src/TNL/Containers/ConstSharedArray.h b/src/TNL/Containers/ConstSharedArray.h
index d9435f843739dd6e3b300a590adaabea198eeb99..ec719634f604e95ac92f06ca2259f6990972ba2e 100644
--- a/src/TNL/Containers/ConstSharedArray.h
+++ b/src/TNL/Containers/ConstSharedArray.h
@@ -91,13 +91,8 @@ class tnlConstSharedArray : public Object
* Every time one touches this grid touches * size * sizeof( Real ) bytes are added
* to transfered bytes in tnlStatistics.
*/
-#ifdef HAVE_NOT_CXX11
- template< typename IndexType2 >
- void touch( IndexType2 touches = 1 ) const;
-#else
template< typename IndexType2 = Index >
void touch( IndexType2 touches = 1 ) const;
-#endif
//! Method for saving the object to a file as a binary data.
bool save( File& file ) const;
diff --git a/src/TNL/Containers/ConstSharedArray_impl.h b/src/TNL/Containers/ConstSharedArray_impl.h
index 93b662269f064e45c8dfb5b7024f0f9fbdb27c4f..fde53b35807267d85495f34d0d90ca38f69ba358 100644
--- a/src/TNL/Containers/ConstSharedArray_impl.h
+++ b/src/TNL/Containers/ConstSharedArray_impl.h
@@ -232,11 +232,7 @@ bool tnlConstSharedArray< Element, Device, Index > :: save( File& file ) const
std::cerr << "You try to save empty array." );
if( ! Object :: save( file ) )
return false;
-#ifdef HAVE_NOT_CXX11
- if( ! file. write< const Index, Device >( &this->size ) )
-#else
if( ! file. write( &this->size ) )
-#endif
return false;
if( ! file. write< Element, Device, Index >( this->data, this->size ) )
{
diff --git a/src/TNL/Containers/IndexedMap.h b/src/TNL/Containers/IndexedMap.h
new file mode 100644
index 0000000000000000000000000000000000000000..2c0e2d99471726fd08ee4c2529c8ca64ce63925d
--- /dev/null
+++ b/src/TNL/Containers/IndexedMap.h
@@ -0,0 +1,82 @@
+/***************************************************************************
+ IndexedMap.h - description
+ -------------------
+ begin : Feb 15, 2014
+ copyright : (C) 2014 by Tomas Oberhuber
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#pragma once
+
+#include <map>
+#include <stdexcept>
+
+namespace TNL {
+namespace Containers {
+
+template< typename Element,
+ typename Index,
+ typename Key >
+class IndexedMap
+{
+ public:
+
+ typedef Element ElementType;
+ typedef Index IndexType;
+ typedef Key KeyType;
+
+ void reset();
+
+ IndexType getSize() const;
+
+ IndexType insert( const ElementType &data );
+
+ bool find( const ElementType &data, IndexType& index ) const;
+
+ template< typename ArrayType >
+ void toArray( ArrayType& array ) const;
+
+ const Element& getElement( KeyType key ) const;
+
+ Element& getElement( KeyType key );
+
+ void print( std::ostream& str ) const;
+
+ protected:
+
+ struct DataWithIndex
+ {
+ // This constructor is here only because of bug in g++, we might fix it later.
+ // http://stackoverflow.com/questions/22357887/comparing-two-mapiterators-why-does-it-need-the-copy-constructor-of-stdpair
+ DataWithIndex(){};
+
+ DataWithIndex( const DataWithIndex& d ) : data( d.data ), index( d.index) {}
+
+ explicit DataWithIndex( const Element data) : data( data ) {}
+
+ DataWithIndex( const Element data,
+ const Index index) : data(data), index(index) {}
+
+ Element data;
+ Index index;
+ };
+
+ typedef std::map< Key, DataWithIndex > STDMapType;
+ typedef typename STDMapType::value_type STDMapValueType;
+ typedef typename STDMapType::const_iterator STDMapIteratorType;
+
+ STDMapType map;
+
+};
+
+template< typename Element,
+ typename Index,
+ typename Key >
+std::ostream& operator <<( std::ostream& str, IndexedMap< Element, Index, Key >& set );
+
+} // namespace Containers
+} // namespace TNL
+
+#include <TNL/Containers/IndexedMap_impl.h>
diff --git a/src/TNL/Containers/IndexedMap_impl.h b/src/TNL/Containers/IndexedMap_impl.h
new file mode 100644
index 0000000000000000000000000000000000000000..2adb8872cd488d1ea2448024cff3521928b531a4
--- /dev/null
+++ b/src/TNL/Containers/IndexedMap_impl.h
@@ -0,0 +1,111 @@
+/***************************************************************************
+ IndexedMap_impl.h - description
+ -------------------
+ begin : Feb 15, 2014
+ copyright : (C) 2014 by Tomas Oberhuber
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#pragma once
+
+namespace TNL {
+namespace Containers {
+
+template< typename Element,
+ typename Index,
+ typename Key >
+void IndexedMap< Element, Index, Key >::reset()
+{
+ map.clear();
+}
+
+template< typename Element,
+ typename Index,
+ typename Key >
+Index IndexedMap< Element, Index, Key >::getSize() const
+{
+ return map.size();
+}
+
+template< typename Element,
+ typename Index,
+ typename Key >
+Index IndexedMap< Element, Index, Key >::insert( const Element &data )
+{
+ STDMapIteratorType iter = map.insert( STDMapValueType( Key( data ),
+ DataWithIndex( data, getSize() ) ) ).first;
+ return iter->second.index;
+}
+
+template< typename Element,
+ typename Index,
+ typename Key >
+bool IndexedMap< Element, Index, Key >::find( const Element &data, Index& index ) const
+{
+ STDMapIteratorType iter = map.find( Key( data ) );
+ if (iter == map.end())
+ return false;
+ index = iter->second.index;
+ return true;
+}
+
+template< typename Element,
+ typename Index,
+ typename Key >
+ template<typename ArrayType>
+void IndexedMap< Element, Index, Key >::toArray( ArrayType& array ) const
+{
+ Assert( array.getSize() == getSize(),
+ std::cerr << "array.getSize() = " << array.getSize()
+ << " getSize() = " << getSize() );
+
+ for( STDMapIteratorType iter = map.begin();
+ iter != map.end();
+ ++iter)
+ array[ iter->second.index ] = iter->second.data;
+}
+
+template< typename Element,
+ typename Index,
+ typename Key >
+const Element& IndexedMap< Element, Index, Key >::getElement( KeyType key ) const
+{
+ return map[ key ];
+}
+
+template< typename Element,
+ typename Index,
+ typename Key >
+Element& IndexedMap< Element, Index, Key >::getElement( KeyType key )
+{
+ return map[ key ];
+}
+
+template< typename Element,
+ typename Index,
+ typename Key >
+void IndexedMap< Element, Index, Key >::print( std::ostream& str ) const
+{
+ STDMapIteratorType iter = map.begin();
+ str << iter->second.data;
+ iter++;
+ while( iter != map.end() )
+ {
+ str << ", " << iter->second.data;
+ iter++;
+ }
+}
+
+template< typename Element,
+ typename Index,
+ typename Key >
+std::ostream& operator<<( std::ostream& str, IndexedMap< Element, Index, Key >& set )
+{
+ set.print( str );
+ return str;
+}
+
+} // namespace Containers
+} // namespace TNL
diff --git a/src/TNL/Containers/IndexedSet.h b/src/TNL/Containers/IndexedSet.h
index e2c7e887ec534ba206c559aab73e684b14a793dd..087918a516999ec8a9d012c1aa7e98d2f412a02a 100644
--- a/src/TNL/Containers/IndexedSet.h
+++ b/src/TNL/Containers/IndexedSet.h
@@ -11,70 +11,45 @@
#pragma once
#include <map>
-#include <stdexcept>
+#include <ostream>
namespace TNL {
namespace Containers {
-template< typename Element,
- typename Index,
- typename Key >
+template< class Key,
+ class Index,
+ class Compare = std::less< Key >,
+ class Allocator = std::allocator< std::pair< const Key, Index > > >
class IndexedSet
{
- public:
+protected:
+ using map_type = std::map< Key, Index, Compare, Allocator >;
+ map_type map;
- typedef Element ElementType;
- typedef Index IndexType;
- typedef Key KeyType;
+public:
+ using key_type = Key;
+ using index_type = Index;
+ using value_type = typename map_type::value_type;
+ using size_type = typename map_type::size_type;
- void reset();
+ void clear();
- IndexType getSize() const;
+ size_type size() const;
- IndexType insert( const ElementType &data );
+ Index insert( const Key& key );
- bool find( const ElementType &data, IndexType& index ) const;
+ bool find( const Key& key, Index& index ) const;
- template< typename ArrayType >
- void toArray( ArrayType& array ) const;
+ size_type count( const Key& key ) const;
- const Element& getElement( KeyType key ) const;
+ size_type erase( const Key& key );
- Element& getElement( KeyType key );
-
void print( std::ostream& str ) const;
-
- protected:
-
- struct DataWithIndex
- {
- // This constructor is here only because of bug in g++, we might fix it later.
- // http://stackoverflow.com/questions/22357887/comparing-two-mapiterators-why-does-it-need-the-copy-constructor-of-stdpair
- DataWithIndex(){};
-
- DataWithIndex( const DataWithIndex& d ) : data( d.data ), index( d.index) {}
-
- explicit DataWithIndex( const Element data) : data( data ) {}
-
- DataWithIndex( const Element data,
- const Index index) : data(data), index(index) {}
-
- Element data;
- Index index;
- };
-
- typedef std::map< Key, DataWithIndex > STDMapType;
- typedef typename STDMapType::value_type STDMapValueType;
- typedef typename STDMapType::const_iterator STDMapIteratorType;
-
- STDMapType map;
-
};
template< typename Element,
- typename Index,
- typename Key >
-std::ostream& operator <<( std::ostream& str, IndexedSet< Element, Index, Key >& set );
+ typename Index >
+std::ostream& operator <<( std::ostream& str, IndexedSet< Element, Index >& set );
} // namespace Containers
} // namespace TNL
diff --git a/src/TNL/Containers/IndexedSet_impl.h b/src/TNL/Containers/IndexedSet_impl.h
index 54c326f65bb24c877fd001c51c8a558cabebc27b..12ac49dba43c8da77a4f201451d96a9eba583e1a 100644
--- a/src/TNL/Containers/IndexedSet_impl.h
+++ b/src/TNL/Containers/IndexedSet_impl.h
@@ -10,85 +10,83 @@
#pragma once
+#include <TNL/Containers/IndexedSet.h>
+
namespace TNL {
namespace Containers {
-template< typename Element,
- typename Index,
- typename Key >
-void IndexedSet< Element, Index, Key >::reset()
+template< class Key,
+ class Index,
+ class Compare,
+ class Allocator >
+void
+IndexedSet< Key, Index, Compare, Allocator >::clear()
{
map.clear();
}
-template< typename Element,
- typename Index,
- typename Key >
-Index IndexedSet< Element, Index, Key >::getSize() const
+template< class Key,
+ class Index,
+ class Compare,
+ class Allocator >
+typename IndexedSet< Key, Index, Compare, Allocator >::size_type
+IndexedSet< Key, Index, Compare, Allocator >::size() const
{
return map.size();
}
-template< typename Element,
- typename Index,
- typename Key >
-Index IndexedSet< Element, Index, Key >::insert( const Element &data )
+template< class Key,
+ class Index,
+ class Compare,
+ class Allocator >
+Index
+IndexedSet< Key, Index, Compare, Allocator >::insert( const Key& key )
{
- STDMapIteratorType iter = map.insert( STDMapValueType( Key( data ),
- DataWithIndex( data, getSize() ) ) ).first;
- return iter->second.index;
+ auto iter = map.insert( value_type( key, size() ) ).first;
+ return iter->second;
}
-template< typename Element,
- typename Index,
- typename Key >
-bool IndexedSet< Element, Index, Key >::find( const Element &data, Index& index ) const
+template< class Key,
+ class Index,
+ class Compare,
+ class Allocator >
+bool
+IndexedSet< Key, Index, Compare, Allocator >::find( const Key& key, Index& index ) const
{
- STDMapIteratorType iter = map.find( Key( data ) );
- if (iter == map.end())
+ auto iter = map.find( Key( key ) );
+ if( iter == map.end() )
return false;
index = iter->second.index;
return true;
}
-template< typename Element,
- typename Index,
- typename Key >
- template<typename ArrayType>
-void IndexedSet< Element, Index, Key >::toArray( ArrayType& array ) const
-{
- TNL_ASSERT( array.getSize() == getSize(),
- std::cerr << "array.getSize() = " << array.getSize()
- << " getSize() = " << getSize() );
-
- for( STDMapIteratorType iter = map.begin();
- iter != map.end();
- ++iter)
- array[ iter->second.index ] = iter->second.data;
-}
-
-template< typename Element,
- typename Index,
- typename Key >
-const Element& IndexedSet< Element, Index, Key >::getElement( KeyType key ) const
+template< class Key,
+ class Index,
+ class Compare,
+ class Allocator >
+typename IndexedSet< Key, Index, Compare, Allocator >::size_type
+IndexedSet< Key, Index, Compare, Allocator >::count( const Key& key ) const
{
- return map[ key ];
+ return map.count( key );
}
-template< typename Element,
- typename Index,
- typename Key >
-Element& IndexedSet< Element, Index, Key >::getElement( KeyType key )
+template< class Key,
+ class Index,
+ class Compare,
+ class Allocator >
+typename IndexedSet< Key, Index, Compare, Allocator >::size_type
+IndexedSet< Key, Index, Compare, Allocator >::erase( const Key& key )
{
- return map[ key ];
+ return map.erase( key );
}
-template< typename Element,
- typename Index,
- typename Key >
-void IndexedSet< Element, Index, Key >::print( std::ostream& str ) const
+template< class Key,
+ class Index,
+ class Compare,
+ class Allocator >
+void IndexedSet< Key, Index, Compare, Allocator >::print( std::ostream& str ) const
{
- STDMapIteratorType iter = map.begin();
+ auto iter = map.begin();
str << iter->second.data;
iter++;
while( iter != map.end() )
@@ -98,10 +96,11 @@ void IndexedSet< Element, Index, Key >::print( std::ostream& str ) const
}
}
-template< typename Element,
- typename Index,
- typename Key >
-std::ostream& operator<<( std::ostream& str, IndexedSet< Element, Index, Key >& set )
+template< class Key,
+ class Index,
+ class Compare,
+ class Allocator >
+std::ostream& operator<<( std::ostream& str, IndexedSet< Key, Index, Compare, Allocator >& set )
{
set.print( str );
return str;
diff --git a/src/TNL/Containers/List.h b/src/TNL/Containers/List.h
index adb163374dc7a3e9d7bbb626605bbebeb2e0c92d..0ec42106de2b49f5296e8a0aede1328715eea3f9 100644
--- a/src/TNL/Containers/List.h
+++ b/src/TNL/Containers/List.h
@@ -10,7 +10,6 @@
#pragma once
-#include <stdlib.h>
#include <iostream>
#include <TNL/Assert.h>
@@ -41,9 +40,7 @@ template< class T > class ListDataElement;
*/
template< class T > class List
{
-
public:
-
typedef T ElementType;
//! Basic constructor
@@ -118,7 +115,6 @@ template< class T > class List
bool DeepLoad( File& file );
protected:
-
//! Pointer to the first element
ListDataElement< T >* first;
@@ -135,8 +131,6 @@ template< class T > class List
//! Iterator index
mutable int index;
-
-
};
template< typename T > std::ostream& operator << ( std::ostream& str, const List< T >& list );
diff --git a/src/TNL/Containers/List_impl.h b/src/TNL/Containers/List_impl.h
index f91f30ca2b7fd7578a109d0786bfcbc93c91a100..136f4cc986a35063a8be9f52ab4fa4e705b6c273 100644
--- a/src/TNL/Containers/List_impl.h
+++ b/src/TNL/Containers/List_impl.h
@@ -1,5 +1,5 @@
/***************************************************************************
- tnlList_impl.h - description
+ List_impl.h - description
-------------------
begin : Mar, 5 Apr 2016 12:46 PM
copyright : (C) 2016 by Tomas Oberhuber
@@ -131,12 +131,10 @@ bool List< T >::Append( const T& data )
{
TNL_ASSERT( ! last, );
first = last = new ListDataElement< T >( data );
- if( ! first ) return false;
}
else
{
ListDataElement< T >* new_element = new ListDataElement< T >( data, last, 0 );
- if( ! new_element ) return false;
TNL_ASSERT( last, );
last = last -> Next() = new_element;
}
@@ -151,12 +149,10 @@ bool List< T >::Prepend( const T& data )
{
TNL_ASSERT( ! last, );
first = last = new ListDataElement< T >( data );
- if( ! first ) return false;
}
else
{
ListDataElement< T >* new_element = new ListDataElement< T >( data, 0, first );
- if( ! new_element ) return false;
first = first -> Previous() = new_element;
}
size ++;
@@ -175,7 +171,6 @@ bool List< T >::Insert( const T& data, const int& ind )
new ListDataElement< T >( data,
iterator -> Previous(),
iterator );
- if( ! new_el ) return false;
iterator -> Previous() -> Next() = new_el;
iterator -> Previous() = new_el;
iterator = new_el;
@@ -208,9 +203,7 @@ template< typename T >
template< typename Array >
void List< T >::toArray( Array& array )
{
- TNL_ASSERT( this->getSize() <= array.getSize(),
- std::cerr << "this->getSize() = " << this->getSize()
- << " array.getSize() = " << array.getSize() << std::endl; );
+ array.setSize( this->getSize() );
for( int i = 0; i < this->getSize(); i++ )
array[ i ] = ( *this )[ i ];
}
@@ -281,59 +274,25 @@ void List< T >::DeepEraseAll()
template< typename T >
bool List< T >::Save( File& file ) const
{
-#ifdef HAVE_NOT_CXX11
- file.write< const int, Devices::Host >( &size );
- for( int i = 0; i < size; i ++ )
- if( ! file. write< int, Devices::Host, int >( &operator[]( i ), 1 ) )
- return false;
- return true;
-#else
file.write( &size );
for( int i = 0; i < size; i ++ )
if( ! file. write( &operator[]( i ), 1 ) )
return false;
return true;
-
-#endif
}
template< typename T >
bool List< T >::DeepSave( File& file ) const
{
-#ifdef HAVE_NOT_CXX11
- file. write< const int, Devices::Host >( &size );
- for( int i = 0; i < size; i ++ )
- if( ! operator[]( i ). save( file ) ) return false;
- return true;
-#else
file. write( &size );
for( int i = 0; i < size; i ++ )
if( ! operator[]( i ). save( file ) ) return false;
return true;
-#endif
}
template< typename T >
bool List< T >::Load( File& file )
{
-#ifdef HAVE_NOT_CXX11
- reset();
- int _size;
- file. read< int, Devices::Host >( &_size );
- if( _size < 0 )
- {
- std::cerr << "The curve size is negative." << std::endl;
- return false;
- }
- T t;
- for( int i = 0; i < _size; i ++ )
- {
- if( ! file. read< T, Devices::Host >( &t ) )
- return false;
- Append( t );
- }
- return true;
-#else
reset();
int _size;
file. read( &_size, 1 );
@@ -350,29 +309,11 @@ bool List< T >::Load( File& file )
Append( t );
}
return true;
-#endif
};
template< typename T >
bool List< T >::DeepLoad( File& file )
{
-#ifdef HAVE_NOT_CXX11
- reset();
- int _size;
- file. read< int, Devices::Host >( &_size );
- if( _size < 0 )
- {
- std::cerr << "The list size is negative." << std::endl;
- return false;
- }
- for( int i = 0; i < _size; i ++ )
- {
- T t;
- if( ! t. load( file ) ) return false;
- Append( t );
- }
- return true;
-#else
reset();
int _size;
file. read( &_size );
@@ -388,7 +329,6 @@ bool List< T >::DeepLoad( File& file )
Append( t );
}
return true;
-#endif
};
template< typename T >
diff --git a/src/TNL/Containers/MultiArray.h b/src/TNL/Containers/MultiArray.h
index 1aabf733bd3bae912818eb1768c39a2b28402c76..f9de98a81f1141f244f9492a3d3cfeb3e85bbabe 100644
--- a/src/TNL/Containers/MultiArray.h
+++ b/src/TNL/Containers/MultiArray.h
@@ -45,9 +45,9 @@ class MultiArray< 1, Element, Device, Index > : public Array< Element, Device, I
virtual String getSerializationTypeVirtual() const;
- bool setDimensions( const Index iSize );
+ void setDimensions( const Index iSize );
- bool setDimensions( const Containers::StaticVector< 1, Index >& dimensions );
+ void setDimensions( const Containers::StaticVector< 1, Index >& dimensions );
__cuda_callable__ void getDimensions( Index& iSize ) const;
@@ -55,7 +55,7 @@ class MultiArray< 1, Element, Device, Index > : public Array< Element, Device, I
//! Set dimensions of the array using another array as a template
template< typename MultiArray >
- bool setLike( const MultiArray& v );
+ void setLike( const MultiArray& v );
void reset();
@@ -124,9 +124,9 @@ class MultiArray< 2, Element, Device, Index > : public Array< Element, Device, I
virtual String getSerializationTypeVirtual() const;
- bool setDimensions( const Index jSize, const Index iSize );
+ void setDimensions( const Index jSize, const Index iSize );
- bool setDimensions( const Containers::StaticVector< 2, Index >& dimensions );
+ void setDimensions( const Containers::StaticVector< 2, Index >& dimensions );
__cuda_callable__ void getDimensions( Index& jSize, Index& iSize ) const;
@@ -134,7 +134,7 @@ class MultiArray< 2, Element, Device, Index > : public Array< Element, Device, I
//! Set dimensions of the array using another array as a template
template< typename MultiArray >
- bool setLike( const MultiArray& v );
+ void setLike( const MultiArray& v );
void reset();
@@ -207,9 +207,9 @@ class MultiArray< 3, Element, Device, Index > : public Array< Element, Device, I
virtual String getSerializationTypeVirtual() const;
- bool setDimensions( const Index k, const Index j, const Index iSize );
+ void setDimensions( const Index k, const Index j, const Index iSize );
- bool setDimensions( const Containers::StaticVector< 3, Index >& dimensions );
+ void setDimensions( const Containers::StaticVector< 3, Index >& dimensions );
__cuda_callable__ void getDimensions( Index& k, Index& j, Index& iSize ) const;
@@ -217,7 +217,7 @@ class MultiArray< 3, Element, Device, Index > : public Array< Element, Device, I
//! Set dimensions of the array using another array as a template
template< typename MultiArrayT >
- bool setLike( const MultiArrayT& v );
+ void setLike( const MultiArrayT& v );
void reset();
@@ -290,9 +290,9 @@ class MultiArray< 4, Element, Device, Index > : public Array< Element, Device, I
virtual String getSerializationTypeVirtual() const;
- bool setDimensions( const Index l, const Index k, const Index j, const Index iSize );
+ void setDimensions( const Index l, const Index k, const Index j, const Index iSize );
- bool setDimensions( const Containers::StaticVector< 4, Index >& dimensions );
+ void setDimensions( const Containers::StaticVector< 4, Index >& dimensions );
__cuda_callable__ void getDimensions( Index& l, Index& k, Index& j, Index& iSize ) const;
@@ -300,7 +300,7 @@ class MultiArray< 4, Element, Device, Index > : public Array< Element, Device, I
//! Set dimensions of the array using another array as a template
template< typename MultiArrayT >
- bool setLike( const MultiArrayT& v );
+ void setLike( const MultiArrayT& v );
void reset();
diff --git a/src/TNL/Containers/MultiArray1D_impl.h b/src/TNL/Containers/MultiArray1D_impl.h
index ec64e060053a57cbfbcae0240961df0619fc4d79..d380aa831fdba0b302dee9743de4dbb7574d3d4c 100644
--- a/src/TNL/Containers/MultiArray1D_impl.h
+++ b/src/TNL/Containers/MultiArray1D_impl.h
@@ -57,28 +57,28 @@ String MultiArray< 1, Element, Device, Index > :: getSerializationTypeVirtual()
};
template< typename Element, typename Device, typename Index >
-bool MultiArray< 1, Element, Device, Index > :: setDimensions( const Index iSize )
+void MultiArray< 1, Element, Device, Index > :: setDimensions( const Index iSize )
{
TNL_ASSERT( iSize > 0,
std::cerr << "iSize = " << iSize );
dimensions[ 0 ] = iSize;
- return Array< Element, Device, Index >::setSize( iSize );
+ Array< Element, Device, Index >::setSize( iSize );
}
template< typename Element, typename Device, typename Index >
-bool MultiArray< 1, Element, Device, Index > :: setDimensions( const Containers::StaticVector< 1, Index >& dimensions )
+void MultiArray< 1, Element, Device, Index > :: setDimensions( const Containers::StaticVector< 1, Index >& dimensions )
{
TNL_ASSERT( dimensions[ 0 ] > 0,
std::cerr << " dimensions[ 0 ] = " << dimensions[ 0 ] );
this->dimensions = dimensions;
- return Array< Element, Device, Index >::setSize( this->dimensions[ 0 ] );
+ Array< Element, Device, Index >::setSize( this->dimensions[ 0 ] );
}
template< typename Element, typename Device, typename Index >
template< typename MultiArrayT >
-bool MultiArray< 1, Element, Device, Index > :: setLike( const MultiArrayT& multiArray )
+void MultiArray< 1, Element, Device, Index > :: setLike( const MultiArrayT& multiArray )
{
- return setDimensions( multiArray. getDimensions() );
+ setDimensions( multiArray. getDimensions() );
}
template< typename Element, typename Device, typename Index >
diff --git a/src/TNL/Containers/MultiArray2D_impl.h b/src/TNL/Containers/MultiArray2D_impl.h
index cbaca350a01a8a84fdac2924c01c137610941597..3812429c8260d75874b74b249876a5d9e8e56cf1 100644
--- a/src/TNL/Containers/MultiArray2D_impl.h
+++ b/src/TNL/Containers/MultiArray2D_impl.h
@@ -57,7 +57,7 @@ String MultiArray< 2, Element, Device, Index > :: getSerializationTypeVirtual()
};
template< typename Element, typename Device, typename Index >
-bool MultiArray< 2, Element, Device, Index > :: setDimensions( const Index jSize,
+void MultiArray< 2, Element, Device, Index > :: setDimensions( const Index jSize,
const Index iSize )
{
TNL_ASSERT( iSize > 0 && jSize > 0,
@@ -66,11 +66,11 @@ bool MultiArray< 2, Element, Device, Index > :: setDimensions( const Index jSize
dimensions[ 0 ] = iSize;
dimensions[ 1 ] = jSize;
- return Array< Element, Device, Index > :: setSize( iSize * jSize );
+ Array< Element, Device, Index > :: setSize( iSize * jSize );
}
template< typename Element, typename Device, typename Index >
-bool MultiArray< 2, Element, Device, Index > :: setDimensions( const Containers::StaticVector< 2, Index >& dimensions )
+void MultiArray< 2, Element, Device, Index > :: setDimensions( const Containers::StaticVector< 2, Index >& dimensions )
{
TNL_ASSERT( dimensions[ 0 ] > 0 && dimensions[ 1 ] > 0,
std::cerr << "dimensions = " << dimensions );
@@ -79,14 +79,14 @@ bool MultiArray< 2, Element, Device, Index > :: setDimensions( const Containers:
*/
this->dimensions. x() = dimensions. y();
this->dimensions. y() = dimensions. x();
- return Array< Element, Device, Index > :: setSize( this->dimensions[ 1 ] * this->dimensions[ 0 ] );
+ Array< Element, Device, Index > :: setSize( this->dimensions[ 1 ] * this->dimensions[ 0 ] );
}
template< typename Element, typename Device, typename Index >
template< typename MultiArrayT >
-bool MultiArray< 2, Element, Device, Index > :: setLike( const MultiArrayT& multiArray )
+void MultiArray< 2, Element, Device, Index > :: setLike( const MultiArrayT& multiArray )
{
- return setDimensions( multiArray. getDimensions() );
+ setDimensions( multiArray. getDimensions() );
}
template< typename Element, typename Device, typename Index >
diff --git a/src/TNL/Containers/MultiArray3D_impl.h b/src/TNL/Containers/MultiArray3D_impl.h
index 750efab1b71834ab058744b8eb081a4f4ed21396..6e9fb9d9f2697d997ab32522257c2b1be23185bd 100644
--- a/src/TNL/Containers/MultiArray3D_impl.h
+++ b/src/TNL/Containers/MultiArray3D_impl.h
@@ -57,7 +57,7 @@ String MultiArray< 3, Element, Device, Index > :: getSerializationTypeVirtual()
};
template< typename Element, typename Device, typename Index >
-bool MultiArray< 3, Element, Device, Index > :: setDimensions( const Index kSize,
+void MultiArray< 3, Element, Device, Index > :: setDimensions( const Index kSize,
const Index jSize,
const Index iSize )
{
@@ -69,11 +69,11 @@ bool MultiArray< 3, Element, Device, Index > :: setDimensions( const Index kSize
dimensions[ 0 ] = iSize;
dimensions[ 1 ] = jSize;
dimensions[ 2 ] = kSize;
- return Array< Element, Device, Index > :: setSize( iSize * jSize * kSize );
+ Array< Element, Device, Index > :: setSize( iSize * jSize * kSize );
}
template< typename Element, typename Device, typename Index >
-bool MultiArray< 3, Element, Device, Index > :: setDimensions( const Containers::StaticVector< 3, Index >& dimensions )
+void MultiArray< 3, Element, Device, Index > :: setDimensions( const Containers::StaticVector< 3, Index >& dimensions )
{
TNL_ASSERT( dimensions[ 0 ] > 0 && dimensions[ 1 ] > 0 && dimensions[ 2 ],
std::cerr << "dimensions = " << dimensions );
@@ -83,16 +83,16 @@ bool MultiArray< 3, Element, Device, Index > :: setDimensions( const Containers:
this->dimensions. x() = dimensions. z();
this->dimensions. y() = dimensions. y();
this->dimensions. z() = dimensions. x();
- return Array< Element, Device, Index > :: setSize( this->dimensions[ 2 ] *
- this->dimensions[ 1 ] *
- this->dimensions[ 0 ] );
+ Array< Element, Device, Index > :: setSize( this->dimensions[ 2 ] *
+ this->dimensions[ 1 ] *
+ this->dimensions[ 0 ] );
}
template< typename Element, typename Device, typename Index >
template< typename MultiArrayT >
-bool MultiArray< 3, Element, Device, Index > :: setLike( const MultiArrayT& multiArray )
+void MultiArray< 3, Element, Device, Index > :: setLike( const MultiArrayT& multiArray )
{
- return setDimensions( multiArray. getDimensions() );
+ setDimensions( multiArray. getDimensions() );
}
template< typename Element, typename Device, typename Index >
diff --git a/src/TNL/Containers/MultiArray4D_impl.h b/src/TNL/Containers/MultiArray4D_impl.h
index 5f022247c086461bb11786fda96e4e90f25d7b42..ec034b3d21b6b8de53fa3326c0f240a5a59a8884 100644
--- a/src/TNL/Containers/MultiArray4D_impl.h
+++ b/src/TNL/Containers/MultiArray4D_impl.h
@@ -58,7 +58,7 @@ String MultiArray< 4, Element, Device, Index > :: getSerializationTypeVirtual()
};
template< typename Element, typename Device, typename Index >
-bool MultiArray< 4, Element, Device, Index > :: setDimensions( const Index lSize,
+void MultiArray< 4, Element, Device, Index > :: setDimensions( const Index lSize,
const Index kSize,
const Index jSize,
const Index iSize )
@@ -73,11 +73,11 @@ bool MultiArray< 4, Element, Device, Index > :: setDimensions( const Index lSize
dimensions[ 1 ] = jSize;
dimensions[ 2 ] = kSize;
dimensions[ 3 ] = lSize;
- return Array< Element, Device, Index > :: setSize( iSize * jSize * kSize * lSize );
+ Array< Element, Device, Index > :: setSize( iSize * jSize * kSize * lSize );
}
template< typename Element, typename Device, typename Index >
-bool MultiArray< 4, Element, Device, Index > :: setDimensions( const Containers::StaticVector< 4, Index >& dimensions )
+void MultiArray< 4, Element, Device, Index > :: setDimensions( const Containers::StaticVector< 4, Index >& dimensions )
{
TNL_ASSERT( dimensions[ 0 ] > 0 && dimensions[ 1 ] > 0 && dimensions[ 2 ] && dimensions[ 3 ] > 0,
std::cerr << "dimensions = " << dimensions );
@@ -88,17 +88,17 @@ bool MultiArray< 4, Element, Device, Index > :: setDimensions( const Containers:
this->dimensions[ 1 ] = dimensions[ 2 ];
this->dimensions[ 2 ] = dimensions[ 1 ];
this->dimensions[ 3 ] = dimensions[ 0 ];
- return Array< Element, Device, Index > :: setSize( this->dimensions[ 3 ] *
- this->dimensions[ 2 ] *
- this->dimensions[ 1 ] *
- this->dimensions[ 0 ] );
+ Array< Element, Device, Index > :: setSize( this->dimensions[ 3 ] *
+ this->dimensions[ 2 ] *
+ this->dimensions[ 1 ] *
+ this->dimensions[ 0 ] );
}
template< typename Element, typename Device, typename Index >
template< typename MultiArrayT >
-bool MultiArray< 4, Element, Device, Index > :: setLike( const MultiArrayT& multiArray )
+void MultiArray< 4, Element, Device, Index > :: setLike( const MultiArrayT& multiArray )
{
- return setDimensions( multiArray. getDimensions() );
+ setDimensions( multiArray. getDimensions() );
}
template< typename Element, typename Device, typename Index >
diff --git a/src/TNL/Containers/MultiVector.h b/src/TNL/Containers/MultiVector.h
index fd89f2c40cbf054ba484d01782ca5d1e1801efd6..344321ce59fc10a6a04672631f5449c92cd3ca9d 100644
--- a/src/TNL/Containers/MultiVector.h
+++ b/src/TNL/Containers/MultiVector.h
@@ -45,9 +45,9 @@ class MultiVector< 1, Real, Device, Index > : public Vector< Real, Device, Index
virtual String getSerializationTypeVirtual() const;
- bool setDimensions( const Index iSize );
+ void setDimensions( const Index iSize );
- bool setDimensions( const StaticVector< Dimension, Index >& dimensions );
+ void setDimensions( const StaticVector< Dimension, Index >& dimensions );
void getDimensions( Index& iSize ) const;
@@ -55,7 +55,7 @@ class MultiVector< 1, Real, Device, Index > : public Vector< Real, Device, Index
//! Set dimensions of the Vector using another Vector as a template
template< typename MultiVector >
- bool setLike( const MultiVector& v );
+ void setLike( const MultiVector& v );
Index getElementIndex( const Index i ) const;
@@ -126,9 +126,9 @@ class MultiVector< 2, Real, Device, Index > : public Vector< Real, Device, Index
virtual String getSerializationTypeVirtual() const;
- bool setDimensions( const Index jSize, const Index iSize );
+ void setDimensions( const Index jSize, const Index iSize );
- bool setDimensions( const StaticVector< 2, Index >& dimensions );
+ void setDimensions( const StaticVector< 2, Index >& dimensions );
void getDimensions( Index& jSize, Index& iSize ) const;
@@ -136,7 +136,7 @@ class MultiVector< 2, Real, Device, Index > : public Vector< Real, Device, Index
//! Set dimensions of the Vector using another Vector as a template
template< typename MultiVector >
- bool setLike( const MultiVector& v );
+ void setLike( const MultiVector& v );
Index getElementIndex( const Index j, const Index i ) const;
@@ -208,9 +208,9 @@ class MultiVector< 3, Real, Device, Index > : public Vector< Real, Device, Index
virtual String getSerializationTypeVirtual() const;
- bool setDimensions( const Index k, const Index j, const Index iSize );
+ void setDimensions( const Index k, const Index j, const Index iSize );
- bool setDimensions( const StaticVector< 3, Index >& dimensions );
+ void setDimensions( const StaticVector< 3, Index >& dimensions );
void getDimensions( Index& k, Index& j, Index& iSize ) const;
@@ -218,7 +218,7 @@ class MultiVector< 3, Real, Device, Index > : public Vector< Real, Device, Index
//! Set dimensions of the Vector using another Vector as a template
template< typename MultiVector >
- bool setLike( const MultiVector& v );
+ void setLike( const MultiVector& v );
Index getElementIndex( const Index k, const Index j, const Index i ) const;
@@ -290,9 +290,9 @@ class MultiVector< 4, Real, Device, Index > : public Vector< Real, Device, Index
virtual String getSerializationTypeVirtual() const;
- bool setDimensions( const Index l, const Index k, const Index j, const Index iSize );
+ void setDimensions( const Index l, const Index k, const Index j, const Index iSize );
- bool setDimensions( const StaticVector< 4, Index >& dimensions );
+ void setDimensions( const StaticVector< 4, Index >& dimensions );
void getDimensions( Index& l, Index& k, Index& j, Index& iSize ) const;
@@ -300,7 +300,7 @@ class MultiVector< 4, Real, Device, Index > : public Vector< Real, Device, Index
//! Set dimensions of the Vector using another Vector as a template
template< typename MultiVector >
- bool setLike( const MultiVector& v );
+ void setLike( const MultiVector& v );
Index getElementIndex( const Index l, const Index k, const Index j, const Index i ) const;
diff --git a/src/TNL/Containers/MultiVector1D_impl.h b/src/TNL/Containers/MultiVector1D_impl.h
index 93ee232f34c6331d716982f4be4e2358fe776050..eb0e3109f636fbe5c45f396dc63a9e3deb8d0561 100644
--- a/src/TNL/Containers/MultiVector1D_impl.h
+++ b/src/TNL/Containers/MultiVector1D_impl.h
@@ -57,28 +57,28 @@ String MultiVector< 1, Real, Device, Index > :: getSerializationTypeVirtual() co
};
template< typename Real, typename Device, typename Index >
-bool MultiVector< 1, Real, Device, Index > :: setDimensions( const Index iSize )
+void MultiVector< 1, Real, Device, Index > :: setDimensions( const Index iSize )
{
TNL_ASSERT( iSize > 0,
std::cerr << "iSize = " << iSize );
dimensions[ 0 ] = iSize;
- return Vector< Real, Device, Index > :: setSize( iSize );
+ Vector< Real, Device, Index > :: setSize( iSize );
}
template< typename Real, typename Device, typename Index >
-bool MultiVector< 1, Real, Device, Index > :: setDimensions( const StaticVector< Dimension, Index >& dimensions )
+void MultiVector< 1, Real, Device, Index > :: setDimensions( const StaticVector< Dimension, Index >& dimensions )
{
TNL_ASSERT( dimensions[ 0 ] > 0,
std::cerr << " dimensions[ 0 ] = " << dimensions[ 0 ] );
this->dimensions = dimensions;
- return Vector< Real, Device, Index > :: setSize( this->dimensions[ 0 ] );
+ Vector< Real, Device, Index > :: setSize( this->dimensions[ 0 ] );
}
template< typename Real, typename Device, typename Index >
template< typename MultiVectorT >
-bool MultiVector< 1, Real, Device, Index > :: setLike( const MultiVectorT& multiVector )
+void MultiVector< 1, Real, Device, Index > :: setLike( const MultiVectorT& multiVector )
{
- return setDimensions( multiVector. getDimensions() );
+ setDimensions( multiVector. getDimensions() );
}
template< typename Real, typename Device, typename Index >
diff --git a/src/TNL/Containers/MultiVector2D_impl.h b/src/TNL/Containers/MultiVector2D_impl.h
index 0abbd6cc04bf19e1b3b5402d7a57d73219a30db7..7eb483e9ed122ecff6a6e4d67ce2d9d25d06fb15 100644
--- a/src/TNL/Containers/MultiVector2D_impl.h
+++ b/src/TNL/Containers/MultiVector2D_impl.h
@@ -57,8 +57,8 @@ String MultiVector< 2, Real, Device, Index > :: getSerializationTypeVirtual() co
};
template< typename Real, typename Device, typename Index >
-bool MultiVector< 2, Real, Device, Index > :: setDimensions( const Index jSize,
- const Index iSize )
+void MultiVector< 2, Real, Device, Index > :: setDimensions( const Index jSize,
+ const Index iSize )
{
TNL_ASSERT( iSize > 0 && jSize > 0,
std::cerr << "iSize = " << iSize
@@ -66,23 +66,23 @@ bool MultiVector< 2, Real, Device, Index > :: setDimensions( const Index jSize,
dimensions[ 0 ] = iSize;
dimensions[ 1 ] = jSize;
- return Vector< Real, Device, Index > :: setSize( iSize * jSize );
+ Vector< Real, Device, Index > :: setSize( iSize * jSize );
}
template< typename Real, typename Device, typename Index >
-bool MultiVector< 2, Real, Device, Index > :: setDimensions( const StaticVector< 2, Index >& dimensions )
+void MultiVector< 2, Real, Device, Index > :: setDimensions( const StaticVector< 2, Index >& dimensions )
{
TNL_ASSERT( dimensions[ 0 ] > 0 && dimensions[ 1 ] > 0,
std::cerr << "dimensions = " << dimensions );
this->dimensions = dimensions;
- return Vector< Real, Device, Index > :: setSize( this->dimensions[ 1 ] * this->dimensions[ 0 ] );
+ Vector< Real, Device, Index > :: setSize( this->dimensions[ 1 ] * this->dimensions[ 0 ] );
}
template< typename Real, typename Device, typename Index >
template< typename MultiVectorT >
-bool MultiVector< 2, Real, Device, Index > :: setLike( const MultiVectorT& multiVector )
+void MultiVector< 2, Real, Device, Index > :: setLike( const MultiVectorT& multiVector )
{
- return setDimensions( multiVector. getDimensions() );
+ setDimensions( multiVector. getDimensions() );
}
template< typename Real, typename Device, typename Index >
diff --git a/src/TNL/Containers/MultiVector3D_impl.h b/src/TNL/Containers/MultiVector3D_impl.h
index 0ee9304075c5c03e3a2f974535dc317904db2922..24d92f0e262a32e77b8c375f07f13bb40742d5e5 100644
--- a/src/TNL/Containers/MultiVector3D_impl.h
+++ b/src/TNL/Containers/MultiVector3D_impl.h
@@ -57,9 +57,9 @@ String MultiVector< 3, Real, Device, Index > :: getSerializationTypeVirtual() co
};
template< typename Real, typename Device, typename Index >
-bool MultiVector< 3, Real, Device, Index > :: setDimensions( const Index kSize,
- const Index jSize,
- const Index iSize )
+void MultiVector< 3, Real, Device, Index > :: setDimensions( const Index kSize,
+ const Index jSize,
+ const Index iSize )
{
TNL_ASSERT( iSize > 0 && jSize > 0 && kSize > 0,
std::cerr << "iSize = " << iSize
@@ -73,21 +73,21 @@ bool MultiVector< 3, Real, Device, Index > :: setDimensions( const Index kSize,
}
template< typename Real, typename Device, typename Index >
-bool MultiVector< 3, Real, Device, Index > :: setDimensions( const StaticVector< 3, Index >& dimensions )
+void MultiVector< 3, Real, Device, Index > :: setDimensions( const StaticVector< 3, Index >& dimensions )
{
TNL_ASSERT( dimensions[ 0 ] > 0 && dimensions[ 1 ] > 0 && dimensions[ 2 ],
std::cerr << "dimensions = " << dimensions );
this->dimensions = dimensions;
- return Vector< Real, Device, Index > :: setSize( this->dimensions[ 2 ] *
- this->dimensions[ 1 ] *
- this->dimensions[ 0 ] );
+ Vector< Real, Device, Index > :: setSize( this->dimensions[ 2 ] *
+ this->dimensions[ 1 ] *
+ this->dimensions[ 0 ] );
}
template< typename Real, typename Device, typename Index >
template< typename MultiVectorT >
-bool MultiVector< 3, Real, Device, Index > :: setLike( const MultiVectorT& multiVector )
+void MultiVector< 3, Real, Device, Index > :: setLike( const MultiVectorT& multiVector )
{
- return setDimensions( multiVector. getDimensions() );
+ setDimensions( multiVector. getDimensions() );
}
template< typename Real, typename Device, typename Index >
diff --git a/src/TNL/Containers/MultiVector4D_impl.h b/src/TNL/Containers/MultiVector4D_impl.h
index 3db8a363fc55efcdf618a68b3534393aceac0a79..f9c883371f53439e9921280a844cb6c21175ff19 100644
--- a/src/TNL/Containers/MultiVector4D_impl.h
+++ b/src/TNL/Containers/MultiVector4D_impl.h
@@ -57,10 +57,10 @@ String MultiVector< 4, Real, Device, Index > :: getSerializationTypeVirtual() co
};
template< typename Real, typename Device, typename Index >
-bool MultiVector< 4, Real, Device, Index > :: setDimensions( const Index lSize,
- const Index kSize,
- const Index jSize,
- const Index iSize )
+void MultiVector< 4, Real, Device, Index > :: setDimensions( const Index lSize,
+ const Index kSize,
+ const Index jSize,
+ const Index iSize )
{
TNL_ASSERT( iSize > 0 && jSize > 0 && kSize > 0 && lSize > 0,
std::cerr << "iSize = " << iSize
@@ -72,26 +72,26 @@ bool MultiVector< 4, Real, Device, Index > :: setDimensions( const Index lSize,
dimensions[ 1 ] = jSize;
dimensions[ 2 ] = kSize;
dimensions[ 3 ] = lSize;
- return Vector< Real, Device, Index > :: setSize( iSize * jSize * kSize * lSize );
+ Vector< Real, Device, Index > :: setSize( iSize * jSize * kSize * lSize );
}
template< typename Real, typename Device, typename Index >
-bool MultiVector< 4, Real, Device, Index > :: setDimensions( const StaticVector< 4, Index >& dimensions )
+void MultiVector< 4, Real, Device, Index > :: setDimensions( const StaticVector< 4, Index >& dimensions )
{
TNL_ASSERT( dimensions[ 0 ] > 0 && dimensions[ 1 ] > 0 && dimensions[ 2 ] && dimensions[ 3 ] > 0,
std::cerr << "dimensions = " << dimensions );
this->dimensions = dimensions;
- return Vector< Real, Device, Index > :: setSize( this->dimensions[ 3 ] *
- this->dimensions[ 2 ] *
- this->dimensions[ 1 ] *
- this->dimensions[ 0 ] );
+ Vector< Real, Device, Index > :: setSize( this->dimensions[ 3 ] *
+ this->dimensions[ 2 ] *
+ this->dimensions[ 1 ] *
+ this->dimensions[ 0 ] );
}
template< typename Real, typename Device, typename Index >
template< typename MultiVectorT >
-bool MultiVector< 4, Real, Device, Index > :: setLike( const MultiVectorT& multiVector )
+void MultiVector< 4, Real, Device, Index > :: setLike( const MultiVectorT& multiVector )
{
- return setDimensions( multiVector. getDimensions() );
+ setDimensions( multiVector. getDimensions() );
}
template< typename Real, typename Device, typename Index >
diff --git a/src/TNL/Containers/SharedArray.h b/src/TNL/Containers/SharedArray.h
index 0c4684830b9addd29e55a8d37fcbf34ceda9aafe..4feaf7862bd7fc35a17e35e69a6230466d0cdf49 100644
--- a/src/TNL/Containers/SharedArray.h
+++ b/src/TNL/Containers/SharedArray.h
@@ -44,17 +44,25 @@ class SharedArray : public Object
typedef SharedArray< Element, Devices::Host, Index > HostType;
typedef SharedArray< Element, Devices::Cuda, Index > CudaType;
+ #ifndef HAVE_MIC
__cuda_callable__
+ #endif
SharedArray();
+ #ifndef HAVE_MIC
__cuda_callable__
+ #endif
SharedArray( Element* _data,
const Index _size );
+ #ifndef HAVE_MIC
__cuda_callable__
+ #endif
SharedArray( Array< Element, Device, Index >& array );
+ #ifndef HAVE_MIC
__cuda_callable__
+ #endif
SharedArray( SharedArray< Element, Device, Index >& array );
static String getType();
@@ -113,6 +121,7 @@ class SharedArray : public Object
__cuda_callable__ Element* getData();
+
/*!
* Returns true if non-zero size is set.
*/
@@ -123,13 +132,8 @@ class SharedArray : public Object
* Every time one touches this grid touches * size * sizeof( Real ) bytes are added
* to transfered bytes in tnlStatistics.
*/
-#ifdef HAVE_NOT_CXX11
- template< typename IndexType2 >
- void touch( IndexType2 touches = 1 ) const;
-#else
template< typename IndexType2 = Index >
void touch( IndexType2 touches = 1 ) const;
-#endif
//! Method for saving the object to a file as a binary data.
bool save( File& file ) const;
@@ -149,6 +153,7 @@ class SharedArray : public Object
Element* data;
};
+
template< typename Element, typename Device, typename Index >
std::ostream& operator << ( std::ostream& str, const SharedArray< Element, Device, Index >& v );
diff --git a/src/TNL/Containers/SharedArray_impl.h b/src/TNL/Containers/SharedArray_impl.h
index 558552a803baee2f454f617538d8a8b2d892d08f..44eb48a02177e01f4ed553fab33f5bfa6dd40100 100644
--- a/src/TNL/Containers/SharedArray_impl.h
+++ b/src/TNL/Containers/SharedArray_impl.h
@@ -21,10 +21,13 @@
namespace TNL {
namespace Containers {
+
template< typename Element,
typename Device,
typename Index >
+#ifndef HAVE_MIC
__cuda_callable__
+#endif
SharedArray< Element, Device, Index >::SharedArray()
: size( 0 ), data( 0 )
{
@@ -33,7 +36,9 @@ SharedArray< Element, Device, Index >::SharedArray()
template< typename Element,
typename Device,
typename Index >
+#ifndef HAVE_MIC
__cuda_callable__
+#endif
SharedArray< Element, Device, Index >::SharedArray( Element* _data,
const Index _size )
{
@@ -43,7 +48,9 @@ SharedArray< Element, Device, Index >::SharedArray( Element* _data,
template< typename Element,
typename Device,
typename Index >
+#ifndef HAVE_MIC
__cuda_callable__
+#endif
SharedArray< Element, Device, Index >::SharedArray( Array< Element, Device, Index >& array )
{
this->bind( array );
@@ -52,7 +59,9 @@ SharedArray< Element, Device, Index >::SharedArray( Array< Element, Device, Inde
template< typename Element,
typename Device,
typename Index >
+#ifndef HAVE_MIC
__cuda_callable__
+#endif
SharedArray< Element, Device, Index >::SharedArray( SharedArray< Element, Device, Index >& array )
{
this->bind( array );
@@ -352,11 +361,7 @@ bool SharedArray< Element, Device, Index > :: save( File& file ) const
std::cerr << "You try to save empty array." << std::endl );
if( ! Object :: save( file ) )
return false;
-#ifdef HAVE_NOT_CXX11
- if( ! file. write< const Index, Devices::Host >( &this->size ) )
-#else
if( ! file. write( &this->size ) )
-#endif
return false;
if( ! file. write< Element, Device, Index >( this->data, this->size ) )
{
@@ -382,13 +387,8 @@ bool SharedArray< Element, Device, Index > :: load( File& file )
if( ! Object :: load( file ) )
return false;
Index _size;
-#ifdef HAVE_NOT_CXX11
- if( ! file. read< Index, Devices::Host >( &_size ) )
- return false;
-#else
if( ! file. read( &_size, 1 ) )
return false;
-#endif
if( _size != this->size )
{
std::cerr << "Error: The size " << _size << " of the data to be load is different from the " <<
diff --git a/src/TNL/Containers/SharedVector.h b/src/TNL/Containers/SharedVector.h
index c4a22ded20ba860502c408c3ddd038966146064c..6526f1b4b7aac8241325f70a866a803b20e98158 100644
--- a/src/TNL/Containers/SharedVector.h
+++ b/src/TNL/Containers/SharedVector.h
@@ -39,17 +39,25 @@ class SharedVector : public Containers::SharedArray< Real, Device, Index >
typedef SharedVector< Real, Devices::Cuda, Index > CudaType;
+ #ifndef HAVE_MIC
__cuda_callable__
+ #endif
SharedVector();
+ #ifndef HAVE_MIC
__cuda_callable__
+ #endif
SharedVector( Real* data,
const Index size );
+ #ifndef HAVE_MIC
__cuda_callable__
+ #endif
SharedVector( Vector< Real, Device, Index >& vector );
+ #ifndef HAVE_MIC
__cuda_callable__
+ #endif
SharedVector( SharedVector< Real, Device, Index >& vector );
static String getType();
diff --git a/src/TNL/Containers/SharedVector_impl.h b/src/TNL/Containers/SharedVector_impl.h
index 869f0aeae455b9b3891d9147f8d3ea4285503a02..446612f133408a6560b2e89dcaf60e4d35259ed6 100644
--- a/src/TNL/Containers/SharedVector_impl.h
+++ b/src/TNL/Containers/SharedVector_impl.h
@@ -19,7 +19,9 @@ namespace Containers {
template< typename Real,
typename Device,
typename Index >
+#ifndef HAVE_MIC
__cuda_callable__
+#endif
SharedVector< Real, Device, Index >::SharedVector()
{
}
@@ -27,7 +29,9 @@ SharedVector< Real, Device, Index >::SharedVector()
template< typename Real,
typename Device,
typename Index >
+#ifndef HAVE_MIC
__cuda_callable__
+#endif
SharedVector< Real, Device, Index >::SharedVector( Real* data,
const Index size )
: Containers::SharedArray< Real, Device, Index >( data, size )
@@ -37,7 +41,9 @@ SharedVector< Real, Device, Index >::SharedVector( Real* data,
template< typename Real,
typename Device,
typename Index >
+#ifndef HAVE_MIC
__cuda_callable__
+#endif
SharedVector< Real, Device, Index >::SharedVector( Vector< Real, Device, Index >& vector )
: Containers::SharedArray< Real, Device, Index >( vector )
{
@@ -46,7 +52,9 @@ SharedVector< Real, Device, Index >::SharedVector( Vector< Real, Device, Index >
template< typename Real,
typename Device,
typename Index >
+#ifndef HAVE_MIC
__cuda_callable__
+#endif
SharedVector< Real, Device, Index >::SharedVector( SharedVector< Real, Device, Index >& vector )
: Containers::SharedArray< Real, Device, Index >( vector )
{
diff --git a/src/TNL/Containers/StaticArray.h b/src/TNL/Containers/StaticArray.h
index 63ac6be40399a2ccec1948fc54bfaf5a32181ade..32f4528dbf5aaedf1c9561c849e42747bd5f1e4a 100644
--- a/src/TNL/Containers/StaticArray.h
+++ b/src/TNL/Containers/StaticArray.h
@@ -16,10 +16,6 @@
namespace TNL {
namespace Containers {
-//! Aliases for the coordinates
-// TODO: Remove this - it is here only because of some legact code
-enum { tnlX = 0, tnlY, tnlZ };
-
template< int Size, typename Element >
class StaticArray
{
@@ -31,6 +27,9 @@ class StaticArray
__cuda_callable__
inline StaticArray();
+ // Note: the template avoids ambiguity of overloaded functions with literal 0 and pointer
+ // reference: https://stackoverflow.com/q/4610503
+ template< typename _unused = void >
__cuda_callable__
inline StaticArray( const Element v[ Size ] );
@@ -91,7 +90,6 @@ class StaticArray
protected:
Element data[ Size ];
-
};
template< typename Element >
@@ -105,6 +103,9 @@ class StaticArray< 1, Element >
__cuda_callable__
inline StaticArray();
+ // Note: the template avoids ambiguity of overloaded functions with literal 0 and pointer
+ // reference: https://stackoverflow.com/q/4610503
+ template< typename _unused = void >
__cuda_callable__
inline StaticArray( const Element v[ size ] );
@@ -186,6 +187,9 @@ class StaticArray< 2, Element >
__cuda_callable__
inline StaticArray();
+ // Note: the template avoids ambiguity of overloaded functions with literal 0 and pointer
+ // reference: https://stackoverflow.com/q/4610503
+ template< typename _unused = void >
__cuda_callable__
inline StaticArray( const Element v[ size ] );
@@ -278,6 +282,9 @@ class StaticArray< 3, Element >
__cuda_callable__
inline StaticArray();
+ // Note: the template avoids ambiguity of overloaded functions with literal 0 and pointer
+ // reference: https://stackoverflow.com/q/4610503
+ template< typename _unused = void >
__cuda_callable__
inline StaticArray( const Element v[ size ] );
@@ -365,7 +372,6 @@ class StaticArray< 3, Element >
protected:
Element data[ size ];
-
};
template< int Size, typename Element >
@@ -378,4 +384,3 @@ std::ostream& operator << ( std::ostream& str, const StaticArray< Size, Element
#include <TNL/Containers/StaticArray1D_impl.h>
#include <TNL/Containers/StaticArray2D_impl.h>
#include <TNL/Containers/StaticArray3D_impl.h>
-
diff --git a/src/TNL/Containers/StaticArray1D_impl.h b/src/TNL/Containers/StaticArray1D_impl.h
index 0b368dbf58c188611fa32173cfb49b6210f98f35..a23849cca338b609ad65a3c68d99d0c712ccf099 100644
--- a/src/TNL/Containers/StaticArray1D_impl.h
+++ b/src/TNL/Containers/StaticArray1D_impl.h
@@ -11,6 +11,7 @@
#pragma once
#include <TNL/param-types.h>
+#include <TNL/Containers/StaticArray.h>
namespace TNL {
namespace Containers {
@@ -22,6 +23,7 @@ inline StaticArray< 1, Element >::StaticArray()
}
template< typename Element >
+ template< typename _unused >
__cuda_callable__
inline StaticArray< 1, Element >::StaticArray( const Element v[ size ] )
{
@@ -77,8 +79,8 @@ template< typename Element >
__cuda_callable__
inline const Element& StaticArray< 1, Element >::operator[]( int i ) const
{
- TNL_ASSERT( i >= 0 && i < size,
- std::cerr << "i = " << i << " size = " << size << std::endl; );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, size, "Element index is out of bounds." );
return data[ i ];
}
@@ -86,8 +88,8 @@ template< typename Element >
__cuda_callable__
inline Element& StaticArray< 1, Element >::operator[]( int i )
{
- TNL_ASSERT( i >= 0 && i < size,
- std::cerr << "i = " << i << " size = " << size << std::endl; );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, size, "Element index is out of bounds." );
return data[ i ];
}
@@ -171,11 +173,7 @@ bool StaticArray< 1, Element >::save( File& file ) const
template< typename Element >
bool StaticArray< 1, Element >::load( File& file)
{
-#ifdef HAVE_NOT_CXX11
- if( ! file.read< Element, Devices::Host, int >( data, size ) )
-#else
if( ! file.read( data, size ) )
-#endif
{
std::cerr << "Unable to read " << getType() << "." << std::endl;
return false;
diff --git a/src/TNL/Containers/StaticArray2D_impl.h b/src/TNL/Containers/StaticArray2D_impl.h
index a0301a4bbf13f19abe1f78577de2c45cc2ece4c7..44ed90b0a6c42f5d9ef28cc2a72959a8c17e586e 100644
--- a/src/TNL/Containers/StaticArray2D_impl.h
+++ b/src/TNL/Containers/StaticArray2D_impl.h
@@ -12,6 +12,7 @@
#include <TNL/param-types.h>
#include <TNL/Math.h>
+#include <TNL/Containers/StaticArray.h>
namespace TNL {
namespace Containers {
@@ -23,6 +24,7 @@ inline StaticArray< 2, Element >::StaticArray()
}
template< typename Element >
+ template< typename _unused >
__cuda_callable__
inline StaticArray< 2, Element >::StaticArray( const Element v[ size ] )
{
@@ -89,8 +91,8 @@ template< typename Element >
__cuda_callable__
inline const Element& StaticArray< 2, Element >::operator[]( int i ) const
{
- TNL_ASSERT( i >= 0 && i < size,
- std::cerr << "i = " << i << " size = " << size << std::endl; );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, size, "Element index is out of bounds." );
return data[ i ];
}
@@ -98,8 +100,8 @@ template< typename Element >
__cuda_callable__
inline Element& StaticArray< 2, Element >::operator[]( int i )
{
- TNL_ASSERT( i >= 0 && i < size,
- std::cerr << "i = " << i << " size = " << size << std::endl; );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, size, "Element index is out of bounds." );
return data[ i ];
}
diff --git a/src/TNL/Containers/StaticArray3D_impl.h b/src/TNL/Containers/StaticArray3D_impl.h
index a246b2e3e6e2e1a26bca5c7c922a0b0371fb3dc1..4e89783ca84dd6279930cd92f21526a34aa54cb3 100644
--- a/src/TNL/Containers/StaticArray3D_impl.h
+++ b/src/TNL/Containers/StaticArray3D_impl.h
@@ -11,6 +11,8 @@
#pragma once
#include <TNL/param-types.h>
+#include <TNL/Math.h>
+#include <TNL/Containers/StaticArray.h>
namespace TNL {
namespace Containers {
@@ -22,6 +24,7 @@ inline StaticArray< 3, Element >::StaticArray()
}
template< typename Element >
+ template< typename _unused >
__cuda_callable__
inline StaticArray< 3, Element >::StaticArray( const Element v[ size ] )
{
@@ -92,8 +95,8 @@ template< typename Element >
__cuda_callable__
inline const Element& StaticArray< 3, Element >::operator[]( int i ) const
{
- TNL_ASSERT( i >= 0 && i < size,
- std::cerr << "i = " << i << " size = " << size << std::endl; );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, size, "Element index is out of bounds." );
return data[ i ];
}
@@ -101,8 +104,8 @@ template< typename Element >
__cuda_callable__
inline Element& StaticArray< 3, Element >::operator[]( int i )
{
- TNL_ASSERT( i >= 0 && i < size,
- std::cerr << "i = " << i << " size = " << size << std::endl; );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, size, "Element index is out of bounds." );
return data[ i ];
}
diff --git a/src/TNL/Containers/StaticArray_impl.h b/src/TNL/Containers/StaticArray_impl.h
index d651749e0f8f46012f7b968e0f8c5e6f0d1fc620..9be36764a5b8e316900db466fa1a284fcf0ac4ab 100644
--- a/src/TNL/Containers/StaticArray_impl.h
+++ b/src/TNL/Containers/StaticArray_impl.h
@@ -12,6 +12,7 @@
#include <TNL/param-types.h>
#include <TNL/Math.h>
+#include <TNL/Containers/StaticArray.h>
namespace TNL {
namespace Containers {
@@ -23,6 +24,7 @@ inline StaticArray< Size, Element >::StaticArray()
};
template< int Size, typename Element >
+ template< typename _unused >
__cuda_callable__
inline StaticArray< Size, Element >::StaticArray( const Element v[ Size ] )
{
@@ -81,8 +83,8 @@ template< int Size, typename Element >
__cuda_callable__
inline const Element& StaticArray< Size, Element >::operator[]( int i ) const
{
- TNL_ASSERT( i >= 0 && i < size,
- std::cerr << "i = " << i << " size = " << size << std::endl; );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, size, "Element index is out of bounds." );
return data[ i ];
}
@@ -90,8 +92,8 @@ template< int Size, typename Element >
__cuda_callable__
inline Element& StaticArray< Size, Element >::operator[]( int i )
{
- TNL_ASSERT( i >= 0 && i < size,
- std::cerr << "i = " << i << " size = " << size << std::endl; );
+ TNL_ASSERT_GE( i, 0, "Element index must be non-negative." );
+ TNL_ASSERT_LT( i, size, "Element index is out of bounds." );
return data[ i ];
}
@@ -203,12 +205,9 @@ std::ostream& StaticArray< Size, Element >::write( std::ostream& str, const char
template< int Size, typename Element >
std::ostream& operator << ( std::ostream& str, const StaticArray< Size, Element >& a )
{
- a.write( str, "," );
- /*for( int i = 0; i < Size - 1; i ++ )
- {
- str << a[ i ] << ", ";
- }
- str << a[ Size - 1 ];*/
+ str << "[ ";
+ a.write( str, ", " );
+ str << " ]";
return str;
};
diff --git a/src/TNL/Containers/StaticVector.h b/src/TNL/Containers/StaticVector.h
index a75f34d3b88f0390caec5ea0e10bd60c810bb8f0..095c9c1ffefd2c9364562f6e9a0d65ee15c33387 100644
--- a/src/TNL/Containers/StaticVector.h
+++ b/src/TNL/Containers/StaticVector.h
@@ -11,23 +11,27 @@
#pragma once
#include <TNL/Containers/StaticArray.h>
+#include <TNL/Config/ParameterContainer.h>
namespace TNL {
namespace Containers {
template< int Size, typename Real = double >
-class StaticVector : public Containers::StaticArray< Size, Real >
+class StaticVector : public StaticArray< Size, Real >
{
public:
typedef Real RealType;
typedef StaticVector< Size, Real > ThisType;
enum { size = Size };
- using Containers::StaticArray< Size, Real >::operator=;
-
+ using StaticArray< Size, Real >::operator=;
+
__cuda_callable__
StaticVector();
+ // Note: the template avoids ambiguity of overloaded functions with literal 0 and pointer
+ // reference: https://stackoverflow.com/q/4610503
+ template< typename _unused = void >
__cuda_callable__
StaticVector( const Real v[ Size ] );
@@ -38,7 +42,7 @@ class StaticVector : public Containers::StaticArray< Size, Real >
//! Copy constructor
__cuda_callable__
StaticVector( const StaticVector< Size, Real >& v );
-
+
bool setup( const Config::ParameterContainer& parameters,
const String& prefix = "" );
@@ -87,25 +91,33 @@ class StaticVector : public Containers::StaticArray< Size, Real >
template< typename OtherReal >
__cuda_callable__
operator StaticVector< Size, OtherReal >() const;
-
+
__cuda_callable__
ThisType abs() const;
-
+
__cuda_callable__
Real lpNorm( const Real& p ) const;
};
template< typename Real >
-class StaticVector< 1, Real > : public Containers::StaticArray< 1, Real >
+class StaticVector< 1, Real > : public StaticArray< 1, Real >
{
public:
typedef Real RealType;
typedef StaticVector< 1, Real > ThisType;
enum { size = 1 };
-
+
+ using StaticArray< 1, Real >::operator=;
+
__cuda_callable__
StaticVector();
+ // Note: the template avoids ambiguity of overloaded functions with literal 0 and pointer
+ // reference: https://stackoverflow.com/q/4610503
+ template< typename _unused = void >
+ __cuda_callable__
+ StaticVector( const Real v[ 1 ] );
+
//! This sets all vector components to v
__cuda_callable__
StaticVector( const Real& v );
@@ -165,22 +177,27 @@ class StaticVector< 1, Real > : public Containers::StaticArray< 1, Real >
__cuda_callable__
ThisType abs() const;
-
+
__cuda_callable__
Real lpNorm( const Real& p ) const;
};
template< typename Real >
-class StaticVector< 2, Real > : public Containers::StaticArray< 2, Real >
+class StaticVector< 2, Real > : public StaticArray< 2, Real >
{
public:
typedef Real RealType;
typedef StaticVector< 2, Real > ThisType;
enum { size = 2 };
-
+
+ using StaticArray< 2, Real >::operator=;
+
__cuda_callable__
StaticVector();
+ // Note: the template avoids ambiguity of overloaded functions with literal 0 and pointer
+ // reference: https://stackoverflow.com/q/4610503
+ template< typename _unused = void >
__cuda_callable__
StaticVector( const Real v[ 2 ] );
@@ -246,22 +263,27 @@ class StaticVector< 2, Real > : public Containers::StaticArray< 2, Real >
__cuda_callable__
ThisType abs() const;
-
+
__cuda_callable__
Real lpNorm( const Real& p ) const;
};
template< typename Real >
-class StaticVector< 3, Real > : public Containers::StaticArray< 3, Real >
+class StaticVector< 3, Real > : public StaticArray< 3, Real >
{
public:
typedef Real RealType;
typedef StaticVector< 3, Real > ThisType;
enum { size = 3 };
-
+
+ using StaticArray< 3, Real >::operator=;
+
__cuda_callable__
StaticVector();
+ // Note: the template avoids ambiguity of overloaded functions with literal 0 and pointer
+ // reference: https://stackoverflow.com/q/4610503
+ template< typename _unused = void >
__cuda_callable__
StaticVector( const Real v[ 3 ] );
@@ -327,7 +349,7 @@ class StaticVector< 3, Real > : public Containers::StaticArray< 3, Real >
__cuda_callable__
ThisType abs() const;
-
+
__cuda_callable__
Real lpNorm( const Real& p ) const;
};
@@ -362,21 +384,21 @@ StaticVector< 3, Real > VectorProduct( const StaticVector< 3, Real >& u,
p[ 1 ] = u[ 2 ] * v[ 0 ] - u[ 0 ] * v[ 2 ];
p[ 2 ] = u[ 0 ] * v[ 1 ] - u[ 1 ] * v[ 0 ];
return p;
-};
+}
template< typename Real >
Real tnlScalarProduct( const StaticVector< 2, Real >& u,
const StaticVector< 2, Real >& v )
{
return u[ 0 ] * v[ 0 ] + u[ 1 ] * v[ 1 ];
-};
+}
template< typename Real >
Real tnlScalarProduct( const StaticVector< 3, Real >& u,
const StaticVector< 3, Real >& v )
{
return u[ 0 ] * v[ 0 ] + u[ 1 ] * v[ 1 ] + u[ 2 ] * v[ 2 ];
-};
+}
template< typename Real >
Real tnlTriangleArea( const StaticVector< 2, Real >& a,
@@ -392,8 +414,8 @@ Real tnlTriangleArea( const StaticVector< 2, Real >& a,
u2. z() = 0;
const StaticVector< 3, Real > v = VectorProduct( u1, u2 );
- return 0.5 * ::sqrt( tnlScalarProduct( v, v ) );
-};
+ return 0.5 * TNL::sqrt( tnlScalarProduct( v, v ) );
+}
template< typename Real >
Real tnlTriangleArea( const StaticVector< 3, Real >& a,
@@ -409,8 +431,8 @@ Real tnlTriangleArea( const StaticVector< 3, Real >& a,
u2. z() = c. z() - a. z();
const StaticVector< 3, Real > v = VectorProduct( u1, u2 );
- return 0.5 * ::sqrt( tnlScalarProduct( v, v ) );
-};
+ return 0.5 * TNL::sqrt( tnlScalarProduct( v, v ) );
+}
} // namespace Containers
} // namespace TNL
diff --git a/src/TNL/Containers/StaticVector1D_impl.h b/src/TNL/Containers/StaticVector1D_impl.h
index ab2391eddacb43a5667d7c82d9c120a940bb1e6e..3127a621910111ff1490ccdc94675c8e737d7fc8 100644
--- a/src/TNL/Containers/StaticVector1D_impl.h
+++ b/src/TNL/Containers/StaticVector1D_impl.h
@@ -10,6 +10,8 @@
#pragma once
+#include <TNL/Containers/StaticVector.h>
+
namespace TNL {
namespace Containers {
@@ -19,17 +21,25 @@ StaticVector< 1, Real >::StaticVector()
{
}
+template< typename Real >
+ template< typename _unused >
+__cuda_callable__
+StaticVector< 1, Real >::StaticVector( const Real v[ 1 ] )
+: StaticArray< 1, Real >( v )
+{
+}
+
template< typename Real >
__cuda_callable__
StaticVector< 1, Real >::StaticVector( const Real& v )
-: Containers::StaticArray< 1, Real >( v )
+: StaticArray< 1, Real >( v )
{
}
template< typename Real >
__cuda_callable__
StaticVector< 1, Real >::StaticVector( const StaticVector< 1, Real >& v )
-: Containers::StaticArray< 1, Real >( v )
+: StaticArray< 1, Real >( v )
{
}
diff --git a/src/TNL/Containers/StaticVector2D_impl.h b/src/TNL/Containers/StaticVector2D_impl.h
index d5f63f283129ae11e50fcdd4352cff91eda26ea3..d66979bf7a3be3bed89929b3d84cdf5da61199d4 100644
--- a/src/TNL/Containers/StaticVector2D_impl.h
+++ b/src/TNL/Containers/StaticVector2D_impl.h
@@ -10,7 +10,7 @@
#pragma once
-#include <TNL/Math.h>
+#include <TNL/Containers/StaticVector.h>
namespace TNL {
namespace Containers {
@@ -22,30 +22,31 @@ StaticVector< 2, Real >::StaticVector()
}
template< typename Real >
+ template< typename _unused >
__cuda_callable__
StaticVector< 2, Real >::StaticVector( const Real v[ 2 ] )
-: Containers::StaticArray< 2, Real >( v )
+: StaticArray< 2, Real >( v )
{
}
template< typename Real >
__cuda_callable__
StaticVector< 2, Real >::StaticVector( const Real& v )
-: Containers::StaticArray< 2, Real >( v )
+: StaticArray< 2, Real >( v )
{
}
template< typename Real >
__cuda_callable__
StaticVector< 2, Real >::StaticVector( const Real& v1, const Real& v2 )
-: Containers::StaticArray< 2, Real >( v1, v2 )
+: StaticArray< 2, Real >( v1, v2 )
{
}
template< typename Real >
__cuda_callable__
StaticVector< 2, Real >::StaticVector( const StaticVector< 2, Real >& v )
-: Containers::StaticArray< 2, Real >( v )
+: StaticArray< 2, Real >( v )
{
}
@@ -183,8 +184,8 @@ __cuda_callable__
StaticVector< 2, Real >
StaticVector< 2, Real >::abs() const
{
- return StaticVector< 2, Real >( ::abs( this->data[ 0 ] ),
- ::abs( this->data[ 1 ] ) );
+ return StaticVector< 2, Real >( TNL::abs( this->data[ 0 ] ),
+ TNL::abs( this->data[ 1 ] ) );
}
template< typename Real >
@@ -195,10 +196,10 @@ StaticVector< 2, Real >::lpNorm( const Real& p ) const
if( p == 1.0 )
return TNL::abs( this->data[ 0 ] ) + TNL::abs( this->data[ 1 ] );
if( p == 2.0 )
- return std::sqrt( this->data[ 0 ] * this->data[ 0 ] +
+ return TNL::sqrt( this->data[ 0 ] * this->data[ 0 ] +
this->data[ 1 ] * this->data[ 1 ] );
- return std::pow( std::pow( TNL::abs( this->data[ 0 ] ), p ) +
- std::pow( TNL::abs( this->data[ 1 ] ), p ), 1.0 / p );
+ return TNL::pow( TNL::pow( TNL::abs( this->data[ 0 ] ), p ) +
+ TNL::pow( TNL::abs( this->data[ 1 ] ), p ), 1.0 / p );
}
#ifdef UNDEF //TEMPLATE_EXPLICIT_INSTANTIATION
diff --git a/src/TNL/Containers/StaticVector3D_impl.h b/src/TNL/Containers/StaticVector3D_impl.h
index 68f5802e2772884305e3a6e6e50db913e1b8319d..d01e82077afd4feba62657d769803a68e1fe8fa1 100644
--- a/src/TNL/Containers/StaticVector3D_impl.h
+++ b/src/TNL/Containers/StaticVector3D_impl.h
@@ -10,6 +10,8 @@
#pragma once
+#include <TNL/Containers/StaticVector.h>
+
namespace TNL {
namespace Containers {
@@ -20,30 +22,31 @@ StaticVector< 3, Real >::StaticVector()
}
template< typename Real >
+ template< typename _unused >
__cuda_callable__
StaticVector< 3, Real >::StaticVector( const Real v[ 3 ] )
-: Containers::StaticArray< 3, Real >( v )
+: StaticArray< 3, Real >( v )
{
}
template< typename Real >
__cuda_callable__
StaticVector< 3, Real >::StaticVector( const Real& v )
-: Containers::StaticArray< 3, Real >( v )
+: StaticArray< 3, Real >( v )
{
}
template< typename Real >
__cuda_callable__
StaticVector< 3, Real >::StaticVector( const Real& v1, const Real& v2, const Real& v3 )
-: Containers::StaticArray< 3, Real >( v1, v2, v3 )
+: StaticArray< 3, Real >( v1, v2, v3 )
{
}
template< typename Real >
__cuda_callable__
StaticVector< 3, Real >::StaticVector( const StaticVector< 3, Real >& v )
-: Containers::StaticArray< 3, Real >( v )
+: StaticArray< 3, Real >( v )
{
}
@@ -175,6 +178,7 @@ bool StaticVector< 3, Real >::operator >= ( const StaticVector& v ) const
this->data[ 1 ] >= v[ 1 ] &&
this->data[ 2 ] >= v[ 2 ] );
}
+
template< typename Real >
template< typename OtherReal >
__cuda_callable__
@@ -193,9 +197,9 @@ __cuda_callable__
StaticVector< 3, Real >
StaticVector< 3, Real >::abs() const
{
- return StaticVector< 3, Real >( ::abs( this->data[ 0 ] ),
- ::abs( this->data[ 1 ] ),
- ::abs( this->data[ 2 ] ) );
+ return StaticVector< 3, Real >( TNL::abs( this->data[ 0 ] ),
+ TNL::abs( this->data[ 1 ] ),
+ TNL::abs( this->data[ 2 ] ) );
}
template< typename Real >
@@ -208,12 +212,12 @@ StaticVector< 3, Real >::lpNorm( const Real& p ) const
TNL::abs( this->data[ 1 ] ) +
TNL::abs( this->data[ 2 ] );
if( p == 2.0 )
- return std::sqrt( this->data[ 0 ] * this->data[ 0 ] +
+ return TNL::sqrt( this->data[ 0 ] * this->data[ 0 ] +
this->data[ 1 ] * this->data[ 1 ] +
this->data[ 2 ] * this->data[ 2 ] );
- return std::pow( std::pow( TNL::abs( this->data[ 0 ] ), p ) +
- std::pow( TNL::abs( this->data[ 1 ] ), p ) +
- std::pow( TNL::abs( this->data[ 2 ] ), p ), 1.0 / p );
+ return TNL::pow( TNL::pow( TNL::abs( this->data[ 0 ] ), p ) +
+ TNL::pow( TNL::abs( this->data[ 1 ] ), p ) +
+ TNL::pow( TNL::abs( this->data[ 2 ] ), p ), 1.0 / p );
}
diff --git a/src/TNL/Containers/StaticVector_impl.h b/src/TNL/Containers/StaticVector_impl.h
index 8c8d37b9c0c0e5b3f58a5f182f75f2837eddf156..45e680a098f36aeeed127295f210dcad5d629d28 100644
--- a/src/TNL/Containers/StaticVector_impl.h
+++ b/src/TNL/Containers/StaticVector_impl.h
@@ -10,7 +10,7 @@
#pragma once
-#include <TNL/Config/ParameterContainer.h>
+#include <TNL/Containers/StaticVector.h>
namespace TNL {
namespace Containers {
@@ -22,23 +22,24 @@ StaticVector< Size, Real >::StaticVector()
}
template< int Size, typename Real >
+ template< typename _unused >
__cuda_callable__
StaticVector< Size, Real >::StaticVector( const Real v[ Size ] )
-: Containers::StaticArray< Size, Real >( v )
+: StaticArray< Size, Real >( v )
{
}
template< int Size, typename Real >
__cuda_callable__
StaticVector< Size, Real >::StaticVector( const Real& v )
-: Containers::StaticArray< Size, Real >( v )
+: StaticArray< Size, Real >( v )
{
}
template< int Size, typename Real >
__cuda_callable__
StaticVector< Size, Real >::StaticVector( const StaticVector< Size, Real >& v )
-: Containers::StaticArray< Size, Real >( v )
+: StaticArray< Size, Real >( v )
{
}
@@ -210,12 +211,12 @@ StaticVector< Size, Real >::lpNorm( const Real& p ) const
Real aux = this->data[ 0 ] * this->data[ 0 ];
for( int i = 1; i < Size; i++ )
aux += this->data[ i ] * this->data[ i ];
- return std::sqrt( aux );
+ return TNL::sqrt( aux );
}
- Real aux = std::pow( TNL::abs( this->data[ 0 ] ), p );
+ Real aux = TNL::pow( TNL::abs( this->data[ 0 ] ), p );
for( int i = 1; i < Size; i++ )
- aux += std::pow( TNL::abs( this->data[ i ] ), p );
- return std::pow( aux, 1.0 / p );
+ aux += TNL::pow( TNL::abs( this->data[ i ] ), p );
+ return TNL::pow( aux, 1.0 / p );
}
template< int Size, typename Real, typename Scalar >
diff --git a/src/TNL/Curve.h b/src/TNL/Curve.h
index 6704b8a98865aabf36a4b5b8ce971c32c1802ff6..f6f0408db82016340bbce152de4fdeb3111ab4f8 100644
--- a/src/TNL/Curve.h
+++ b/src/TNL/Curve.h
@@ -35,36 +35,20 @@ class CurveElement
bool save( File& file ) const
{
-#ifdef HAVE_NOT_CXX11
- if( ! file. write< const T, Devices::Host >( &position ) )
- return false;
- if( ! file. write< const bool, Devices::Host >( &separator ) )
- return false;
- return true;
-#else
if( ! file. write( &position ) )
return false;
if( ! file. write( &separator ) )
return false;
return true;
-#endif
};
bool load( File& file )
{
-#ifdef HAVE_NOT_CXX11
- if( ! file. read< T, Devices::Host >( &position ) )
- return false;
- if( ! file. read< bool, Devices::Host >( &separator ) )
- return false;
- return true;
-#else
if( ! file. read( &position ) )
return false;
if( ! file. read( &separator ) )
return false;
return true;
-#endif
};
T position;
@@ -182,7 +166,7 @@ template< class T > bool Write( const Curve< T >& curve,
if( strncmp( format, "tnl",3 ) == 0 )
{
File file;
- if( ! file. open( String( file_name ) + String( ".tnl" ), tnlWriteMode ) )
+ if( ! file. open( String( file_name ) + String( ".tnl" ), IOMode::write ) )
{
std::cerr << "I am not able to open the file " << file_name << " for drawing curve." << std::endl;
return false;
@@ -218,7 +202,7 @@ template< class T > bool Read( Curve< T >& crv,
const char* input_file )
{
File file;
- if( ! file. open( String( input_file ), tnlReadMode ) )
+ if( ! file. open( String( input_file ), IOMode::read ) )
{
std::cout << " unable to open file " << input_file << std::endl;
return false;
diff --git a/src/TNL/Debugging/CMakeLists.txt b/src/TNL/Debugging/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/DevicePointer.h b/src/TNL/DevicePointer.h
index 3a8bd43ff2a03e390446156952fc6571be362454..956973529cbc39ad4e9bc42366ae118298b03ca2 100644
--- a/src/TNL/DevicePointer.h
+++ b/src/TNL/DevicePointer.h
@@ -18,6 +18,8 @@
#include <cstring>
+#include "Devices/MIC.h"
+
namespace TNL {
@@ -113,11 +115,18 @@ class DevicePointer< Object, Devices::Host > : public SmartPointer
return *( this->pointer );
}
- operator bool()
+ __cuda_callable__
+ operator bool() const
{
return this->pointer;
}
+ __cuda_callable__
+ bool operator!() const
+ {
+ return ! this->pointer;
+ }
+
template< typename Device = Devices::Host >
__cuda_callable__
const Object& getData() const
@@ -280,11 +289,18 @@ class DevicePointer< Object, Devices::Cuda > : public SmartPointer
return *( this->pointer );
}
- operator bool()
+ __cuda_callable__
+ operator bool() const
{
return this->pd;
}
+ __cuda_callable__
+ bool operator!() const
+ {
+ return ! this->pd;
+ }
+
template< typename Device = Devices::Host >
__cuda_callable__
const Object& getData() const
@@ -378,7 +394,7 @@ class DevicePointer< Object, Devices::Cuda > : public SmartPointer
TNL_ASSERT( this->pointer, );
TNL_ASSERT( this->cuda_pointer, );
cudaMemcpy( (void*) this->cuda_pointer, (void*) this->pointer, sizeof( ObjectType ), cudaMemcpyHostToDevice );
- if( ! checkCudaDevice ) {
+ if( ! TNL_CHECK_CUDA_DEVICE ) {
return false;
}
this->set_last_sync_state();
@@ -409,12 +425,8 @@ class DevicePointer< Object, Devices::Cuda > : public SmartPointer
{
this->pointer = &obj;
this->pd = new PointerData();
- if( ! this->pd )
- return false;
// pass to device
this->cuda_pointer = Devices::Cuda::passToDevice( *this->pointer );
- if( ! this->cuda_pointer )
- return false;
// set last-sync state
this->set_last_sync_state();
Devices::Cuda::insertSmartPointer( this );
@@ -462,4 +474,304 @@ class DevicePointer< Object, Devices::Cuda > : public SmartPointer
Object* cuda_pointer;
};
+/****
+ * Specialization for MIC
+ */
+
+#ifdef HAVE_MIC
+template< typename Object >
+class DevicePointer< Object, Devices::MIC > : public SmartPointer
+{
+ private:
+ // Convenient template alias for controlling the selection of copy- and
+ // move-constructors and assignment operators using SFINAE.
+ // The type Object_ is "enabled" iff Object_ and Object are not the same,
+ // but after removing const and volatile qualifiers they are the same.
+ template< typename Object_ >
+ using Enabler = std::enable_if< ! std::is_same< Object_, Object >::value &&
+ std::is_same< typename std::remove_cv< Object >::type, Object_ >::value >;
+
+ // friend class will be needed for templated assignment operators
+ template< typename Object_, typename Device_ >
+ friend class DevicePointer;
+
+ public:
+
+ typedef Object ObjectType;
+ typedef Devices::MIC DeviceType;
+ typedef DevicePointer< Object, Devices::MIC > ThisType;
+
+ explicit DevicePointer( ObjectType& obj )
+ : pointer( nullptr ),
+ pd( nullptr ),
+ mic_pointer( nullptr )
+ {
+ this->allocate( obj );
+ }
+
+ // this is needed only to avoid the default compiler-generated constructor
+ DevicePointer( const ThisType& pointer )
+ : pointer( pointer.pointer ),
+ pd( (PointerData*) pointer.pd ),
+ mic_pointer( pointer.mic_pointer )
+ {
+ this->pd->counter += 1;
+ }
+
+ // conditional constructor for non-const -> const data
+ template< typename Object_,
+ typename = typename Enabler< Object_ >::type >
+ DevicePointer( const DevicePointer< Object_, DeviceType >& pointer )
+ : pointer( pointer.pointer ),
+ pd( (PointerData*) pointer.pd ),
+ mic_pointer( pointer.mic_pointer )
+ {
+ this->pd->counter += 1;
+ }
+
+ // this is needed only to avoid the default compiler-generated constructor
+ DevicePointer( ThisType&& pointer )
+ : pointer( pointer.pointer ),
+ pd( (PointerData*) pointer.pd ),
+ mic_pointer( pointer.mic_pointer )
+ {
+ pointer.pointer = nullptr;
+ pointer.pd = nullptr;
+ pointer.mic_pointer = nullptr;
+ }
+
+ // conditional constructor for non-const -> const data
+ template< typename Object_,
+ typename = typename Enabler< Object_ >::type >
+ DevicePointer( DevicePointer< Object_, DeviceType >&& pointer )
+ : pointer( pointer.pointer ),
+ pd( (PointerData*) pointer.pd ),
+ mic_pointer( pointer.mic_pointer )
+ {
+ pointer.pointer = nullptr;
+ pointer.pd = nullptr;
+ pointer.mic_pointer = nullptr;
+ }
+
+ const Object* operator->() const
+ {
+ return this->pointer;
+ }
+
+ Object* operator->()
+ {
+ this->pd->maybe_modified = true;
+ return this->pointer;
+ }
+
+ const Object& operator *() const
+ {
+ return *( this->pointer );
+ }
+
+ Object& operator *()
+ {
+ this->pd->maybe_modified = true;
+ return *( this->pointer );
+ }
+
+ operator bool()
+ {
+ return this->pd;
+ }
+
+ template< typename Device = Devices::Host >
+ __cuda_callable__
+ const Object& getData() const
+ {
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::MIC >::value, "Only Devices::Host or Devices::MIC devices are accepted here." );
+ TNL_ASSERT( this->pointer, );
+ TNL_ASSERT( this->pd, );
+ TNL_ASSERT( this->mic_pointer, );
+ if( std::is_same< Device, Devices::Host >::value )
+ return *( this->pointer );
+ if( std::is_same< Device, Devices::MIC >::value )
+ return *( this->mic_pointer );
+ }
+
+ template< typename Device = Devices::Host >
+ __cuda_callable__
+ Object& modifyData()
+ {
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::MIC >::value, "Only Devices::Host or Devices::MIC devices are accepted here." );
+ TNL_ASSERT( this->pointer, );
+ TNL_ASSERT( this->pd, );
+ TNL_ASSERT( this->mic_pointer, );
+ if( std::is_same< Device, Devices::Host >::value )
+ {
+ this->pd->maybe_modified = true;
+ return *( this->pointer );
+ }
+ if( std::is_same< Device, Devices::MIC >::value )
+ return *( this->mic_pointer );
+ }
+
+ // this is needed only to avoid the default compiler-generated operator
+ const ThisType& operator=( const ThisType& ptr )
+ {
+ this->free();
+ this->pointer = ptr.pointer;
+ this->pd = (PointerData*) ptr.pd;
+ this->mic_pointer = ptr.mic_pointer;
+ this->pd->counter += 1;
+ return *this;
+ }
+
+ // conditional operator for non-const -> const data
+ template< typename Object_,
+ typename = typename Enabler< Object_ >::type >
+ const ThisType& operator=( const DevicePointer< Object_, DeviceType >& ptr )
+ {
+ this->free();
+ this->pointer = ptr.pointer;
+ this->pd = (PointerData*) ptr.pd;
+ this->mic_pointer = ptr.mic_pointer;
+ this->pd->counter += 1;
+ return *this;
+ }
+
+ // this is needed only to avoid the default compiler-generated operator
+ const ThisType& operator=( ThisType&& ptr )
+ {
+ this->free();
+ this->pointer = ptr.pointer;
+ this->pd = (PointerData*) ptr.pd;
+ this->mic_pointer = ptr.mic_pointer;
+ ptr.pointer = nullptr;
+ ptr.pd = nullptr;
+ ptr.mic_pointer = nullptr;
+ return *this;
+ }
+
+ // conditional operator for non-const -> const data
+ template< typename Object_,
+ typename = typename Enabler< Object_ >::type >
+ const ThisType& operator=( DevicePointer< Object_, DeviceType >&& ptr )
+ {
+ this->free();
+ this->pointer = ptr.pointer;
+ this->pd = (PointerData*) ptr.pd;
+ this->mic_pointer = ptr.mic_pointer;
+ ptr.pointer = nullptr;
+ ptr.pd = nullptr;
+ ptr.mic_pointer = nullptr;
+ return *this;
+ }
+
+ bool synchronize()
+ {
+ if( ! this->pd )
+ return true;
+ if( this->modified() )
+ {
+ TNL_ASSERT( this->pointer, );
+ TNL_ASSERT( this->mic_pointer, );
+ Devices::MIC::CopyToMIC((void*) this->mic_pointer, (void*) this->pointer, sizeof( ObjectType ));
+ this->set_last_sync_state();
+ return true;
+ }
+ return true;
+
+ }
+
+ ~DevicePointer()
+ {
+ this->free();
+ Devices::MIC::removeSmartPointer( this );
+ }
+
+ protected:
+
+ struct PointerData
+ {
+ char data_image[ sizeof(Object) ];
+ int counter = 1;
+ bool maybe_modified = false;
+ };
+
+ bool allocate( ObjectType& obj )
+ {
+ this->pointer = &obj;
+ this->pd = new PointerData();
+ if( ! this->pd )
+ return false;
+ // pass to device
+ this->mic_pointer = (ObjectType*)Devices::MIC::AllocMIC(sizeof(ObjectType));
+ if( ! this->mic_pointer )
+ return false;
+ Devices::MIC::CopyToMIC((void*)this->mic_pointer,(void*)this->pointer,sizeof(ObjectType));
+
+ // set last-sync state
+ this->set_last_sync_state();
+ Devices::MIC::insertSmartPointer( this );
+ return true;
+ }
+
+ void set_last_sync_state()
+ {
+ TNL_ASSERT( this->pointer, );
+ TNL_ASSERT( this->pd, );
+ std::memcpy( (void*) &this->pd->data_image, (void*) this->pointer, sizeof( Object ) );
+ this->pd->maybe_modified = false;
+ }
+
+ bool modified()
+ {
+ TNL_ASSERT( this->pointer, );
+ TNL_ASSERT( this->pd, );
+ // optimization: skip bitwise comparison if we're sure that the data is the same
+ if( ! this->pd->maybe_modified )
+ return false;
+ return std::memcmp( (void*) &this->pd->data_image, (void*) this->pointer, sizeof( Object ) ) != 0;
+ }
+
+ void free()
+ {
+ if( this->pd )
+ {
+ if( ! --this->pd->counter )
+ {
+ delete this->pd;
+ this->pd = nullptr;
+ if( this->mic_pointer )
+ Devices::MIC::FreeMIC( (void*) this->mic_pointer );
+ }
+ }
+ }
+
+ Object* pointer;
+
+ PointerData* pd;
+
+ // mic_pointer can't be part of PointerData structure, since we would be
+ // unable to dereference this-pd on the device
+ Object* mic_pointer;
+};
+#endif
+
+
+#if (!defined(NDEBUG)) && (!defined(HAVE_MIC))
+namespace Assert {
+
+template< typename Object, typename Device >
+struct Formatter< DevicePointer< Object, Device > >
+{
+ static std::string
+ printToString( const DevicePointer< Object, Device >& value )
+ {
+ ::std::stringstream ss;
+ ss << "(DevicePointer< " << Object::getType() << ", " << Device::getDeviceType()
+ << " > object at " << &value << ")";
+ return ss.str();
+ }
+};
+
+} // namespace Assert
+#endif
+
} // namespace TNL
diff --git a/src/TNL/Devices/CMakeLists.txt b/src/TNL/Devices/CMakeLists.txt
old mode 100755
new mode 100644
index f09e2b47947ea21af93619167ea415f6b62491e6..7b4babe28e75b075edecad776b68cd18983aa0bb
--- a/src/TNL/Devices/CMakeLists.txt
+++ b/src/TNL/Devices/CMakeLists.txt
@@ -1,12 +1,15 @@
set (headers Cuda.h
Cuda_impl.h
+ CudaCallable.h
CudaDeviceInfo.h
- Host.h )
+ Host.h
+ MIC.h )
SET( CURRENT_DIR ${CMAKE_SOURCE_DIR}/src/TNL/Devices )
set( common_SOURCES
${CURRENT_DIR}/Cuda.cpp
- ${CURRENT_DIR}/Host.cpp )
+ ${CURRENT_DIR}/Host.cpp
+ ${CURRENT_DIR}/MIC.cpp )
IF( BUILD_CUDA )
set( tnl_devices_CUDA__SOURCES
diff --git a/src/TNL/Devices/Cuda.cu b/src/TNL/Devices/Cuda.cu
index 0e59e2da5c75ede9f037a470bc16fdacaadcb855..2605e6dca83290eb59db54618b7bf91ed1e59150 100644
--- a/src/TNL/Devices/Cuda.cu
+++ b/src/TNL/Devices/Cuda.cu
@@ -9,6 +9,7 @@
/* See Copyright Notice in tnl/Copyright */
#include <TNL/Devices/Cuda.h>
+#include <TNL/Exceptions/CudaRuntimeError.h>
#include <TNL/Config/ConfigDescription.h>
#include <TNL/Config/ParameterContainer.h>
@@ -106,382 +107,7 @@ bool Cuda::checkDevice( const char* file_name, int line, cudaError error )
{
if( error == cudaSuccess )
return true;
- std::cerr << "CUDA ERROR(" << error << ") at line " << line << " in " << file_name << ":" << std::endl;
- std::cerr << cudaGetErrorString( error ) << std::endl;
- std::cerr << "Detailed description is: " << std::endl;
- switch( error )
- {
- // 1
- case cudaErrorMissingConfiguration:
- std::cerr
- << "The device function being invoked (usually via ::cudaLaunch()) was not " << std::endl
- << "previously configured via the ::cudaConfigureCall() function. " << std::endl;
- break;
-
- // 2
- case cudaErrorMemoryAllocation:
- std::cerr
- << "The API call failed because it was unable to allocate enough memory to " << std::endl
- << "perform the requested operation. " << std::endl;
- break;
-
- // 3
- case cudaErrorInitializationError:
- std::cerr
- << "The API call failed because the CUDA driver and runtime could not be " << std::endl
- << "initialized. " << std::endl;
- break;
-
- // 4
- case cudaErrorLaunchFailure:
- std::cerr
- << "An exception occurred on the device while executing a kernel. Common " << std::endl
- << "causes include dereferencing an invalid device pointer and accessing " << std::endl
- << "out of bounds shared memory. The device cannot be used until " << std::endl
- << "::cudaThreadExit() is called. All existing device memory allocations " << std::endl
- << "are invalid and must be reconstructed if the program is to continue " << std::endl
- << "using CUDA. " << std::endl;
- break;
-
- // 5
- case cudaErrorPriorLaunchFailure:
- std::cerr
- << "This indicated that a previous kernel launch failed. This was previously " << std::endl
- << "used for device emulation of kernel launches. " << std::endl
- << "This error return is deprecated as of CUDA 3.1. Device emulation mode was " << std::endl
- << "removed with the CUDA 3.1 release. " << std::endl;
- break;
-
- // 6
- case cudaErrorLaunchTimeout:
- std::cerr
- << "This indicates that the device kernel took too long to execute. This can " << std::endl
- << "only occur if timeouts are enabled - see the device property " << std::endl
- << "ref ::cudaDeviceProp::kernelExecTimeoutEnabled \"kernelExecTimeoutEnabled\" " << std::endl
- << "for more information. The device cannot be used until ::cudaThreadExit() " << std::endl
- << "is called. All existing device memory allocations are invalid and must be " << std::endl
- << "reconstructed if the program is to continue using CUDA. " << std::endl;
- break;
-
- // 7
- case cudaErrorLaunchOutOfResources:
- std::cerr
- << "This indicates that a launch did not occur because it did not have " << std::endl
- << "appropriate resources. Although this error is similar to " << std::endl
- << "::cudaErrorInvalidConfiguration, this error usually indicates that the " << std::endl
- << "user has attempted to pass too many arguments to the device kernel, or the " << std::endl
- << "kernel launch specifies too many threads for the kernel's register count. " << std::endl;
- break;
-
- // 8
- case cudaErrorInvalidDeviceFunction:
- std::cerr
- << "The requested device function does not exist or is not compiled for the " << std::endl
- << "proper device architecture. " << std::endl;
- break;
-
- // 9
- case cudaErrorInvalidConfiguration:
- std::cerr
- << "This indicates that a kernel launch is requesting resources that can " << std::endl
- << "never be satisfied by the current device. Requesting more shared memory " << std::endl
- << "per block than the device supports will trigger this error, as will " << std::endl
- << "requesting too many threads or blocks. See ::cudaDeviceProp for more " << std::endl
- << "device limitations. " << std::endl;
- break;
-
- // 10
- case cudaErrorInvalidDevice:
- std::cerr
- << "This indicates that the device ordinal supplied by the user does not " << std::endl
- << "correspond to a valid CUDA device. " << std::endl;
- break;
-
- // 11
- case cudaErrorInvalidValue:
- std::cerr
- << "This indicates that one or more of the parameters passed to the API call " << std::endl
- << "is not within an acceptable range of values. " << std::endl;
- break;
-
- // 12
- case cudaErrorInvalidPitchValue:
- std::cerr
- << "This indicates that one or more of the pitch-related parameters passed " << std::endl
- << "to the API call is not within the acceptable range for pitch. " << std::endl;
- break;
-
- // 13
- case cudaErrorInvalidSymbol:
- std::cerr
- << "This indicates that the symbol name/identifier passed to the API call " << std::endl
- << "is not a valid name or identifier. " << std::endl;
- break;
-
- // 14
- case cudaErrorMapBufferObjectFailed:
- std::cerr
- << "This indicates that the buffer object could not be mapped. " << std::endl;
- break;
-
- // 15
- case cudaErrorUnmapBufferObjectFailed:
- std::cerr
- << "This indicates that the buffer object could not be unmapped. " << std::endl;
- break;
-
- // 16
- case cudaErrorInvalidHostPointer:
- std::cerr
- << "This indicates that at least one host pointer passed to the API call is " << std::endl
- << "not a valid host pointer. " << std::endl;
- break;
-
- // 17
- case cudaErrorInvalidDevicePointer:
- std::cerr
- << "This indicates that at least one device pointer passed to the API call is " << std::endl
- << "not a valid device pointer. " << std::endl;
- break;
-
- case cudaErrorInvalidTexture:
- std::cerr
- << "This indicates that the texture passed to the API call is not a valid " << std::endl
- << "texture. " << std::endl;
- break;
-
- case cudaErrorInvalidTextureBinding:
- std::cerr
- << "This indicates that the texture binding is not valid. This occurs if you " << std::endl
- << "call ::cudaGetTextureAlignmentOffset() with an unbound texture. " << std::endl;
- break;
-
- case cudaErrorInvalidChannelDescriptor:
- std::cerr
- << "This indicates that the channel descriptor passed to the API call is not " << std::endl
- << "valid. This occurs if the format is not one of the formats specified by " << std::endl
- << "::cudaChannelFormatKind, or if one of the dimensions is invalid. " << std::endl;
- break;
-
- case cudaErrorInvalidMemcpyDirection:
- std::cerr
- << "This indicates that the direction of the memcpy passed to the API call is " << std::endl
- << "not one of the types specified by ::cudaMemcpyKind. " << std::endl;
- break;
-
- case cudaErrorAddressOfConstant:
- std::cerr
- << "This indicated that the user has taken the address of a constant variable, " << std::endl
- << "which was forbidden up until the CUDA 3.1 release. " << std::endl
- << "This error return is deprecated as of CUDA 3.1. Variables in constant " << std::endl
- << "memory may now have their address taken by the runtime via " << std::endl
- << "::cudaGetSymbolAddress(). " << std::endl;
- break;
-
- case cudaErrorTextureFetchFailed:
- std::cerr
- << "This indicated that a texture fetch was not able to be performed. " << std::endl
- << "This was previously used for device emulation of texture operations. " << std::endl
- << "This error return is deprecated as of CUDA 3.1. Device emulation mode was " << std::endl
- << "removed with the CUDA 3.1 release. " << std::endl;
- break;
-
- case cudaErrorTextureNotBound:
- std::cerr
- << "This indicated that a texture was not bound for access. " << std::endl
- << "This was previously used for device emulation of texture operations. " << std::endl
- << "This error return is deprecated as of CUDA 3.1. Device emulation mode was " << std::endl
- << "removed with the CUDA 3.1 release. " << std::endl;
- break;
-
- case cudaErrorSynchronizationError:
- std::cerr
- << "This indicated that a synchronization operation had failed. " << std::endl
- << "This was previously used for some device emulation functions. " << std::endl
- << "This error return is deprecated as of CUDA 3.1. Device emulation mode was " << std::endl
- << "removed with the CUDA 3.1 release. " << std::endl;
- break;
-
- case cudaErrorInvalidFilterSetting:
- std::cerr
- << "This indicates that a non-float texture was being accessed with linear " << std::endl
- << "filtering. This is not supported by CUDA. " << std::endl;
- break;
-
- case cudaErrorInvalidNormSetting:
- std::cerr
- << "This indicates that an attempt was made to read a non-float texture as a " << std::endl
- << "normalized float. This is not supported by CUDA. " << std::endl;
- break;
-
- case cudaErrorMixedDeviceExecution:
- std::cerr
- << "Mixing of device and device emulation code was not allowed. " << std::endl
- << "This error return is deprecated as of CUDA 3.1. Device emulation mode was " << std::endl
- << "removed with the CUDA 3.1 release. " << std::endl;
- break;
-
- case cudaErrorCudartUnloading:
- std::cerr
- << "This indicated an issue with calling API functions during the unload " << std::endl
- << "process of the CUDA runtime in prior releases. " << std::endl
- << "This error return is deprecated as of CUDA 3.2. " << std::endl;
- break;
-
- case cudaErrorUnknown:
- std::cerr
- << "This indicates that an unknown internal error has occurred. " << std::endl;
- break;
-
- case cudaErrorNotYetImplemented:
- std::cerr
- << "This indicates that the API call is not yet implemented. Production " << std::endl
- << "releases of CUDA will never return this error. " << std::endl;
- break;
-
- case cudaErrorMemoryValueTooLarge:
- std::cerr
- << "This indicated that an emulated device pointer exceeded the 32-bit address " << std::endl
- << "range. " << std::endl
- << "This error return is deprecated as of CUDA 3.1. Device emulation mode was " << std::endl
- << "removed with the CUDA 3.1 release. " << std::endl;
- break;
-
- case cudaErrorInvalidResourceHandle:
- std::cerr
- << "This indicates that a resource handle passed to the API call was not " << std::endl
- << "valid. Resource handles are opaque types like ::cudaStream_t and " << std::endl
- << "::cudaEvent_t. " << std::endl;
- break;
-
- case cudaErrorNotReady:
- std::cerr
- << "This indicates that asynchronous operations issued previously have not " << std::endl
- << "completed yet. This result is not actually an error, but must be indicated " << std::endl
- << "differently than ::cudaSuccess (which indicates completion). Calls that " << std::endl
- << "may return this value include ::cudaEventQuery() and ::cudaStreamQuery(). " << std::endl;
- break;
-
- case cudaErrorInsufficientDriver:
- std::cerr
- << "This indicates that the installed NVIDIA CUDA driver is older than the " << std::endl
- << "CUDA runtime library. This is not a supported configuration. Users should " << std::endl
- << "install an updated NVIDIA display driver to allow the application to run. " << std::endl;
- break;
-
- case cudaErrorSetOnActiveProcess:
- std::cerr
- << "This indicates that the user has called ::cudaSetDevice(), " << std::endl
- << "::cudaSetValidDevices(), ::cudaSetDeviceFlags(), " << std::endl
- << "::cudaD3D9SetDirect3DDevice(), ::cudaD3D10SetDirect3DDevice, " << std::endl
- << "::cudaD3D11SetDirect3DDevice(), * or ::cudaVDPAUSetVDPAUDevice() after " << std::endl
- << "initializing the CUDA runtime by calling non-device management operations " << std::endl
- << "(allocating memory and launching kernels are examples of non-device " << std::endl
- << "management operations). This error can also be returned if using " << std::endl
- << "runtime/driver interoperability and there is an existing ::CUcontext " << std::endl
- << "active on the host thread. " << std::endl;
- break;
-
- case cudaErrorInvalidSurface:
- std::cerr
- << "This indicates that the surface passed to the API call is not a valid " << std::endl
- << "surface. " << std::endl;
- break;
-
- case cudaErrorNoDevice:
- std::cerr
- << "This indicates that no CUDA-capable devices were detected by the installed " << std::endl
- << "CUDA driver. " << std::endl;
- break;
-
- case cudaErrorECCUncorrectable:
- std::cerr
- << "This indicates that an uncorrectable ECC error was detected during " << std::endl
- << "execution. " << std::endl;
- break;
-
- case cudaErrorSharedObjectSymbolNotFound:
- std::cerr
- << "This indicates that a link to a shared object failed to resolve. " << std::endl;
- break;
-
- case cudaErrorSharedObjectInitFailed:
- std::cerr
- << "This indicates that initialization of a shared object failed. " << std::endl;
- break;
-
- case cudaErrorUnsupportedLimit:
- std::cerr
- << "This indicates that the ::cudaLimit passed to the API call is not " << std::endl
- << "supported by the active device. " << std::endl;
- break;
-
- case cudaErrorDuplicateVariableName:
- std::cerr
- << "This indicates that multiple global or constant variables (across separate " << std::endl
- << "CUDA source files in the application) share the same string name. " << std::endl;
- break;
-
- case cudaErrorDuplicateTextureName:
- std::cerr
- << "This indicates that multiple textures (across separate CUDA source " << std::endl
- << "files in the application) share the same string name. " << std::endl;
- break;
-
- case cudaErrorDuplicateSurfaceName:
- std::cerr
- << "This indicates that multiple surfaces (across separate CUDA source " << std::endl
- << "files in the application) share the same string name. " << std::endl;
- break;
-
- case cudaErrorDevicesUnavailable:
- std::cerr
- << "This indicates that all CUDA devices are busy or unavailable at the current " << std::endl
- << "time. Devices are often busy/unavailable due to use of " << std::endl
- << "::cudaComputeModeExclusive or ::cudaComputeModeProhibited. They can also " << std::endl
- << "be unavailable due to memory constraints on a device that already has " << std::endl
- << "active CUDA work being performed. " << std::endl;
- break;
-
- case cudaErrorInvalidKernelImage:
- std::cerr
- << "This indicates that the device kernel image is invalid. " << std::endl;
- break;
-
- case cudaErrorNoKernelImageForDevice:
- std::cerr
- << "This indicates that there is no kernel image available that is suitable " << std::endl
- << "for the device. This can occur when a user specifies code generation " << std::endl
- << "options for a particular CUDA source file that do not include the " << std::endl
- << "corresponding device configuration. " << std::endl;
- break;
-
- case cudaErrorIncompatibleDriverContext:
- std::cerr
- << "This indicates that the current context is not compatible with this " << std::endl
- << "version of the CUDA Runtime. This can only occur if you are using CUDA " << std::endl
- << "Runtime/Driver interoperability and have created an existing Driver " << std::endl
- << "context using an older API. Please see \\ref CUDART_DRIVER " << std::endl
- << "\"Interactions with the CUDA Driver API\" for more information. " << std::endl;
- break;
-
- case cudaErrorStartupFailure:
- std::cerr
- << "This indicates an internal startup failure in the CUDA runtime. " << std::endl;
- break;
-
- case cudaErrorApiFailureBase:
- std::cerr
- << "Any unhandled CUDA driver error is added to this value and returned via " << std::endl
- << "the runtime. Production releases of CUDA should not return such errors. " << std::endl;
- break;
-
- default:
- std::cerr << "(detailed description is not available)" << std::endl;
- break;
- }
- throw EXIT_FAILURE;
- return false;
+ throw Exceptions::CudaRuntimeError( error, file_name, line );
}
std::ostream& operator << ( std::ostream& str, const dim3& d )
diff --git a/src/TNL/Devices/Cuda.h b/src/TNL/Devices/Cuda.h
index 6333b1ab5fb4f695815515cf742d370e0a454dca..805a9c8a35845d2a48a583ecd28fc9559194e676 100644
--- a/src/TNL/Devices/Cuda.h
+++ b/src/TNL/Devices/Cuda.h
@@ -1,5 +1,5 @@
/***************************************************************************
- Devices::Cuda.h - description
+ Cuda.h - description
-------------------
begin : Nov 7, 2012
copyright : (C) 2012 by Tomas Oberhuber
@@ -16,6 +16,7 @@
#include <TNL/Assert.h>
#include <TNL/SmartPointersRegister.h>
#include <TNL/Timer.h>
+#include <TNL/Devices/CudaCallable.h>
namespace TNL {
@@ -26,12 +27,6 @@ namespace Config {
namespace Devices {
-#ifdef HAVE_CUDA
-#define __cuda_callable__ __device__ __host__
-#else
-#define __cuda_callable__
-#endif
-
class Cuda
{
public:
@@ -156,7 +151,7 @@ class Cuda
/****
* I do not know why, but it is more reliable to pass the error code instead
* of calling cudaGetLastError() inside the method.
- * We recommend to use macro 'checkCudaDevice' defined bellow.
+ * We recommend to use macro 'TNL_CHECK_CUDA_DEVICE' defined bellow.
*/
static bool checkDevice( const char* file_name, int line, cudaError error );
#else
@@ -184,14 +179,14 @@ class Cuda
};
#ifdef HAVE_CUDA
-#define checkCudaDevice ::TNL::Devices::Cuda::checkDevice( __FILE__, __LINE__, cudaGetLastError() )
-std::ostream& operator << ( std::ostream& str, const dim3& d );
+#define TNL_CHECK_CUDA_DEVICE ::TNL::Devices::Cuda::checkDevice( __FILE__, __LINE__, cudaGetLastError() )
#else
-#define checkCudaDevice ::TNL::Devices::Cuda::checkDevice()
+#define TNL_CHECK_CUDA_DEVICE ::TNL::Devices::Cuda::checkDevice()
#endif
-#define CudaSupportMissingMessage \
- std::cerr << "The CUDA support is missing in the source file " << __FILE__ << " at line " << __LINE__ << ". Please set WITH_CUDA=yes in the install script. " << std::endl;
+#ifdef HAVE_CUDA
+std::ostream& operator << ( std::ostream& str, const dim3& d );
+#endif
} // namespace Devices
} // namespace TNL
diff --git a/src/TNL/Devices/CudaCallable.h b/src/TNL/Devices/CudaCallable.h
new file mode 100644
index 0000000000000000000000000000000000000000..e0a86a3e4f01a8ae112a45c7528f77e152289b04
--- /dev/null
+++ b/src/TNL/Devices/CudaCallable.h
@@ -0,0 +1,25 @@
+/***************************************************************************
+ CudaCallable.h - description
+ -------------------
+ begin : Jun 20, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#pragma once
+
+// The __cuda_callable__ macro has to be in a separate header file to avoid
+// infinite loops by the #include directives.
+//
+// For example, the implementation of Devices::Cuda needs TNL_ASSERT_*
+// macros, which need __cuda_callable__ functions.
+
+#ifdef HAVE_MIC
+ #define __cuda_callable__ __attribute__((target(mic)))
+#elif HAVE_CUDA
+ #define __cuda_callable__ __device__ __host__
+#else
+ #define __cuda_callable__
+#endif
diff --git a/src/TNL/Devices/CudaDeviceInfo.cpp b/src/TNL/Devices/CudaDeviceInfo.cpp
index 45199ecda1fa986dc68d0fd5386b06c09e8b7ecc..85a6604d87bba8655aac46609d4ae0db8dc24934 100644
--- a/src/TNL/Devices/CudaDeviceInfo.cpp
+++ b/src/TNL/Devices/CudaDeviceInfo.cpp
@@ -107,6 +107,13 @@ getCudaCores( int deviceNum )
return 0;
}
+int
+CudaDeviceInfo::
+getRegistersPerMultiprocessor( int deviceNum )
+{
+ return 0;
+}
+
void
CudaDeviceInfo::
writeDeviceInfo( Logger& logger )
diff --git a/src/TNL/Devices/CudaDeviceInfo.cu b/src/TNL/Devices/CudaDeviceInfo.cu
index 84096561dcb11de412828b07e43403132af493f4..56905df9675ef10b07ba40d1876215376ade611c 100644
--- a/src/TNL/Devices/CudaDeviceInfo.cu
+++ b/src/TNL/Devices/CudaDeviceInfo.cu
@@ -135,17 +135,26 @@ getCudaCoresPerMultiprocessors( int deviceNum )
case 1: // Tesla generation, G80, G8x, G9x classes
return 8;
case 2: // Fermi generation
- switch( minor )
- {
- case 0: // GF100 class
- return 32;
- case 1: // GF10x class
- return 48;
- }
+ switch( minor )
+ {
+ case 0: // GF100 class
+ return 32;
+ case 1: // GF10x class
+ return 48;
+ }
case 3: // Kepler generation -- GK10x, GK11x classes
return 192;
case 5: // Maxwell generation -- GM10x, GM20x classes
return 128;
+ case 6: // Pascal generation
+ switch( minor )
+ {
+ case 0: // GP100 class
+ return 64;
+ case 1: // GP10x classes
+ case 2:
+ return 128;
+ }
default:
return -1;
}
@@ -159,6 +168,21 @@ getCudaCores( int deviceNum )
CudaDeviceInfo::getCudaCoresPerMultiprocessors( deviceNum );
}
+int
+CudaDeviceInfo::
+getRegistersPerMultiprocessor( int deviceNum )
+{
+ // results are cached because they are used for configuration of some kernels
+ static std::unordered_map< int, int > results;
+ if( results.count( deviceNum ) == 0 ) {
+ cudaDeviceProp properties;
+ cudaGetDeviceProperties( &properties, deviceNum );
+ results.emplace( deviceNum, properties.regsPerMultiprocessor );
+ return properties.regsPerMultiprocessor;
+ }
+ return results[ deviceNum ];
+}
+
void
CudaDeviceInfo::
writeDeviceInfo( Logger& logger )
diff --git a/src/TNL/Devices/CudaDeviceInfo.h b/src/TNL/Devices/CudaDeviceInfo.h
index b658e917703f8d97a7caca76c8055a845670506d..0b02daee53071d5899f4de14041e40f2da96adaa 100644
--- a/src/TNL/Devices/CudaDeviceInfo.h
+++ b/src/TNL/Devices/CudaDeviceInfo.h
@@ -50,6 +50,8 @@ class CudaDeviceInfo
static int getCudaCores( int deviceNum );
+ static int getRegistersPerMultiprocessor( int deviceNum );
+
static void writeDeviceInfo( Logger& logger );
};
diff --git a/src/TNL/Devices/Cuda_impl.h b/src/TNL/Devices/Cuda_impl.h
index cea827a03c3dbda07fb0f5b347bbb61ca6067bf1..19bb9db873e3706cfbd9c3de18e9e69bf9bf7527 100644
--- a/src/TNL/Devices/Cuda_impl.h
+++ b/src/TNL/Devices/Cuda_impl.h
@@ -11,6 +11,8 @@
#pragma once
#include <TNL/Devices/Cuda.h>
+#include <TNL/Exceptions/CudaBadAlloc.h>
+#include <TNL/Exceptions/CudaSupportMissing.h>
namespace TNL {
namespace Devices {
@@ -68,7 +70,6 @@ __device__ inline int Cuda::getGlobalThreadIdx_z( const dim3& gridIdx )
{
return ( gridIdx.z * getMaxGridSize() + blockIdx.z ) * blockDim.z + threadIdx.z;
}
-
#endif
@@ -79,23 +80,20 @@ ObjectType* Cuda::passToDevice( const ObjectType& object )
ObjectType* deviceObject;
if( cudaMalloc( ( void** ) &deviceObject,
( size_t ) sizeof( ObjectType ) ) != cudaSuccess )
- {
- checkCudaDevice;
- return 0;
- }
+ throw Exceptions::CudaBadAlloc();
if( cudaMemcpy( ( void* ) deviceObject,
( void* ) &object,
sizeof( ObjectType ),
cudaMemcpyHostToDevice ) != cudaSuccess )
{
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cudaFree( ( void* ) deviceObject );
+ TNL_CHECK_CUDA_DEVICE;
return 0;
}
return deviceObject;
#else
- TNL_ASSERT( false, std::cerr << "CUDA support is missing." );
- return 0;
+ throw Exceptions::CudaSupportMissing();
#endif
}
@@ -108,26 +106,25 @@ ObjectType Cuda::passFromDevice( const ObjectType* object )
( void* ) &object,
sizeof( ObjectType ),
cudaMemcpyDeviceToHost );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
return aux;
#else
- TNL_ASSERT( false, std::cerr << "CUDA support is missing." );
- return 0;
+ throw Exceptions::CudaSupportMissing();
#endif
}
template< typename ObjectType >
void Cuda::passFromDevice( const ObjectType* deviceObject,
- ObjectType& hostObject )
+ ObjectType& hostObject )
{
#ifdef HAVE_CUDA
cudaMemcpy( ( void* ) &hostObject,
( void* ) deviceObject,
sizeof( ObjectType ),
cudaMemcpyDeviceToHost );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#else
- TNL_ASSERT( false, std::cerr << "CUDA support is missing." );
+ throw Exceptions::CudaSupportMissing();
#endif
}
@@ -147,9 +144,9 @@ void Cuda::freeFromDevice( ObjectType* deviceObject )
{
#ifdef HAVE_CUDA
cudaFree( ( void* ) deviceObject );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#else
- TNL_ASSERT( false, std::cerr << "CUDA support is missing." );
+ throw Exceptions::CudaSupportMissing();
#endif
}
diff --git a/src/TNL/Devices/Host.cpp b/src/TNL/Devices/Host.cpp
index db56392bcb6ad8218e7f3a45f3d1173eae541d94..60f266734b1d6e1d8353bc8767f5b31a17bab48d 100644
--- a/src/TNL/Devices/Host.cpp
+++ b/src/TNL/Devices/Host.cpp
@@ -304,14 +304,22 @@ void Host::configSetup( Config::ConfigDescription& config, const String& prefix
bool Host::setup( const Config::ParameterContainer& parameters,
const String& prefix )
{
- if( parameters.getParameter< bool >( prefix + "openmp-enabled" ) )
+ if( parameters.getParameter< bool >( prefix + "openmp-enabled" ) ) {
+#ifdef HAVE_OPENMP
enableOMP();
+#else
+ std::cerr << "OpenMP is not supported - please recompile the TNL library with OpenMP." << std::endl;
+ return false;
+#endif
+ }
else
disableOMP();
- setMaxThreadsCount( parameters.getParameter< int >( prefix + "openmp-max-threads" ) );
+ const int threadsCount = parameters.getParameter< int >( prefix + "openmp-max-threads" );
+ if( threadsCount > 1 && ! isOMPEnabled() )
+ std::cerr << "Warning: openmp-max-threads was set to " << threadsCount << ", but OpenMP is disabled." << std::endl;
+ setMaxThreadsCount( threadsCount );
return true;
}
} // namespace Devices
} // namespace TNL
-
diff --git a/src/TNL/Devices/MIC.cpp b/src/TNL/Devices/MIC.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..b16ac7854dd29ced0ed433228f5aa783d1f017ac
--- /dev/null
+++ b/src/TNL/Devices/MIC.cpp
@@ -0,0 +1,41 @@
+/***************************************************************************
+ MIC.cpp - description
+ -------------------
+ begin : Feb 10, 2017
+ copyright : (C) 2017 by Tomas Oberhuber
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Vit Hanousek
+
+#include <TNL/Devices/MIC.h>
+
+namespace TNL {
+namespace Devices {
+
+SmartPointersRegister MIC::smartPointersRegister;
+Timer MIC::smartPointersSynchronizationTimer;
+
+void MIC::insertSmartPointer( SmartPointer* pointer )
+{
+ smartPointersRegister.insert( pointer, -1 );
+}
+
+void MIC::removeSmartPointer( SmartPointer* pointer )
+{
+ smartPointersRegister.remove( pointer, -1 );
+}
+
+bool MIC::synchronizeDevice( int deviceId )
+{
+ smartPointersSynchronizationTimer.start();
+ bool b = smartPointersRegister.synchronizeDevice( deviceId );
+ smartPointersSynchronizationTimer.stop();
+ return b;
+}
+
+
+} // namespace Devices
+} // namespace TNL
\ No newline at end of file
diff --git a/src/TNL/Devices/MIC.h b/src/TNL/Devices/MIC.h
new file mode 100644
index 0000000000000000000000000000000000000000..36678c0d27da1c873ee4bf0da2e71616c012d2bc
--- /dev/null
+++ b/src/TNL/Devices/MIC.h
@@ -0,0 +1,179 @@
+/***************************************************************************
+ MIC.h - description
+ -------------------
+ begin : Nov 7, 2016
+ copyright : (C) 2016 by Tomas Oberhuber
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Vit Hanousek
+
+#pragma once
+
+#include <iostream>
+#include <cstring>
+#include <unistd.h>
+#include <TNL/String.h>
+#include <TNL/Assert.h>
+#include <TNL/SmartPointersRegister.h>
+#include <TNL/Timer.h>
+
+#include <TNL/Devices/CudaCallable.h>
+
+
+namespace TNL {
+
+namespace Devices {
+
+//useful macros from Intel's tutorials -- but we do not use it, becaouse it is tricky (system of maping variables CPU-MIC)
+#define ALLOC alloc_if(1) //alloac variable at begining of offloaded block -- default
+#define FREE free_if(1) // delete variable at the end of offloaded block -- default
+#define RETAIN free_if(0) //do not delete variable at the end of offladed block
+#define REUSE alloc_if(0) //do not alloc variable at begin of offloaded block, reuse variable on MIC which was not deleted befeore
+
+//structure which hides pointer - bypass mapping of variables and addresses of arrays and allow get RAW addres of MIC memory to RAM
+template< typename Type >
+struct MICHider{
+ Type *pointer;
+};
+
+//inflatable structure -- structures can be copied to MIC - classes not (viz paper published after CSJP 2016 in Krakow)
+//object can be copied in side this structure and then copied into MIC memory
+template <unsigned int VELIKOST>
+struct MICStruct{
+ uint8_t data[VELIKOST];
+};
+
+//Macros which can make code better readeble --but they are tricky, creating variables with specific names...
+//version using inflatable structure
+#define TNLMICSTRUCT(bb,typ) Devices::MICStruct<sizeof(typ)> s ## bb; \
+ memcpy((void*)& s ## bb,(void*)& bb,sizeof(typ));
+#define TNLMICSTRUCTOFF(bb,typ) s ## bb
+#define TNLMICSTRUCTUSE(bb,typ) typ * kernel ## bb = (typ*) &s ## bb;
+#define TNLMICSTRUCTALLOC(bb,typ) typ * kernel ## bb = (typ*) malloc (sizeof(typ)); \
+ memcpy((void*)kernel ## bb,(void*) & s ## bb, sizeof(typ));
+
+//version which retypes pointer of object to pointer to array of uint8_t,
+//object can be copied using uint8_t pointer as array with same length as object size
+#define TNLMICHIDE(bb,typ) uint8_t * u ## bb=(uint8_t *)&bb; \
+ MICHider<typ> kernel ## bb;
+#define TNLMICHIDEALLOCOFF(bb,typ) in(u ## bb:length(sizeof(typ))) out(kernel ## bb)
+#define TNLMICHIDEALLOC(bb,typ) kernel ## bb.pointer=(typ*)malloc(sizeof(typ)); \
+ memcpy((void*)kernel ## bb.pointer,(void*)u ## bb,sizeof(typ));
+#define TNLMICHIDEFREEOFF(bb,typ) in(kernel ## bb)
+#define TNLMICHIDEFREE(bb,typ) free((void*)kernel ## bb.pointer
+
+class MIC
+{
+ public:
+
+ static String getDeviceType()
+ {
+ return String( "MIC" );
+ };
+
+#ifdef HAVE_MIC
+
+ //useful debuging -- but produce warning
+ __cuda_callable__ static inline void CheckMIC(void)
+ {
+ #ifdef __MIC__
+ std::cout<<"ON MIC"<<std::endl;
+ #else
+ std::cout<<"ON CPU" <<std::endl;
+ #endif
+ };
+
+
+ //old copying funciton -- deprecated
+ template <typename TYP>
+ static
+ TYP * passToDevice(TYP &objektCPU)
+ {
+ uint8_t * uk=(uint8_t *)&objektCPU;
+ MICHider<TYP> ret;
+
+ #pragma offload target(mic) in(uk:length(sizeof(TYP))) out(ret)
+ {
+ ret.pointer=(TYP*)malloc(sizeof(TYP));
+ std::memcpy((void*)ret.pointer,(void*)uk,sizeof(TYP));
+ }
+ return ret.pointer;
+
+ std::cout << "Někdo mně volá :-D" <<std::endl;
+ };
+
+ //old cleaning function -- deprecated
+ template <typename TYP>
+ static
+ void freeFromDevice(TYP *objektMIC)
+ {
+ MICHider<TYP> ptr;
+ ptr.pointer=objektMIC;
+ #pragma offload target(mic) in(ptr)
+ {
+ free((void*)ptr.pointer);
+ }
+ };
+
+ static inline
+ void CopyToMIC(void* mic_ptr,void* ptr,size_t size)
+ {
+ uint8_t image[size];
+ std::memcpy((void*)&image,ptr,size);
+ Devices::MICHider<void> hide_ptr;
+ hide_ptr.pointer=mic_ptr;
+ #pragma offload target(mic) in(hide_ptr) in(image) in(size)
+ {
+ std::memcpy((void*)hide_ptr.pointer,(void*)&image,size);
+ }
+ };
+
+ static inline
+ void* AllocMIC(size_t size)
+ {
+ Devices::MICHider<void> hide_ptr;
+ #pragma offload target(mic) out(hide_ptr) in(size)
+ {
+ hide_ptr.pointer=malloc(size);
+ }
+ return hide_ptr.pointer;
+ };
+
+ static inline
+ void FreeMIC(void* ptr)
+ {
+ Devices::MICHider<void> hide_ptr;
+ hide_ptr.pointer=ptr;
+ #pragma offload target(mic) in(hide_ptr)
+ {
+ free(hide_ptr.pointer);
+ }
+ };
+
+
+#endif
+
+ static void insertSmartPointer( SmartPointer* pointer );
+
+ static void removeSmartPointer( SmartPointer* pointer );
+
+ // Negative deviceId means that CudaDeviceInfo::getActiveDevice will be
+ // called to get the device ID.
+ static bool synchronizeDevice( int deviceId = -1 );
+
+ static Timer smartPointersSynchronizationTimer;
+
+ protected:
+
+ static SmartPointersRegister smartPointersRegister;
+
+};
+
+}//namespace Devices
+
+}//namespace TNL
+
+
diff --git a/src/TNL/Exceptions/CMakeLists.txt b/src/TNL/Exceptions/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..856f8c2fc73fd86f8e1530392bcb158dbaca5568
--- /dev/null
+++ b/src/TNL/Exceptions/CMakeLists.txt
@@ -0,0 +1,7 @@
+SET( headers CudaBadAlloc.h
+ CudaRuntimeError.h
+ CudaSupportMissing.h
+ MICBadAlloc.h
+ MICSupportMissing.h )
+
+INSTALL( FILES ${headers} DESTINATION include/tnl-${tnlVersion}/TNL/Exceptions )
diff --git a/src/TNL/Exceptions/CudaBadAlloc.h b/src/TNL/Exceptions/CudaBadAlloc.h
new file mode 100644
index 0000000000000000000000000000000000000000..4dc78c488e65fe6e57fcfb8d5672d2554f2262b9
--- /dev/null
+++ b/src/TNL/Exceptions/CudaBadAlloc.h
@@ -0,0 +1,40 @@
+/***************************************************************************
+ CudaBadAlloc.h - description
+ -------------------
+ begin : Jun 18, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Jakub Klinkovsky
+
+#pragma once
+
+#include <new>
+
+namespace TNL {
+namespace Exceptions {
+
+struct CudaBadAlloc
+ : public std::bad_alloc
+{
+ CudaBadAlloc()
+ {
+#ifdef HAVE_CUDA
+ // Make sure to clear the CUDA error, otherwise the exception handler
+ // might throw another exception with the same error.
+ cudaGetLastError();
+#endif
+ }
+
+ const char* what() const throw()
+ {
+ return "Failed to allocate memory on the CUDA device: "
+ "most likely there is not enough space on the device memory.";
+ }
+};
+
+} // namespace Exceptions
+} // namespace TNL
diff --git a/src/TNL/Exceptions/CudaRuntimeError.h b/src/TNL/Exceptions/CudaRuntimeError.h
new file mode 100644
index 0000000000000000000000000000000000000000..a6773c23c03ac0ab97ef5e9b6ecc8884ac324255
--- /dev/null
+++ b/src/TNL/Exceptions/CudaRuntimeError.h
@@ -0,0 +1,77 @@
+/***************************************************************************
+ CudaRuntimeError.h - description
+ -------------------
+ begin : Jun 18, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Jakub Klinkovsky
+
+#pragma once
+
+#include "CudaSupportMissing.h"
+
+namespace TNL {
+namespace Exceptions {
+
+#ifdef HAVE_CUDA
+ using CudaStatusType = cudaError;
+#else
+ using CudaStatusType = int;
+#endif
+
+class CudaRuntimeError
+ : public std::runtime_error
+{
+public:
+ CudaRuntimeError( CudaStatusType error_code )
+ : std::runtime_error( "CUDA ERROR " + std::to_string( (int) error_code ) + " (" + name( error_code ) + "): "
+ + description( error_code ) + "." ),
+ code_( error_code )
+ {}
+
+ CudaRuntimeError( CudaStatusType error_code, const std::string& what_arg )
+ : std::runtime_error( "CUDA ERROR " + std::to_string( (int) error_code ) + " (" + name( error_code ) + "): "
+ + description( error_code ) + ".\nDetails: " + what_arg ),
+ code_(error_code)
+ {}
+
+ CudaRuntimeError( CudaStatusType error_code, const char* file_name, int line )
+ : std::runtime_error( "CUDA ERROR " + std::to_string( (int) error_code ) + " (" + name( error_code ) + "): "
+ + description( error_code ) + ".\nSource: line " + std::to_string( line )
+ + " in " + file_name + ": " + description( error_code ) ),
+ code_(error_code)
+ {}
+
+ CudaStatusType code() const
+ {
+ return code_;
+ }
+
+private:
+ static std::string name( CudaStatusType error_code )
+ {
+#ifdef HAVE_CUDA
+ return cudaGetErrorName( error_code );
+#else
+ throw CudaSupportMissing();
+#endif
+ }
+
+ static std::string description( CudaStatusType error_code )
+ {
+#ifdef HAVE_CUDA
+ return cudaGetErrorString( error_code );
+#else
+ throw CudaSupportMissing();
+#endif
+ }
+
+ CudaStatusType code_;
+};
+
+} // namespace Exceptions
+} // namespace TNL
diff --git a/src/TNL/Exceptions/CudaSupportMissing.h b/src/TNL/Exceptions/CudaSupportMissing.h
new file mode 100644
index 0000000000000000000000000000000000000000..7a7f978a122a0c0f53f82a06d3dc03bbde40fd36
--- /dev/null
+++ b/src/TNL/Exceptions/CudaSupportMissing.h
@@ -0,0 +1,30 @@
+/***************************************************************************
+ CudaSupportMissing.h - description
+ -------------------
+ begin : Jun 18, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Jakub Klinkovsky
+
+#pragma once
+
+#include <stdexcept>
+
+namespace TNL {
+namespace Exceptions {
+
+struct CudaSupportMissing
+ : public std::runtime_error
+{
+ CudaSupportMissing()
+ : std::runtime_error( "CUDA support is missing, but the program called a function which needs it. "
+ "Please recompile the program with CUDA support." )
+ {}
+};
+
+} // namespace Exceptions
+} // namespace TNL
diff --git a/src/TNL/Exceptions/MICBadAlloc.h b/src/TNL/Exceptions/MICBadAlloc.h
new file mode 100644
index 0000000000000000000000000000000000000000..b8f3a9157c54d8155652a42a700ad71a221aa201
--- /dev/null
+++ b/src/TNL/Exceptions/MICBadAlloc.h
@@ -0,0 +1,31 @@
+/***************************************************************************
+ MICBadAlloc.h - description
+ -------------------
+ begin : Jul 31, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Jakub Klinkovsky
+
+#pragma once
+
+#include <new>
+
+namespace TNL {
+namespace Exceptions {
+
+struct MICBadAlloc
+ : public std::bad_alloc
+{
+ const char* what() const throw()
+ {
+ return "Failed to allocate memory on the MIC device: "
+ "most likely there is not enough space on the device memory.";
+ }
+};
+
+} // namespace Exceptions
+} // namespace TNL
diff --git a/src/TNL/Exceptions/MICSupportMissing.h b/src/TNL/Exceptions/MICSupportMissing.h
new file mode 100644
index 0000000000000000000000000000000000000000..6d4260e6addbbb9dd89a7c9d5a07833485c6a0c2
--- /dev/null
+++ b/src/TNL/Exceptions/MICSupportMissing.h
@@ -0,0 +1,30 @@
+/***************************************************************************
+ MICSupportMissing.h - description
+ -------------------
+ begin : Jul 31, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Jakub Klinkovsky
+
+#pragma once
+
+#include <stdexcept>
+
+namespace TNL {
+namespace Exceptions {
+
+struct MICSupportMissing
+ : public std::runtime_error
+{
+ MICSupportMissing()
+ : std::runtime_error( "MIC support is missing, but the program called a function which needs it. "
+ "Please recompile the program with MIC support." )
+ {}
+};
+
+} // namespace Exceptions
+} // namespace TNL
diff --git a/src/TNL/Experimental/Arithmetics/CMakeLists.txt b/src/TNL/Experimental/Arithmetics/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Arithmetics/Real.h b/src/TNL/Experimental/Arithmetics/Real.h
index 96d8369ef62d3211cc8182833569c85e7e7b6a44..7f0943147705757f3eb0c745b6e0025dbd763614 100644
--- a/src/TNL/Experimental/Arithmetics/Real.h
+++ b/src/TNL/Experimental/Arithmetics/Real.h
@@ -11,7 +11,7 @@
#pragma once
#include <iostream>
-#include <math.h>
+#include <cmath>
#include <TNL/Experimental/Arithmetics/FlopsCounter.h>
namespace TNL {
@@ -313,105 +313,105 @@ template< class T > const Real< T > fabs( const Real< T >& v )
template< class T > const Real< T > sqrt( const Real< T >& v )
{
tnl_flops_counter. recordFunction();
- return Real< T >( ::sqrt( v. Data() ) );
+ return Real< T >( std::sqrt( v. Data() ) );
};
template< class T > const Real< T > pow( const Real< T >& x, const Real< T >& exp )
{
tnl_flops_counter. recordFunction();
- return Real< T >( ::pow( x. Data(), exp. Data() ) );
+ return Real< T >( std::pow( x. Data(), exp. Data() ) );
};
template< class T > const Real< T > pow( const Real< T >& x, const T& exp )
{
tnl_flops_counter. recordFunction();
- return Real< T >( ::pow( x. Data(), exp ) );
+ return Real< T >( std::pow( x. Data(), exp ) );
};
template< class T > const Real< T > cos( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( ::cos( x. Data() ) );
+ return Real< T >( std::cos( x. Data() ) );
};
template< class T > const Real< T > sin( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( ::sin( x. Data() ) );
+ return Real< T >( std::sin( x. Data() ) );
};
template< class T > const Real< T > tan( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( tan( x. Data() ) );
+ return Real< T >( std::tan( x. Data() ) );
};
template< class T > const Real< T > acos( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( acos( x. Data() ) );
+ return Real< T >( std::acos( x. Data() ) );
};
template< class T > const Real< T > asin( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( asin( x. Data() ) );
+ return Real< T >( std::asin( x. Data() ) );
};
template< class T > const Real< T > atan( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( atan( x. Data() ) );
+ return Real< T >( std::atan( x. Data() ) );
};
template< class T > const Real< T > atan2( const Real< T >& x, const Real< T >& exp )
{
tnl_flops_counter. recordFunction();
- return Real< T >( atan2( x. Data(), exp. Data() ) );
+ return Real< T >( std::atan2( x. Data(), exp. Data() ) );
};
template< class T > const Real< T > atan2( const Real< T >& x, const T& exp )
{
tnl_flops_counter. recordFunction();
- return Real< T >( atan2( x. Data(), exp ) );
+ return Real< T >( std::atan2( x. Data(), exp ) );
};
template< class T > const Real< T > cosh( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( cosh( x. Data() ) );
+ return Real< T >( std::cosh( x. Data() ) );
};
template< class T > const Real< T > sinh( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( sinh( x. Data() ) );
+ return Real< T >( std::sinh( x. Data() ) );
};
template< class T > const Real< T > tanh( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( ::tanh( x. Data() ) );
+ return Real< T >( std::tanh( x. Data() ) );
};
template< class T > const Real< T > exp( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( exp( x. Data() ) );
+ return Real< T >( std::exp( x. Data() ) );
};
template< class T > const Real< T > log( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( log( x. Data() ) );
+ return Real< T >( std::log( x. Data() ) );
};
template< class T > const Real< T > log10( const Real< T >& x )
{
tnl_flops_counter. recordFunction();
- return Real< T >( log10( x. Data() ) );
+ return Real< T >( std::log10( x. Data() ) );
};
template< class T >
diff --git a/src/TNL/Experimental/CMakeLists.txt b/src/TNL/Experimental/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Operators/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Eikonal/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Eikonal/CMakeLists.txt
index 202550179b61cbf4a47bca43f96f751d0e2510eb..115fd1c2cde483c4fbfe64951210d69827976d2b 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Eikonal/CMakeLists.txt
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Eikonal/CMakeLists.txt
@@ -15,4 +15,4 @@ set( tnl_implementation_operators_godunov_SOURCES
${common_SOURCES}
PARENT_SCOPE )
-INSTALL( FILES ${headers} DESTINATION include/tnl-${tnlVersion}/implementation/operators/godunov )
+INSTALL( FILES ${headers} DESTINATION include/tnl-${tnlVersion}/TNL/Operators/Godunov/ )
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Eikonal/godunovEikonal.h b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Eikonal/godunovEikonal.h
index a90c7258438a52c0afebf2d0e9838e1491f16a13..9cebf0ea9a2608526cff1dcf1290e41fa4f7b083 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Eikonal/godunovEikonal.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Eikonal/godunovEikonal.h
@@ -37,7 +37,7 @@ class godunovEikonalScheme< tnlGrid< 1,MeshReal, Device, MeshIndex >, Real, Inde
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 1, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
static String getType();
@@ -49,11 +49,11 @@ class godunovEikonalScheme< tnlGrid< 1,MeshReal, Device, MeshIndex >, Real, Inde
const RealType& f ) const
{
const RealType& hx_inv = entity.getMesh().template getSpaceStepsProducts< -1 >();
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const RealType& u_c = u[ entity.getIndex() ];
- const RealType& u_e = u[ neighbourEntities.template getEntityIndex< 1 >() ];
- const RealType& u_w = u[ neighbourEntities.template getEntityIndex< -1 >() ];
+ const RealType& u_e = u[ neighborEntities.template getEntityIndex< 1 >() ];
+ const RealType& u_w = u[ neighborEntities.template getEntityIndex< -1 >() ];
if( f > 0.0 )
{
@@ -99,7 +99,7 @@ class godunovEikonalScheme< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Inde
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -114,12 +114,12 @@ class godunovEikonalScheme< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Inde
const RealType& hx_inv = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const RealType& hy_inv = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType& u_c = u[ entity.getIndex() ];
- const RealType& u_e = u[ neighbourEntities.template getEntityIndex< 1, 0 >() ];
- const RealType& u_w = u[ neighbourEntities.template getEntityIndex< -1, 0 >() ];
- const RealType& u_n = u[ neighbourEntities.template getEntityIndex< 0, 1 >() ];
- const RealType& u_s = u[ neighbourEntities.template getEntityIndex< 0, -1 >() ];
+ const RealType& u_e = u[ neighborEntities.template getEntityIndex< 1, 0 >() ];
+ const RealType& u_w = u[ neighborEntities.template getEntityIndex< -1, 0 >() ];
+ const RealType& u_n = u[ neighborEntities.template getEntityIndex< 0, 1 >() ];
+ const RealType& u_s = u[ neighborEntities.template getEntityIndex< 0, -1 >() ];
if( f > 0.0 )
{
@@ -180,7 +180,7 @@ class godunovEikonalScheme< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Inde
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
static String getType();
@@ -195,14 +195,14 @@ class godunovEikonalScheme< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Inde
const RealType& hy_inv = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const RealType& hz_inv = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& u_c = u[ entity.getIndex() ];
- const RealType& u_e = u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ];
- const RealType& u_w = u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ];
- const RealType& u_n = u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ];
- const RealType& u_s = u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ];
- const RealType& u_t = u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ];
- const RealType& u_b = u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ];
+ const RealType& u_e = u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ];
+ const RealType& u_w = u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ];
+ const RealType& u_n = u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ];
+ const RealType& u_s = u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ];
+ const RealType& u_t = u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ];
+ const RealType& u_b = u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ];
if( f > 0.0 )
{
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/CMakeLists.txt
old mode 100755
new mode 100644
index 8b74f755ed0a74e7001f1c155c60f61c650679d2..bb9c4e3b4fb36d9e9982f8abcead16a1e5dfdb16
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/CMakeLists.txt
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/CMakeLists.txt
@@ -21,4 +21,4 @@ set( tnl_implementation_operators_godunov-eikonal_SOURCES
${common_SOURCES}
PARENT_SCOPE )
-INSTALL( FILES ${headers} DESTINATION include/tnl-${tnlVersion}/implementation/operators/godunov-eikonal )
+INSTALL( FILES ${headers} DESTINATION include/tnl-${tnlVersion}/TNL/Operators/Godunov-Eikonal/ )
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/godunovEikonal.h b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/godunovEikonal.h
index d4941ba38a992fd76047a7ead2f538b79af2f079..fcff6b36aa023b548972117521b33b772029a838 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/godunovEikonal.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/godunovEikonal.h
@@ -20,8 +20,7 @@
#include <solvers/preconditioners/tnlDummyPreconditioner.h>
#include <solvers/tnlSolverMonitor.h>
#include <core/tnlLogger.h>
-#include <core/vectors/tnlVector.h>
-#include <core/vectors/tnlSharedVector.h>
+#include <TNL/Containers/Vector.h>
#include <core/mfilename.h>
#include <mesh/tnlGrid.h>
@@ -49,7 +48,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 1, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -98,7 +97,7 @@ class godunovEikonalScheme< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Inde
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
static String getType();
@@ -143,7 +142,7 @@ class godunovEikonalScheme< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Inde
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal.h b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal.h
index e2f1859231a8096ae38e0dcaff664f490efa5cfc..f462245848757f6ccf9b9349ac4baa38cb613e4b 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal.h
@@ -21,8 +21,7 @@
#include <solvers/preconditioners/tnlDummyPreconditioner.h>
#include <solvers/tnlSolverMonitor.h>
#include <core/tnlLogger.h>
-#include <core/vectors/tnlVector.h>
-#include <core/vectors/tnlSharedVector.h>
+#include <TNL/Containers/Vector.h>
#include <core/mfilename.h>
#include <mesh/tnlGrid.h>
#include <core/tnlCuda.h>
@@ -52,7 +51,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 1, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -121,7 +120,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
#ifdef HAVE_CUDA
__device__ __host__
@@ -150,7 +149,7 @@ public:
const Vector& u,
const RealType& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities) const;
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities) const;
#ifdef HAVE_CUDA
__device__
@@ -161,7 +160,7 @@ public:
const RealType* u,
const RealType& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities) const;
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities) const;
#ifdef HAVE_CUDA
__device__ __host__
#endif
@@ -197,7 +196,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
#ifdef HAVE_CUDA
@@ -227,7 +226,7 @@ public:
const Vector& u,
const RealType& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities ) const;
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities ) const;
#ifdef HAVE_CUDA
__device__
@@ -238,7 +237,7 @@ public:
const RealType* u,
const RealType& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities ) const;
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities ) const;
#ifdef HAVE_CUDA
__device__ __host__
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal2D_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal2D_impl.h
index 29738ab330e3b951d6b8a9bd6be89b5606c0f5b0..5ba9e3f781b5078f71acee19cd754364bca32b50 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal2D_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal2D_impl.h
@@ -145,7 +145,7 @@ Real parallelGodunovEikonalScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Re
const Vector& u,
const Real& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities ) const
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities ) const
{
@@ -182,24 +182,24 @@ Real parallelGodunovEikonalScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Re
if(coordinates.x() == mesh.getDimensions().x() - 1)
- xf += u[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< -1, 0 >()];
else
- xf += u[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< 1, 0 >()];
if(coordinates.x() == 0)
- xb -= u[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< 1, 0 >()];
else
- xb -= u[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< -1, 0 >()];
if(coordinates.y() == mesh.getDimensions().y() - 1)
- yf += u[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, -1 >()];
else
- yf += u[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, 1 >()];
if(coordinates.y() == 0)
- yb -= u[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, 1 >()];
else
- yb -= u[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, -1 >()];
//xb *= ihx;
@@ -269,9 +269,9 @@ Real parallelGodunovEikonalScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Re
// else *//*if(boundaryCondition & 4)
// xf = 0.0;
// else /**/if(coordinates.x() == mesh.getDimensions().x() - 1)
-// xf = negativePart((u[neighbourEntities.template getEntityIndex< -1, 0 >()] - u[cellIndex])*ihx);
+// xf = negativePart((u[neighborEntities.template getEntityIndex< -1, 0 >()] - u[cellIndex])*ihx);
// else
-// xf = negativePart((u[neighbourEntities.template getEntityIndex< 1, 0 >()] - u[cellIndex])*ihx);
+// xf = negativePart((u[neighborEntities.template getEntityIndex< 1, 0 >()] - u[cellIndex])*ihx);
//
// /**/ /* if(boundaryCondition & 4)
// xb = (u[cellIndex] - u[mesh.getCellXPredecessor( cellIndex )])*ihx;
@@ -280,25 +280,25 @@ Real parallelGodunovEikonalScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Re
// else /**/if(coordinates.x() == 0)
// xb = positivePart((u[cellIndex] - u[mesh.template getCellNextToCell<+1,0>( cellIndex )])*ihx);
// else
-// xb = positivePart((u[cellIndex] - u[neighbourEntities.template getEntityIndex< -1, 0 >()])*ihx);
+// xb = positivePart((u[cellIndex] - u[neighborEntities.template getEntityIndex< -1, 0 >()])*ihx);
//
// /**/ /* if(boundaryCondition & 1)
// yf = (u[mesh.getCellYSuccessor( cellIndex )] - u[cellIndex])*ihy;
// else *//*if(boundaryCondition & 8)
// yf = 0.0;
// else /**/if(coordinates.y() == mesh.getDimensions().y() - 1)
-// yf = negativePart((u[neighbourEntities.template getEntityIndex< 0, -1 >()] - u[cellIndex])*ihy);
+// yf = negativePart((u[neighborEntities.template getEntityIndex< 0, -1 >()] - u[cellIndex])*ihy);
// else
-// yf = negativePart((u[neighbourEntities.template getEntityIndex< 0, 1 >()] - u[cellIndex])*ihy);
+// yf = negativePart((u[neighborEntities.template getEntityIndex< 0, 1 >()] - u[cellIndex])*ihy);
//
// /**/ /* if(boundaryCondition & 8)
// yb = (u[cellIndex] - u[mesh.getCellYPredecessor( cellIndex )])*ihy;
// else *//*if(boundaryCondition & 1)
// yb = 0.0;
// else /**/if(coordinates.y() == 0)
-// yb = positivePart((u[cellIndex] - u[neighbourEntities.template getEntityIndex< 0, 1 >()])*ihy);
+// yb = positivePart((u[cellIndex] - u[neighborEntities.template getEntityIndex< 0, 1 >()])*ihy);
// else
-// yb = positivePart((u[cellIndex] - u[neighbourEntities.template getEntityIndex< 0, -1 >()])*ihy);
+// yb = positivePart((u[cellIndex] - u[neighborEntities.template getEntityIndex< 0, -1 >()])*ihy);
//
// if(xb - xf > 0.0)
// xf = 0.0;
@@ -323,36 +323,36 @@ Real parallelGodunovEikonalScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Re
// else*//* if(boundaryCondition & 4)
// xf = 0.0;
// else /**/if(coordinates.x() == mesh.getDimensions().x() - 1)
-// xf = positivePart((u[neighbourEntities.template getEntityIndex< -1, 0 >()] - u[cellIndex])*ihx);
+// xf = positivePart((u[neighborEntities.template getEntityIndex< -1, 0 >()] - u[cellIndex])*ihx);
// else
-// xf = positivePart((u[neighbourEntities.template getEntityIndex< 1, 0 >()] - u[cellIndex])*ihx);
+// xf = positivePart((u[neighborEntities.template getEntityIndex< 1, 0 >()] - u[cellIndex])*ihx);
//
// /**/ /* if(boundaryCondition & 4)
// xb = (u[cellIndex] - u[mesh.getCellXPredecessor( cellIndex )])*ihx;
// else*//* if(boundaryCondition & 2)
// xb = 0.0;
// else /**/if(coordinates.x() == 0)
-// xb = negativePart((u[cellIndex] - u[neighbourEntities.template getEntityIndex< 1, 0 >()])*ihx);
+// xb = negativePart((u[cellIndex] - u[neighborEntities.template getEntityIndex< 1, 0 >()])*ihx);
// else
-// xb = negativePart((u[cellIndex] - u[neighbourEntities.template getEntityIndex< -1, 0 >()])*ihx);
+// xb = negativePart((u[cellIndex] - u[neighborEntities.template getEntityIndex< -1, 0 >()])*ihx);
//
// /**/ /* if(boundaryCondition & 1)
// yf = (u[mesh.getCellYSuccessor( cellIndex )] - u[cellIndex])*ihy;
// else *//*if(boundaryCondition & 8)
// yf = 0.0;
// else /**/if(coordinates.y() == mesh.getDimensions().y() - 1)
-// yf = positivePart((u[neighbourEntities.template getEntityIndex< 0, -1 >()] - u[cellIndex])*ihy);
+// yf = positivePart((u[neighborEntities.template getEntityIndex< 0, -1 >()] - u[cellIndex])*ihy);
// else
-// yf = positivePart((u[neighbourEntities.template getEntityIndex< 0, 1 >()] - u[cellIndex])*ihy);
+// yf = positivePart((u[neighborEntities.template getEntityIndex< 0, 1 >()] - u[cellIndex])*ihy);
//
// /**/ /* if(boundaryCondition & 8)
// yb = (u[cellIndex] - u[mesh.getCellYPredecessor( cellIndex )])*ihy;
// else*//* if(boundaryCondition & 1)
// yb = 0.0;
// else /**/if(coordinates.y() == 0)
-// yb = negativePart((u[cellIndex] - u[neighbourEntities.template getEntityIndex< 0, 1 >()])*ihy);
+// yb = negativePart((u[cellIndex] - u[neighborEntities.template getEntityIndex< 0, 1 >()])*ihy);
// else
-// yb = negativePart((u[cellIndex] - u[neighbourEntities.template getEntityIndex< 0, -1 >()])*ihy);
+// yb = negativePart((u[cellIndex] - u[neighborEntities.template getEntityIndex< 0, -1 >()])*ihy);
//
//
// if(xb - xf > 0.0)
@@ -401,7 +401,7 @@ Real parallelGodunovEikonalScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Re
const Real* u,
const Real& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities) const
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities) const
{
RealType signui;
@@ -418,24 +418,24 @@ Real parallelGodunovEikonalScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Re
if(coordinates.x() == mesh.getDimensions().x() - 1)
- xf += u[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< -1, 0 >()];
else
- xf += u[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< 1, 0 >()];
if(coordinates.x() == 0)
- xb -= u[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< 1, 0 >()];
else
- xb -= u[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< -1, 0 >()];
if(coordinates.y() == mesh.getDimensions().y() - 1)
- yf += u[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, -1 >()];
else
- yf += u[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, 1 >()];
if(coordinates.y() == 0)
- yb -= u[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, 1 >()];
else
- yb -= u[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, -1 >()];
if(signui > 0.0)
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal3D_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal3D_impl.h
index 3aa8aeed5357e7a4f0b0f176e9002c89a4295a84..a352a07375a6457da3cff77a7f89dde88c94e8f7 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal3D_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovEikonal3D_impl.h
@@ -127,7 +127,7 @@ Real parallelGodunovEikonalScheme< tnlGrid< 3, MeshReal, Device, MeshIndex >, Re
const Vector& u,
const Real& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities ) const
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities ) const
{
if ( ((coordinates.x() == 0 && (boundaryCondition & 4)) or
(coordinates.x() == mesh.getDimensions().x() - 1 && (boundaryCondition & 2)) or
@@ -158,35 +158,35 @@ Real parallelGodunovEikonalScheme< tnlGrid< 3, MeshReal, Device, MeshIndex >, Re
if(coordinates.x() == mesh.getDimensions().x() - 1)
- xf += u[neighbourEntities.template getEntityIndex< -1, 0, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< -1, 0, 0 >()];
else
- xf += u[neighbourEntities.template getEntityIndex< 1, 0, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< 1, 0, 0 >()];
if(coordinates.x() == 0)
- xb -= u[neighbourEntities.template getEntityIndex< 1, 0, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< 1, 0, 0 >()];
else
- xb -= u[neighbourEntities.template getEntityIndex< -1, 0, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< -1, 0, 0 >()];
if(coordinates.y() == mesh.getDimensions().y() - 1)
- yf += u[neighbourEntities.template getEntityIndex< 0, -1, 0 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, -1, 0 >()];
else
- yf += u[neighbourEntities.template getEntityIndex< 0, 1, 0 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, 1, 0 >()];
if(coordinates.y() == 0)
- yb -= u[neighbourEntities.template getEntityIndex< 0, 1, 0 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, 1, 0 >()];
else
- yb -= u[neighbourEntities.template getEntityIndex< 0, -1, 0 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, -1, 0 >()];
if(coordinates.z() == mesh.getDimensions().z() - 1)
- zf += u[neighbourEntities.template getEntityIndex< 0, 0, -1 >()];
+ zf += u[neighborEntities.template getEntityIndex< 0, 0, -1 >()];
else
- zf += u[neighbourEntities.template getEntityIndex< 0, 0, 1 >()];
+ zf += u[neighborEntities.template getEntityIndex< 0, 0, 1 >()];
if(coordinates.z() == 0)
- zb -= u[neighbourEntities.template getEntityIndex< 0, 0, 1 >()];
+ zb -= u[neighborEntities.template getEntityIndex< 0, 0, 1 >()];
else
- zb -= u[neighbourEntities.template getEntityIndex< 0, 0, -1 >()];
+ zb -= u[neighborEntities.template getEntityIndex< 0, 0, -1 >()];
if(signui > 0.0)
{
@@ -262,7 +262,7 @@ Real parallelGodunovEikonalScheme< tnlGrid< 3, MeshReal, Device, MeshIndex >, Re
const Real* u,
const Real& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities
) const
{
RealType signui;
@@ -279,35 +279,35 @@ Real parallelGodunovEikonalScheme< tnlGrid< 3, MeshReal, Device, MeshIndex >, Re
if(coordinates.x() == mesh.getDimensions().x() - 1)
- xf += u[neighbourEntities.template getEntityIndex< -1, 0, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< -1, 0, 0 >()];
else
- xf += u[neighbourEntities.template getEntityIndex< 1, 0, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< 1, 0, 0 >()];
if(coordinates.x() == 0)
- xb -= u[neighbourEntities.template getEntityIndex< 1, 0, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< 1, 0, 0 >()];
else
- xb -= u[neighbourEntities.template getEntityIndex< -1, 0, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< -1, 0, 0 >()];
if(coordinates.y() == mesh.getDimensions().y() - 1)
- yf += u[neighbourEntities.template getEntityIndex< 0, -1, 0 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, -1, 0 >()];
else
- yf += u[neighbourEntities.template getEntityIndex< 0, 1, 0 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, 1, 0 >()];
if(coordinates.y() == 0)
- yb -= u[neighbourEntities.template getEntityIndex< 0, 1, 0 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, 1, 0 >()];
else
- yb -= u[neighbourEntities.template getEntityIndex< 0, -1, 0 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, -1, 0 >()];
if(coordinates.z() == mesh.getDimensions().z() - 1)
- zf += u[neighbourEntities.template getEntityIndex< 0, 0, -1 >()];
+ zf += u[neighborEntities.template getEntityIndex< 0, 0, -1 >()];
else
- zf += u[neighbourEntities.template getEntityIndex< 0, 0, 1 >()];
+ zf += u[neighborEntities.template getEntityIndex< 0, 0, 1 >()];
if(coordinates.z() == 0)
- zb -= u[neighbourEntities.template getEntityIndex< 0, 0, 1 >()];
+ zb -= u[neighborEntities.template getEntityIndex< 0, 0, 1 >()];
else
- zb -= u[neighbourEntities.template getEntityIndex< 0, 0, -1 >()];
+ zb -= u[neighborEntities.template getEntityIndex< 0, 0, -1 >()];
if(signui > 0.0)
{
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovMap.h b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovMap.h
index db7a087224b933150fcda1ed47119d2e23448000..deeff7c08850c41319ea2bfbfd55fa8d83cb9a9d 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovMap.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovMap.h
@@ -21,8 +21,7 @@
#include <solvers/preconditioners/tnlDummyPreconditioner.h>
#include <solvers/tnlSolverMonitor.h>
#include <core/tnlLogger.h>
-#include <core/vectors/tnlVector.h>
-#include <core/vectors/tnlSharedVector.h>
+#include <TNL/Containers/Vector.h>
#include <core/mfilename.h>
#include <mesh/tnlGrid.h>
#include <core/tnlCuda.h>
@@ -52,7 +51,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 1, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -121,7 +120,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
#ifdef HAVE_CUDA
__device__ __host__
@@ -150,7 +149,7 @@ public:
const Vector& u,
const RealType& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities,
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities,
const Vector& map) const;
#ifdef HAVE_CUDA
@@ -162,7 +161,7 @@ public:
const RealType* u,
const RealType& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities,
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities,
const RealType* map) const;
#ifdef HAVE_CUDA
__device__ __host__
@@ -199,7 +198,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
#ifdef HAVE_CUDA
@@ -229,7 +228,7 @@ public:
const Vector& u,
const RealType& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities ) const;
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities ) const;
#ifdef HAVE_CUDA
__device__
@@ -240,7 +239,7 @@ public:
const RealType* u,
const RealType& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities) const;
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities) const;
#ifdef HAVE_CUDA
__device__ __host__
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovMap2D_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovMap2D_impl.h
index c5f7d613497e7f61b449a2e16190c18988a12c3f..9b3117cf25a8d039002db1c6ba7328ea0870d492 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovMap2D_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov-Eikonal/parallelGodunovMap2D_impl.h
@@ -145,7 +145,7 @@ Real parallelGodunovMapScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Real,
const Vector& u,
const Real& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities,
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities,
const Vector& map) const
{
@@ -190,24 +190,24 @@ Real parallelGodunovMapScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Real,
if(coordinates.x() == mesh.getDimensions().x() - 1)
- xf += u[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< -1, 0 >()];
else
- xf += u[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< 1, 0 >()];
if(coordinates.x() == 0)
- xb -= u[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< 1, 0 >()];
else
- xb -= u[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< -1, 0 >()];
if(coordinates.y() == mesh.getDimensions().y() - 1)
- yf += u[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, -1 >()];
else
- yf += u[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, 1 >()];
if(coordinates.y() == 0)
- yb -= u[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, 1 >()];
else
- yb -= u[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, -1 >()];
if(signui > 0.0)
@@ -286,7 +286,7 @@ Real parallelGodunovMapScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Real,
const Real* u,
const Real& time,
const IndexType boundaryCondition,
- const tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities,
+ const tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities,
const Real* map) const
{
// int gid = (blockDim.y*blockIdx.y + threadIdx.y)*blockDim.x*gridDim.x + blockDim.x*blockIdx.x + threadIdx.x;
@@ -316,24 +316,24 @@ Real parallelGodunovMapScheme< tnlGrid< 2, MeshReal, Device, MeshIndex >, Real,
if(coordinates.x() == mesh.getDimensions().x() - 1)
- xf += u[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< -1, 0 >()];
else
- xf += u[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ xf += u[neighborEntities.template getEntityIndex< 1, 0 >()];
if(coordinates.x() == 0)
- xb -= u[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< 1, 0 >()];
else
- xb -= u[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ xb -= u[neighborEntities.template getEntityIndex< -1, 0 >()];
if(coordinates.y() == mesh.getDimensions().y() - 1)
- yf += u[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, -1 >()];
else
- yf += u[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ yf += u[neighborEntities.template getEntityIndex< 0, 1 >()];
if(coordinates.y() == 0)
- yb -= u[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, 1 >()];
else
- yb -= u[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ yb -= u[neighborEntities.template getEntityIndex< 0, -1 >()];
if(signui > 0.0)
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov/CMakeLists.txt
old mode 100755
new mode 100644
index 74f7510d72329f60305bb366604e72b7201a3ea0..98b3eefe79e290a43e25867cf4e2c7cad9a1a419
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov/CMakeLists.txt
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/Godunov/CMakeLists.txt
@@ -14,4 +14,4 @@ set( tnl_implementation_operators_godunov_SOURCES
${common_SOURCES}
PARENT_SCOPE )
-INSTALL( FILES ${headers} DESTINATION include/tnl-${tnlVersion}/implementation/operators/godunov )
+INSTALL( FILES ${headers} DESTINATION include/tnl-${tnlVersion}/TNL/Operators/Godunov/ )
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/upwindEikonal.h b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/upwindEikonal.h
index e8f793260999eba3f7e2082a5237ee3daba6e86a..7467f66715e6dcc5fcf9a04ac0ecd66d1f026a11 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/upwindEikonal.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Operators/Hamilton-Jacobi/upwindEikonal.h
@@ -41,7 +41,7 @@ class upwindEikonalScheme< tnlGrid< 1,MeshReal, Device, MeshIndex >, Real, Index
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 1, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -54,11 +54,11 @@ class upwindEikonalScheme< tnlGrid< 1,MeshReal, Device, MeshIndex >, Real, Index
const RealType& f ) const
{
const RealType& hx_inv = entity.getMesh().template getSpaceStepsProducts< -1 >();
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const RealType& u_c = u[ entity.getIndex() ];
- const RealType& u_e = u[ neighbourEntities.template getEntityIndex< 1 >() ];
- const RealType& u_w = u[ neighbourEntities.template getEntityIndex< -1 >() ];
+ const RealType& u_e = u[ neighborEntities.template getEntityIndex< 1 >() ];
+ const RealType& u_w = u[ neighborEntities.template getEntityIndex< -1 >() ];
if( f > 0.0 )
{
@@ -94,7 +94,7 @@ class upwindEikonalScheme< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
static String getType();
@@ -108,12 +108,12 @@ class upwindEikonalScheme< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index
const RealType& hx_inv = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const RealType& hy_inv = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType& u_c = u[ entity.getIndex() ];
- const RealType& u_e = u[ neighbourEntities.template getEntityIndex< 1, 0 >() ];
- const RealType& u_w = u[ neighbourEntities.template getEntityIndex< -1, 0 >() ];
- const RealType& u_n = u[ neighbourEntities.template getEntityIndex< 0, 1 >() ];
- const RealType& u_s = u[ neighbourEntities.template getEntityIndex< 0, -1 >() ];
+ const RealType& u_e = u[ neighborEntities.template getEntityIndex< 1, 0 >() ];
+ const RealType& u_w = u[ neighborEntities.template getEntityIndex< -1, 0 >() ];
+ const RealType& u_n = u[ neighborEntities.template getEntityIndex< 0, 1 >() ];
+ const RealType& u_s = u[ neighborEntities.template getEntityIndex< 0, -1 >() ];
if( f > 0.0 )
{
const RealType xf = negativePart( ( u_e - u_c ) * hx_inv );
@@ -151,7 +151,7 @@ class upwindEikonalScheme< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -167,14 +167,14 @@ class upwindEikonalScheme< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index
const RealType& hy_inv = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const RealType& hz_inv = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& u_c = u[ entity.getIndex() ];
- const RealType& u_e = u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ];
- const RealType& u_w = u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ];
- const RealType& u_n = u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ];
- const RealType& u_s = u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ];
- const RealType& u_t = u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ];
- const RealType& u_b = u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ];
+ const RealType& u_e = u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ];
+ const RealType& u_w = u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ];
+ const RealType& u_n = u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ];
+ const RealType& u_s = u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ];
+ const RealType& u_t = u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ];
+ const RealType& u_b = u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ];
if( f > 0.0 )
{
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/main.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/main.h
index d17b47ac5125bebf9ac9619a9496b5dc4b46b394..813e99c72cd6de9b597c10a943ab666220453018 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/main.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/main.h
@@ -47,26 +47,26 @@ int main( int argc, char* argv[] )
if(dim == 2)
{
- tnlFastSweepingMap<tnlGrid<2,double,tnlHost, int>, double, int> solver;
+ tnlFastSweepingMap<tnlGrid<2,double,TNL::Devices::Host, int>, double, int> solver;
if(!solver.init(parameters))
{
cerr << "Solver failed to initialize." << endl;
return EXIT_FAILURE;
}
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cout << "-------------------------------------------------------------" << endl;
cout << "Starting solver..." << endl;
solver.run();
}
// else if(dim == 3)
// {
-// tnlFastSweepingMap<tnlGrid<3,double,tnlHost, int>, double, int> solver;
+// tnlFastSweepingMap<tnlGrid<3,double,TNL::Devices::Host, int>, double, int> solver;
// if(!solver.init(parameters))
// {
// cerr << "Solver failed to initialize." << endl;
// return EXIT_FAILURE;
// }
-// checkCudaDevice;
+// TNL_CHECK_CUDA_DEVICE;
// cout << "-------------------------------------------------------------" << endl;
// cout << "Starting solver..." << endl;
// solver.run();
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap.h
index f6f2150a4df3ae6f0fe85efeeb6ea9abac0007d3..c568329ba2aa5fdb8fed303d43b25e73f210c014 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap.h
@@ -17,10 +17,10 @@
#define TNLFASTSWEEPING_H_
#include <TNL/Config/ParameterContainer.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
#include <TNL/Containers/StaticVector.h>
#include <functions/tnlMeshFunction.h>
-#include <core/tnlHost.h>
+#include <TNL/Devices/Host.h>
#include <mesh/tnlGrid.h>
#include <mesh/grids/tnlGridEntity.h>
#include <limits.h>
@@ -55,7 +55,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -137,7 +137,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
tnlFastSweepingMap();
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap2D_CUDA_v4_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap2D_CUDA_v4_impl.h
index 5e9d11b5264e301d8eb9a30b625861bf6a79ad2b..b865bef001413f10d2862e0c16675e5c33309470 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap2D_CUDA_v4_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap2D_CUDA_v4_impl.h
@@ -135,7 +135,7 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
initCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
return true;
}
@@ -164,21 +164,21 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
cudaMemcpy(this->changed, &zero, sizeof(int), cudaMemcpyHostToDevice);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,0,0, this->changed);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cudaMemcpy(&run, this->changed,sizeof(int), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cntr++;
cout << "Finished set of sweeps #" << cntr << " " << run << endl;
}
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//data.setLike(dofVector.getData());
//cudaMemcpy(data.getData(), cudaDofVector2, this->dofVector.getData().getSize()*sizeof(double), cudaMemcpyDeviceToHost);
@@ -207,35 +207,35 @@ template< typename MeshReal,
__device__
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: updateValue( Index i, Index j, Index* something_changed)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
if(map_cuda[Entity.getIndex()] != 0.0)
{
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real value = cudaDofVector2[Entity.getIndex()];
Real im = abs(1.0/map_cuda[Entity.getIndex()]);
Real a,b, tmp;
if( i == 0 )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()];
else if( i == Mesh.getDimensions().x() - 1 )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
else
{
- a = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] );
+ a = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] );
}
if( j == 0 )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()];
else if( j == Mesh.getDimensions().y() - 1 )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
else
{
- b = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] );
+ b = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] );
}
@@ -269,11 +269,11 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
int i = threadIdx.x + blockDim.x*blockIdx.x;
int j = blockDim.y*blockIdx.y + threadIdx.y;
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
int gid = Entity.getIndex();
@@ -294,18 +294,18 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
// {
// if(cudaDofVector[Entity.getIndex()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1111(i,j);
// else
// setupSquare1110(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1101(i,j);
// else
// setupSquare1100(i,j);
@@ -313,16 +313,16 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1011(i,j);
// else
// setupSquare1010(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1001(i,j);
// else
// setupSquare1000(i,j);
@@ -331,18 +331,18 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0111(i,j);
// else
// setupSquare0110(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0101(i,j);
// else
// setupSquare0100(i,j);
@@ -350,16 +350,16 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0011(i,j);
// else
// setupSquare0010(i,j);
// }
// else
// {
-// if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0001(i,j);
// else
// setupSquare0000(i,j);
@@ -397,7 +397,7 @@ Real tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
-__global__ void runCUDA(tnlFastSweepingMap< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int i, int* changed)
+__global__ void runCUDA(tnlFastSweepingMap< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i, int* changed)
{
__shared__ int something_changed;
@@ -512,7 +512,7 @@ __global__ void runCUDA(tnlFastSweepingMap< tnlGrid< 2,double, tnlHost, int >, d
}
-__global__ void initCUDA(tnlFastSweepingMap< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlFastSweepingMap< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
@@ -571,14 +571,14 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1111( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(INT_MAX,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -590,14 +590,14 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0000( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(-INT_MAX,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -609,18 +609,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1110( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -629,9 +629,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -642,18 +642,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1101( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -662,9 +662,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -675,18 +675,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1011( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -695,9 +695,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -708,17 +708,17 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0111( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
be=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
a = be/al;
@@ -728,9 +728,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -742,18 +742,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0001( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -762,9 +762,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -775,18 +775,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0010( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -795,9 +795,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -808,18 +808,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0100( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -828,9 +828,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -841,17 +841,17 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1000( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
be=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
a = be/al;
@@ -861,9 +861,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -878,18 +878,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1100( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -898,9 +898,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -911,18 +911,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1010( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -931,9 +931,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -944,14 +944,14 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1001( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(cudaDofVector[Entity.getIndex()],cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -968,18 +968,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0011( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -988,9 +988,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1001,18 +1001,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0101( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -1021,9 +1021,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
cudaDofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -1034,14 +1034,14 @@ template< typename MeshReal,
typename Index >
void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0110( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=fabsMin(cudaDofVector[Entity.getIndex()],cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
#endif
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap2D_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap2D_impl.h
index adff615fd43cd012ba5e45095140d7c2e0e5fbf0..5f4d6f86c6f2db94e7ed8ac728cb5334216c32a8 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap2D_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap2D_impl.h
@@ -103,7 +103,7 @@ template< typename MeshReal,
bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: initGrid()
{
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
for(int i=0; i< Mesh.getDimensions().x()*Mesh.getDimensions().x();i++)
{
dofVector2[i]=INT_MAX*sign(dofVector[i]);
@@ -122,22 +122,22 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
// {
// this->Entity.setCoordinates(CoordinatesType(i,j));
// this->Entity.refresh();
-// neighbourEntities.refresh(Mesh,Entity.getIndex());
+// neighborEntities.refresh(Mesh,Entity.getIndex());
//
// if(dofVector[this->Entity.getIndex()] > 0)
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1111(i,j);
// else
// setupSquare1110(i,j);
// }
// else
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1101(i,j);
// else
// setupSquare1100(i,j);
@@ -145,16 +145,16 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
// }
// else
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1011(i,j);
// else
// setupSquare1010(i,j);
// }
// else
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare1001(i,j);
// else
// setupSquare1000(i,j);
@@ -163,18 +163,18 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
// }
// else
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0111(i,j);
// else
// setupSquare0110(i,j);
// }
// else
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0101(i,j);
// else
// setupSquare0100(i,j);
@@ -182,16 +182,16 @@ bool tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
// }
// else
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0011(i,j);
// else
// setupSquare0010(i,j);
// }
// else
// {
-// if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+// if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
// setupSquare0001(i,j);
// else
// setupSquare0000(i,j);
@@ -287,30 +287,30 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
this->Entity.refresh();
if(map[Entity.getIndex()] != 0.0)
{
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real value = dofVector2[Entity.getIndex()];
Real im = abs(1.0/map[Entity.getIndex()]);
Real a,b, tmp;
if( i == 0 )
- a = dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ a = dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()];
else if( i == Mesh.getDimensions().x() - 1 )
- a = dofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ a = dofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
else
{
- a = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()],
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] );
+ a = fabsMin( dofVector2[neighborEntities.template getEntityIndex< -1, 0 >()],
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] );
}
if( j == 0 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()];
else if( j == Mesh.getDimensions().y() - 1 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
else
{
- b = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()],
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] );
+ b = fabsMin( dofVector2[neighborEntities.template getEntityIndex< 0, -1 >()],
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] );
}
@@ -362,11 +362,11 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
// this->Entity.setCoordinates(CoordinatesType(i,j));
// this->Entity.refresh();
-// auto neighbourEntities = Entity.getNeighbourEntities();
+// auto neighborEntities = Entity.getNeighborEntities();
// dofVector2[Entity.getIndex()]=fabsMin(INT_MAX,dofVector2[Entity.getIndex()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -380,11 +380,11 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
// this->Entity.setCoordinates(CoordinatesType(i,j));
// this->Entity.refresh();
-// auto neighbourEntities = Entity.getNeighbourEntities();
+// auto neighborEntities = Entity.getNeighborEntities();
// dofVector2[Entity.getIndex()]=fabsMin(-INT_MAX,dofVector2[(Entity.getIndex())]);
-// dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -398,15 +398,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -415,9 +415,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -430,15 +430,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -447,9 +447,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -462,15 +462,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -479,9 +479,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -494,14 +494,14 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
be=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
a = be/al;
@@ -511,9 +511,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -527,15 +527,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -544,9 +544,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -559,15 +559,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -576,9 +576,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -591,15 +591,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -608,9 +608,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -623,14 +623,14 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
be=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
a = be/al;
@@ -640,9 +640,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -659,15 +659,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -676,9 +676,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -691,15 +691,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -708,9 +708,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -723,11 +723,11 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
dofVector2[Entity.getIndex()]=fabsMin(dofVector[Entity.getIndex()],dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -746,15 +746,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -763,9 +763,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -778,15 +778,15 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -795,9 +795,9 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -810,11 +810,11 @@ void tnlFastSweepingMap< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index >
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
dofVector2[Entity.getIndex()]=fabsMin(dofVector[Entity.getIndex()],dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap_CUDA.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap_CUDA.h
index aa606ea47999ce739fb23d7ad6c38f937cf23f26..a23057e78c745e74467db4c4190d6f217024bc5a 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap_CUDA.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping-map/tnlFastSweepingMap_CUDA.h
@@ -17,9 +17,9 @@
#define TNLFASTSWEEPING_H_
#include <TNL/Config/ParameterContainer.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
#include <TNL/Containers/StaticVector.h>
-#include <core/tnlHost.h>
+#include <TNL/Devices/Host.h>
#include <mesh/tnlGrid.h>
#include <mesh/grids/tnlGridEntity.h>
@@ -54,7 +54,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
tnlFastSweepingMap();
@@ -131,7 +131,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -176,11 +176,11 @@ protected:
#ifdef HAVE_CUDA
//template<int sweep_t>
-__global__ void runCUDA(tnlFastSweepingMap< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int i, int* changed);
-//__global__ void runCUDA(tnlFastSweepingMap< tnlGrid< 3,double, tnlHost, int >, double, int >* solver, int sweep, int i);
+__global__ void runCUDA(tnlFastSweepingMap< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i, int* changed);
+//__global__ void runCUDA(tnlFastSweepingMap< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i);
-__global__ void initCUDA(tnlFastSweepingMap< tnlGrid< 2,double, tnlHost, int >, double, int >* solver);
-//__global__ void initCUDA(tnlFastSweepingMap< tnlGrid< 3,double, tnlHost, int >, double, int >* solver);
+__global__ void initCUDA(tnlFastSweepingMap< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver);
+//__global__ void initCUDA(tnlFastSweepingMap< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver);
#endif
/*various implementtions.... choose one*/
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/main.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/main.h
index 217377f3b77645f2ad2ede8db68bdada5f0885fb..8aca8f1b81db3a4a462d231e4724f3eef50e4723 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/main.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/main.h
@@ -47,26 +47,26 @@ int main( int argc, char* argv[] )
if(dim == 2)
{
- tnlFastSweeping<tnlGrid<2,double,tnlHost, int>, double, int> solver;
+ tnlFastSweeping<tnlGrid<2,double,TNL::Devices::Host, int>, double, int> solver;
if(!solver.init(parameters))
{
cerr << "Solver failed to initialize." << endl;
return EXIT_FAILURE;
}
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cout << "-------------------------------------------------------------" << endl;
cout << "Starting solver..." << endl;
solver.run();
}
else if(dim == 3)
{
- tnlFastSweeping<tnlGrid<3,double,tnlHost, int>, double, int> solver;
+ tnlFastSweeping<tnlGrid<3,double,TNL::Devices::Host, int>, double, int> solver;
if(!solver.init(parameters))
{
cerr << "Solver failed to initialize." << endl;
return EXIT_FAILURE;
}
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cout << "-------------------------------------------------------------" << endl;
cout << "Starting solver..." << endl;
solver.run();
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping.h
index 55f145af95deba9752fb5c4f564bbbfa331bf153..96d26db7b5a2077d8e2199292f0e888b0171a5c2 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping.h
@@ -17,10 +17,10 @@
#define TNLFASTSWEEPING_H_
#include <TNL/Config/ParameterContainer.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
#include <TNL/Containers/StaticVector.h>
#include <functions/tnlMeshFunction.h>
-#include <core/tnlHost.h>
+#include <TNL/Devices/Host.h>
#include <mesh/tnlGrid.h>
#include <mesh/grids/tnlGridEntity.h>
#include <limits.h>
@@ -55,7 +55,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -135,7 +135,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
tnlFastSweeping();
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_impl.h
index 48cbba5cb2e247296d41d3b7c2f935c8f75ce960..21e45020924453bd5676650028f6fdc6b210bc42 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_impl.h
@@ -89,7 +89,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
initCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
return true;
}
@@ -158,21 +158,21 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDeviceSynchronize();
}
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
for(int i = 0; i < 2*n ; i++)
{
runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,1,i);
cudaDeviceSynchronize();
}
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
for(int i = 0; i < 2*n ; i++)
{
runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,2,i);
cudaDeviceSynchronize();
}
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
for(int i = 2*n - 1; i > -1; i--)
{
runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,3,i);
@@ -180,7 +180,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
}
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cudaMemcpy(this->dofVector.getData(), cudaDofVector, this->dofVector.getSize()*sizeof(double), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
@@ -439,7 +439,7 @@ Real tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
-__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int i)
+__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i)
{
int gx = threadIdx.x + blockDim.x*blockIdx.x;
@@ -502,7 +502,7 @@ __global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, doub
}
-__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
int gx = threadIdx.x + blockDim.x*blockIdx.x;
int gy = blockDim.y*blockIdx.y + threadIdx.y;
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v2_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v2_impl.h
index e6b20eb4153b04248a3bf72b2198d923aa21d748..1d4ee11b0856a618f7b86452828a2bd2e71bf24e 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v2_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v2_impl.h
@@ -89,7 +89,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
initCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
return true;
}
@@ -158,21 +158,21 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDeviceSynchronize();
}
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//// for(int i = 0; i < 2*n ; i++)
// {
// runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,1,0);
// cudaDeviceSynchronize();
// }
// cudaDeviceSynchronize();
-// checkCudaDevice;
+// TNL_CHECK_CUDA_DEVICE;
//// for(int i = 0; i < 2*n ; i++)
// {
// runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,2,0);
// cudaDeviceSynchronize();
// }
// cudaDeviceSynchronize();
-// checkCudaDevice;
+// TNL_CHECK_CUDA_DEVICE;
//// for(int i = 2*n - 1; i > -1; i--)
// {
// runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,3,0);
@@ -180,7 +180,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
// }
//
// cudaDeviceSynchronize();
-// checkCudaDevice;
+// TNL_CHECK_CUDA_DEVICE;
cudaMemcpy(this->dofVector.getData(), cudaDofVector, this->dofVector.getSize()*sizeof(double), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
@@ -438,7 +438,7 @@ Real tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
-__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int k)
+__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int k)
{
//int gx = threadIdx.x;
@@ -568,7 +568,7 @@ __global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, doub
}
-__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
int gx = threadIdx.x + blockDim.x*blockIdx.x;
int gy = blockDim.y*blockIdx.y + threadIdx.y;
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v3_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v3_impl.h
index d05b649f29d222bc96649d3d6ed92a7fa11cf14e..fad3bc293cc452c1fe077e892b13b51750a78bb6 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v3_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v3_impl.h
@@ -104,7 +104,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
initCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
return true;
}
@@ -174,21 +174,21 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
//cudaDeviceSynchronize();
}
// cudaDeviceSynchronize();
-// checkCudaDevice;
+// TNL_CHECK_CUDA_DEVICE;
// for(int i = 0; i < 2*m -1; i++)
// {
// runCUDA<2><<<numBlocks,threadsPerBlock>>>(this->cudaSolver,2,i);
// cudaDeviceSynchronize();
// }
// cudaDeviceSynchronize();
-// checkCudaDevice;
+// TNL_CHECK_CUDA_DEVICE;
// for(int i = 0; i < 2*m -1; i++)
// {
// runCUDA<4><<<numBlocks,threadsPerBlock>>>(this->cudaSolver,4,i);
// cudaDeviceSynchronize();
// }
// cudaDeviceSynchronize();
-// checkCudaDevice;
+// TNL_CHECK_CUDA_DEVICE;
// for(int i = 0; i < 2*m -1; i++)
// {
// runCUDA<8><<<numBlocks,threadsPerBlock>>>(this->cudaSolver,8,i);
@@ -234,7 +234,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
// cudaDeviceSynchronize();
// }
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cudaMemcpy(this->dofVector.getData(), cudaDofVector, this->dofVector.getSize()*sizeof(double), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
@@ -492,7 +492,7 @@ Real tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
template<>
-__global__ void runCUDA<1>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int k)
+__global__ void runCUDA<1>(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int k)
{
if(blockIdx.x+blockIdx.y == k)
@@ -516,7 +516,7 @@ __global__ void runCUDA<1>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, d
/*---------------------------------------------------------------------------------------------------------------------------*/
}
template<>
- __global__ void runCUDA<2>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int k)
+ __global__ void runCUDA<2>(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int k)
{
if((gridDim.x - blockIdx.x - 1)+blockIdx.y == k)
{
@@ -538,7 +538,7 @@ __global__ void runCUDA<1>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, d
}
} /*---------------------------------------------------------------------------------------------------------------------------*/
template<>
- __global__ void runCUDA<4>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int k)
+ __global__ void runCUDA<4>(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int k)
{
if(blockIdx.x+blockIdx.y == gridDim.x+gridDim.y-k-2)
{
@@ -564,7 +564,7 @@ __global__ void runCUDA<1>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, d
}
template<>
- __global__ void runCUDA<8>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int k)
+ __global__ void runCUDA<8>(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int k)
{
if((gridDim.x - blockIdx.x - 1)+blockIdx.y == gridDim.x+gridDim.y-k-2)
{
@@ -595,7 +595,7 @@ __global__ void runCUDA<1>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, d
template<>
- __global__ void runCUDA<5>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int k)
+ __global__ void runCUDA<5>(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int k)
{
if(blockIdx.x+blockIdx.y == k)
@@ -640,7 +640,7 @@ __global__ void runCUDA<1>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, d
template<>
- __global__ void runCUDA<10>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int k)
+ __global__ void runCUDA<10>(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int k)
{
if((gridDim.x - blockIdx.x - 1)+blockIdx.y == k)
{
@@ -687,7 +687,7 @@ __global__ void runCUDA<1>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, d
template<>
- __global__ void runCUDA<15>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int k)
+ __global__ void runCUDA<15>(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int k)
{
if(blockIdx.x+blockIdx.y == k)
@@ -806,7 +806,7 @@ __global__ void runCUDA<1>(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, d
-__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
int gx = threadIdx.x + blockDim.x*blockIdx.x;
int gy = blockDim.y*blockIdx.y + threadIdx.y;
@@ -825,7 +825,7 @@ __global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, dou
-//__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int k)
+//__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int k)
//{
//
// if(sweep==1 && blockIdx.x+blockIdx.y == k)
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v4_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v4_impl.h
index 230b6ce436bb53c40172eb43cf8b7e668f4bf891..dd878d4953d46ac014622a43e03ad0e8b6b9a492 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v4_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v4_impl.h
@@ -122,7 +122,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
initCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
return true;
}
@@ -147,7 +147,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,0,0);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//data.setLike(dofVector.getData());
//cudaMemcpy(data.getData(), cudaDofVector2, this->dofVector.getData().getSize()*sizeof(double), cudaMemcpyDeviceToHost);
@@ -176,32 +176,32 @@ template< typename MeshReal,
__device__
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: updateValue( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real value = cudaDofVector2[Entity.getIndex()];
Real a,b, tmp;
if( i == 0 )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()];
else if( i == Mesh.getDimensions().x() - 1 )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
else
{
- a = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] );
+ a = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] );
}
if( j == 0 )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()];
else if( j == Mesh.getDimensions().y() - 1 )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
else
{
- b = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] );
+ b = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] );
}
@@ -227,11 +227,11 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
int i = threadIdx.x + blockDim.x*blockIdx.x;
int j = blockDim.y*blockIdx.y + threadIdx.y;
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
int gid = Entity.getIndex();
@@ -248,18 +248,18 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
if(cudaDofVector[Entity.getIndex()] > 0)
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1111(i,j);
else
setupSquare1110(i,j);
}
else
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1101(i,j);
else
setupSquare1100(i,j);
@@ -267,16 +267,16 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
}
else
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1011(i,j);
else
setupSquare1010(i,j);
}
else
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1001(i,j);
else
setupSquare1000(i,j);
@@ -285,18 +285,18 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
}
else
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0111(i,j);
else
setupSquare0110(i,j);
}
else
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0101(i,j);
else
setupSquare0100(i,j);
@@ -304,16 +304,16 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
}
else
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0011(i,j);
else
setupSquare0010(i,j);
}
else
{
- if(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0001(i,j);
else
setupSquare0000(i,j);
@@ -357,7 +357,7 @@ Real tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
-__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int i)
+__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i)
{
int gx = 0;
@@ -464,7 +464,7 @@ __global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, doub
}
-__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
@@ -523,14 +523,14 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1111( Index i, Index j)
{
-// tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+// tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
// Entity.setCoordinates(CoordinatesType(i,j));
// Entity.refresh();
-// tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+// tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
// cudaDofVector2[Entity.getIndex()]=fabsMin(INT_MAX,cudaDofVector2[Entity.getIndex()]);
-// cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
-// cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
-// cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+// cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+// cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+// cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -542,14 +542,14 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0000( Index i, Index j)
{
-// tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+// tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
// Entity.setCoordinates(CoordinatesType(i,j));
// Entity.refresh();
-// tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+// tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
// cudaDofVector2[Entity.getIndex()]=fabsMin(-INT_MAX,cudaDofVector2[Entity.getIndex()]);
-// cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
-// cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
-// cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+// cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+// cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+// cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -561,18 +561,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1110( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -581,9 +581,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=INT_MAX; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -594,18 +594,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1101( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -614,9 +614,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=0.5*h; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=INT_MAX; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=INT_MAX; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -627,18 +627,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1011( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -647,9 +647,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=INT_MAX; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=INT_MAX; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -660,17 +660,17 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0111( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
be=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
a = be/al;
@@ -680,9 +680,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=-0.5*h; //fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=INT_MAX; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=INT_MAX; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -694,18 +694,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0001( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -714,9 +714,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=-INT_MAX; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -727,18 +727,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0010( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -747,9 +747,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=-INT_MAX; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=-INT_MAX; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -760,18 +760,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0100( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
- al=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(cudaDofVector[Entity.getIndex()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -780,9 +780,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=INT_MAX; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=INT_MAX; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -793,17 +793,17 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1000( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
be=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
a = be/al;
@@ -813,9 +813,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=0.5*h; //fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=-INT_MAX; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=-INT_MAX; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -830,18 +830,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1100( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -850,9 +850,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=0.5*h; //fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -863,18 +863,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1010( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -883,9 +883,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=0.5*h; //fabsMin(abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(-abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -896,14 +896,14 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare1001( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=0.5*h; //fabsMin(cudaDofVector[Entity.getIndex()],cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -920,18 +920,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0011( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -940,9 +940,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=-0.5*h; //fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -953,18 +953,18 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0101( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real al,be, a,b,c,s;
al=abs(cudaDofVector[Entity.getIndex()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
cudaDofVector[Entity.getIndex()]));
- be=abs(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b = 1.0;
@@ -973,9 +973,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
cudaDofVector2[Entity.getIndex()]=-0.5*h; //fabsMin(-abs(a*0+b*0+c)*s,cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=-0.5*h; //fabsMin(-abs(a*1+b*0+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=0.5*h; //fabsMin(abs(a*1+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(abs(a*0+b*1+c)*s,cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -986,14 +986,14 @@ template< typename MeshReal,
typename Index >
void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: setupSquare0110( Index i, Index j)
{
- tnlGridEntity< tnlGrid< 2,double, tnlHost, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 2,double, TNL::Devices::Host, int >, 2, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
cudaDofVector2[Entity.getIndex()]=-0.5*h; //fabsMin(cudaDofVector[Entity.getIndex()],cudaDofVector2[Entity.getIndex()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(cudaDofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=0.5*h; //fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 0, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=-0.5*h; //fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 1 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=0.5*h; //fabsMin(cudaDofVector[neighborEntities.template getEntityIndex< 1, 0 >()],cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
#endif
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v5_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v5_impl.h
index f9f3e28f63fc7a3ad88b98980a4454035fd299b4..4f8c90df818b7121663e8053b0577d37fd119055 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v5_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_CUDA_v5_impl.h
@@ -112,7 +112,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
initCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
return true;
}
@@ -179,7 +179,7 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
runCUDA<<<numBlocks,threadsPerBlock,3*(512+1)*sizeof(double)>>>(this->cudaSolver,0,0);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cudaMemcpy(this->dofVector.getData(), cudaDofVector, this->dofVector.getSize()*sizeof(double), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
@@ -491,7 +491,7 @@ Real tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
-__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int i)
+__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i)
{
extern __shared__ double u[];
@@ -677,7 +677,7 @@ __global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, doub
}
-__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver)
{
int gx = threadIdx.x + blockDim.x*blockIdx.x;
int gy = blockDim.y*blockIdx.y + threadIdx.y;
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_impl.h
index 920806c211483d1c8cc3dabd2e194704dbb560bd..8fd78cea97230d16135ddd03903442bb9922e0b8 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping2D_impl.h
@@ -93,7 +93,7 @@ template< typename MeshReal,
bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > :: initGrid()
{
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
for(int i=0; i< Mesh.getDimensions().x()*Mesh.getDimensions().x();i++)
{
dofVector2[i]=INT_MAX*sign(dofVector[i]);
@@ -105,22 +105,22 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- neighbourEntities.refresh(Mesh,Entity.getIndex());
+ neighborEntities.refresh(Mesh,Entity.getIndex());
if(dofVector[this->Entity.getIndex()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1111(i,j);
else
setupSquare1110(i,j);
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1101(i,j);
else
setupSquare1100(i,j);
@@ -128,16 +128,16 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1011(i,j);
else
setupSquare1010(i,j);
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare1001(i,j);
else
setupSquare1000(i,j);
@@ -146,18 +146,18 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0111(i,j);
else
setupSquare0110(i,j);
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0101(i,j);
else
setupSquare0100(i,j);
@@ -165,16 +165,16 @@ bool tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()] > 0)
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0011(i,j);
else
setupSquare0010(i,j);
}
else
{
- if(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()] > 0)
+ if(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()] > 0)
setupSquare0001(i,j);
else
setupSquare0000(i,j);
@@ -397,29 +397,29 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Real value = dofVector2[Entity.getIndex()];
Real a,b, tmp;
if( i == 0 )
- a = dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()];
+ a = dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()];
else if( i == Mesh.getDimensions().x() - 1 )
- a = dofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()];
+ a = dofVector2[neighborEntities.template getEntityIndex< -1, 0 >()];
else
{
- a = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< -1, 0 >()],
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()] );
+ a = fabsMin( dofVector2[neighborEntities.template getEntityIndex< -1, 0 >()],
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()] );
}
if( j == 0 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()];
else if( j == Mesh.getDimensions().y() - 1 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, -1 >()];
else
{
- b = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< 0, -1 >()],
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()] );
+ b = fabsMin( dofVector2[neighborEntities.template getEntityIndex< 0, -1 >()],
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()] );
}
@@ -466,11 +466,11 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
// this->Entity.setCoordinates(CoordinatesType(i,j));
// this->Entity.refresh();
-// auto neighbourEntities = Entity.getNeighbourEntities();
+// auto neighborEntities = Entity.getNeighborEntities();
// dofVector2[Entity.getIndex()]=fabsMin(INT_MAX,dofVector2[Entity.getIndex()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -484,11 +484,11 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
// this->Entity.setCoordinates(CoordinatesType(i,j));
// this->Entity.refresh();
-// auto neighbourEntities = Entity.getNeighbourEntities();
+// auto neighborEntities = Entity.getNeighborEntities();
// dofVector2[Entity.getIndex()]=fabsMin(-INT_MAX,dofVector2[(Entity.getIndex())]);
-// dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
-// dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+// dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-INT_MAX,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -502,15 +502,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -519,9 +519,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -534,15 +534,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -551,9 +551,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -566,15 +566,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -583,9 +583,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -598,14 +598,14 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
be=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
a = be/al;
@@ -615,9 +615,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -631,15 +631,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]));
a = be/al;
b=1.0;
@@ -648,9 +648,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -663,15 +663,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = be/al;
b=1.0;
@@ -680,9 +680,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -695,15 +695,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
- al=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
+ al=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
(dofVector[Entity.getIndex()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = be/al;
b=1.0;
@@ -712,9 +712,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -727,14 +727,14 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
be=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
a = be/al;
@@ -744,9 +744,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -763,15 +763,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -780,9 +780,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -795,15 +795,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -812,9 +812,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(-abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -827,11 +827,11 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
dofVector2[Entity.getIndex()]=fabsMin(dofVector[Entity.getIndex()],dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -850,15 +850,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]));
a = al-be;
b=1.0;
@@ -867,9 +867,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -882,15 +882,15 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
Real al,be, a,b,c,s;
al=abs(dofVector[Entity.getIndex()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()]-
+ (dofVector[neighborEntities.template getEntityIndex< 1, 0 >()]-
dofVector[Entity.getIndex()]));
- be=abs(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]/
- (dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()]-
- dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()]));
+ be=abs(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]/
+ (dofVector[neighborEntities.template getEntityIndex< 1, 1 >()]-
+ dofVector[neighborEntities.template getEntityIndex< 0, 1 >()]));
a = al-be;
b=1.0;
@@ -899,9 +899,9 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
dofVector2[Entity.getIndex()]=fabsMin(-abs(a*0+b*0+c)*s,dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(-abs(a*1+b*0+c)*s,dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(abs(a*1+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(abs(a*0+b*1+c)*s,dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
@@ -914,11 +914,11 @@ void tnlFastSweeping< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j));
this->Entity.refresh();
- auto neighbourEntities = Entity.getNeighbourEntities();
+ auto neighborEntities = Entity.getNeighborEntities();
dofVector2[Entity.getIndex()]=fabsMin(dofVector[Entity.getIndex()],dofVector2[(Entity.getIndex())]);
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 0, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 1 >()]);
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighbourEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighbourEntities.template getEntityIndex< 1, 0 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 0, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 0, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 1 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 1 >()]);
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]=fabsMin(dofVector[neighborEntities.template getEntityIndex< 1, 0 >()],dofVector2[neighborEntities.template getEntityIndex< 1, 0 >()]);
}
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping3D_CUDA_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping3D_CUDA_impl.h
index da045676ae1f119e73cc4539732b31d5d632ca23..51bb61716d1b49e2921d974e93692cdf0a34ddf1 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping3D_CUDA_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping3D_CUDA_impl.h
@@ -111,10 +111,10 @@ bool tnlFastSweeping< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > ::
dim3 numBlocks(n/8 + 1, n/8 +1, n/8 +1);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
return true;
}
@@ -139,7 +139,7 @@ bool tnlFastSweeping< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > ::
runCUDA<<<numBlocks,threadsPerBlock>>>(this->cudaSolver,0,0);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cudaMemcpy(this->dofVector.getData().getData(), cudaDofVector2, this->dofVector.getData().getSize()*sizeof(double), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
@@ -165,41 +165,41 @@ template< typename MeshReal,
__device__
void tnlFastSweeping< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > :: updateValue( Index i, Index j, Index k)
{
- tnlGridEntity< tnlGrid< 3,double, tnlHost, int >, 3, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 3,double, TNL::Devices::Host, int >, 3, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j,k));
Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities(Entity);
Real value = cudaDofVector2[Entity.getIndex()];
Real a,b,c, tmp;
if( i == 0 )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0, 0 >()];
else if( i == Mesh.getDimensions().x() - 1 )
- a = cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0, 0 >()];
+ a = cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0, 0 >()];
else
{
- a = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< -1, 0, 0 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 1, 0, 0 >()] );
+ a = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< -1, 0, 0 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 1, 0, 0 >()] );
}
if( j == 0 )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1, 0 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1, 0 >()];
else if( j == Mesh.getDimensions().y() - 1 )
- b = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1, 0 >()];
+ b = cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1, 0 >()];
else
{
- b = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< 0, -1, 0 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 1, 0 >()] );
+ b = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< 0, -1, 0 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 1, 0 >()] );
}
if( k == 0 )
- c = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 0, 1 >()];
+ c = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 0, 1 >()];
else if( k == Mesh.getDimensions().z() - 1 )
- c = cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 0, -1 >()];
+ c = cudaDofVector2[neighborEntities.template getEntityIndex< 0, 0, -1 >()];
else
{
- c = fabsMin( cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 0, -1 >()],
- cudaDofVector2[neighbourEntities.template getEntityIndex< 0, 0, 1 >()] );
+ c = fabsMin( cudaDofVector2[neighborEntities.template getEntityIndex< 0, 0, -1 >()],
+ cudaDofVector2[neighborEntities.template getEntityIndex< 0, 0, 1 >()] );
}
Real hD = 3.0*h*h - 2.0*(a*a + b*b + c*c - a*b - a*c - b*c);
@@ -222,7 +222,7 @@ template< typename MeshReal,
__device__
bool tnlFastSweeping< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > :: initGrid(int i, int j, int k)
{
- tnlGridEntity< tnlGrid< 3,double, tnlHost, int >, 3, tnlGridEntityNoStencilStorage > Entity(Mesh);
+ tnlGridEntity< tnlGrid< 3,double, TNL::Devices::Host, int >, 3, tnlGridEntityNoStencilStorage > Entity(Mesh);
Entity.setCoordinates(CoordinatesType(i,j,k));
Entity.refresh();
int gid = Entity.getIndex();
@@ -255,7 +255,7 @@ Real tnlFastSweeping< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > ::
-__global__ void runCUDA(tnlFastSweeping< tnlGrid< 3,double, tnlHost, int >, double, int >* solver, int sweep, int i)
+__global__ void runCUDA(tnlFastSweeping< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i)
{
int gx = 0;
@@ -474,7 +474,7 @@ __global__ void runCUDA(tnlFastSweeping< tnlGrid< 3,double, tnlHost, int >, doub
}
-__global__ void initCUDA(tnlFastSweeping< tnlGrid< 3,double, tnlHost, int >, double, int >* solver)
+__global__ void initCUDA(tnlFastSweeping< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver)
{
int gx = threadIdx.x + blockDim.x*blockIdx.x;
int gy = blockDim.y*blockIdx.y + threadIdx.y;
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping3D_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping3D_impl.h
index dd5681e6d88b3fa666f210d0fe24f2423ac78b68..dc4fd4f866c19b09d17ad8b2c19d2907d581d1c8 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping3D_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping3D_impl.h
@@ -237,38 +237,38 @@ void tnlFastSweeping< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > ::
{
this->Entity.setCoordinates(CoordinatesType(i,j,k));
this->Entity.refresh();
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities(Entity);
Real value = dofVector2[Entity.getIndex()];
Real a,b,c, tmp;
if( i == 0 )
- a = dofVector2[neighbourEntities.template getEntityIndex< 1, 0, 0>()];
+ a = dofVector2[neighborEntities.template getEntityIndex< 1, 0, 0>()];
else if( i == Mesh.getDimensions().x() - 1 )
- a = dofVector2[neighbourEntities.template getEntityIndex< -1, 0, 0 >()];
+ a = dofVector2[neighborEntities.template getEntityIndex< -1, 0, 0 >()];
else
{
- a = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< -1, 0, 0>()],
- dofVector2[neighbourEntities.template getEntityIndex< 1, 0, 0>()] );
+ a = fabsMin( dofVector2[neighborEntities.template getEntityIndex< -1, 0, 0>()],
+ dofVector2[neighborEntities.template getEntityIndex< 1, 0, 0>()] );
}
if( j == 0 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, 1, 0>()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, 1, 0>()];
else if( j == Mesh.getDimensions().y() - 1 )
- b = dofVector2[neighbourEntities.template getEntityIndex< 0, -1, 0>()];
+ b = dofVector2[neighborEntities.template getEntityIndex< 0, -1, 0>()];
else
{
- b = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< 0, -1, 0>()],
- dofVector2[neighbourEntities.template getEntityIndex< 0, 1, 0>()] );
+ b = fabsMin( dofVector2[neighborEntities.template getEntityIndex< 0, -1, 0>()],
+ dofVector2[neighborEntities.template getEntityIndex< 0, 1, 0>()] );
}
if( k == 0 )
- c = dofVector2[neighbourEntities.template getEntityIndex< 0, 0, 1>()];
+ c = dofVector2[neighborEntities.template getEntityIndex< 0, 0, 1>()];
else if( k == Mesh.getDimensions().z() - 1 )
- c = dofVector2[neighbourEntities.template getEntityIndex< 0, 0, -1>()];
+ c = dofVector2[neighborEntities.template getEntityIndex< 0, 0, -1>()];
else
{
- c = fabsMin( dofVector2[neighbourEntities.template getEntityIndex< 0, 0, -1>()],
- dofVector2[neighbourEntities.template getEntityIndex< 0, 0, 1>()] );
+ c = fabsMin( dofVector2[neighborEntities.template getEntityIndex< 0, 0, -1>()],
+ dofVector2[neighborEntities.template getEntityIndex< 0, 0, 1>()] );
}
Real hD = 3.0*h*h - 2.0*(a*a+b*b+c*c-a*b-a*c-b*c);
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping_CUDA.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping_CUDA.h
index 310cdf3f3028eb68e71c31d821ffea1538615724..f531da431bfec5d16da8ea7deabe6595031a0873 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping_CUDA.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/fast-sweeping/tnlFastSweeping_CUDA.h
@@ -17,9 +17,9 @@
#define TNLFASTSWEEPING_H_
#include <TNL/Config/ParameterContainer.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
#include <TNL/Containers/StaticVector.h>
-#include <core/tnlHost.h>
+#include <TNL/Devices/Host.h>
#include <mesh/tnlGrid.h>
#include <mesh/grids/tnlGridEntity.h>
@@ -54,7 +54,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 2, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
tnlFastSweeping();
@@ -129,7 +129,7 @@ public:
typedef Device DeviceType;
typedef Index IndexType;
typedef tnlGrid< 3, Real, Device, Index > MeshType;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
+ typedef TNL::Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::CoordinatesType CoordinatesType;
@@ -174,11 +174,11 @@ protected:
#ifdef HAVE_CUDA
//template<int sweep_t>
-__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver, int sweep, int i);
-__global__ void runCUDA(tnlFastSweeping< tnlGrid< 3,double, tnlHost, int >, double, int >* solver, int sweep, int i);
+__global__ void runCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i);
+__global__ void runCUDA(tnlFastSweeping< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver, int sweep, int i);
-__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, tnlHost, int >, double, int >* solver);
-__global__ void initCUDA(tnlFastSweeping< tnlGrid< 3,double, tnlHost, int >, double, int >* solver);
+__global__ void initCUDA(tnlFastSweeping< tnlGrid< 2,double, TNL::Devices::Host, int >, double, int >* solver);
+__global__ void initCUDA(tnlFastSweeping< tnlGrid< 3,double, TNL::Devices::Host, int >, double, int >* solver);
#endif
/*various implementtions.... choose one*/
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/main.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/main.h
index 55a0942f81721c1e90e670aaae18941471f6e13d..911f0a29d9f0a82aa73767cfcee407093bc2eb66 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/main.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/main.h
@@ -41,24 +41,24 @@ int main( int argc, char* argv[] )
tnlDeviceEnum device;
- device = tnlHostDevice;
+ device = TNL::Devices::HostDevice;
const int& dim = parameters.getParameter< int >( "dim" );
if(dim == 2)
{
- typedef parallelGodunovMapScheme< tnlGrid<2,double,tnlHost, int>, double, int > SchemeTypeHost;
+ typedef parallelGodunovMapScheme< tnlGrid<2,double,TNL::Devices::Host, int>, double, int > SchemeTypeHost;
/*#ifdef HAVE_CUDA
typedef parallelGodunovMapScheme< tnlGrid<2,double,tnlCuda, int>, double, int > SchemeTypeDevice;
#endif
#ifndef HAVE_CUDA*/
- typedef parallelGodunovMapScheme< tnlGrid<2,double,tnlHost, int>, double, int > SchemeTypeDevice;
+ typedef parallelGodunovMapScheme< tnlGrid<2,double,TNL::Devices::Host, int>, double, int > SchemeTypeDevice;
/*#endif*/
- if(device==tnlHostDevice)
+ if(device==TNL::Devices::HostDevice)
{
- typedef tnlHost Device;
+ typedef TNL::Devices::Host Device;
tnlParallelMapSolver<2,SchemeTypeHost,SchemeTypeDevice, Device> solver;
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/tnlParallelMapSolver.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/tnlParallelMapSolver.h
index bf0a388166c89946727b95687eca1a74764739c3..400e163c9dcc8d536a478a0952aabf8ccbb1a2d8 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/tnlParallelMapSolver.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/tnlParallelMapSolver.h
@@ -18,10 +18,10 @@
#define TNLPARALLELMAPSOLVER_H_
#include <TNL/Config/ParameterContainer.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
#include <TNL/Containers/StaticVector.h>
#include <functions/tnlMeshFunction.h>
-#include <core/tnlHost.h>
+#include <TNL/Devices/Host.h>
#include <mesh/tnlGrid.h>
#include <mesh/grids/tnlGridEntity.h>
#include <limits.h>
@@ -31,7 +31,6 @@
#include <ctime>
#ifdef HAVE_CUDA
-#include <cuda.h>
#include <core/tnlCuda.h>
#endif
@@ -53,13 +52,13 @@ public:
typedef SchemeDevice SchemeTypeDevice;
typedef SchemeHost SchemeTypeHost;
typedef Device DeviceType;
- typedef tnlVector< double, tnlHost, int > VectorType;
- typedef tnlVector< int, tnlHost, int > IntVectorType;
- typedef tnlGrid< 2, double, tnlHost, int > MeshType;
+ typedef TNL::Containers::Vector< double, TNL::Devices::Host, int > VectorType;
+ typedef TNL::Containers::Vector< int, TNL::Devices::Host, int > IntVectorType;
+ typedef tnlGrid< 2, double, TNL::Devices::Host, int > MeshType;
#ifdef HAVE_CUDA
- typedef tnlVector< double, tnlHost, int > VectorTypeCUDA;
- typedef tnlVector< int, tnlHost, int > IntVectorTypeCUDA;
- typedef tnlGrid< 2, double, tnlHost, int > MeshTypeCUDA;
+ typedef TNL::Containers::Vector< double, TNL::Devices::Host, int > VectorTypeCUDA;
+ typedef TNL::Containers::Vector< int, TNL::Devices::Host, int > IntVectorTypeCUDA;
+ typedef tnlGrid< 2, double, TNL::Devices::Host, int > MeshTypeCUDA;
#endif
tnlParallelMapSolver();
bool init( const Config::ParameterContainer& parameters );
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/tnlParallelMapSolver2D_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/tnlParallelMapSolver2D_impl.h
index 1cb730e42d0b0b72782e61e011b6081d2eddc6ef..2925df8b236f13b0d3c460b5ca9b2970ff730e80 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/tnlParallelMapSolver2D_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel-map/tnlParallelMapSolver2D_impl.h
@@ -31,7 +31,7 @@
template< typename SchemeHost, typename SchemeDevice, typename Device>
tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::tnlParallelMapSolver()
{
- this->device = tnlHostDevice; /////////////// tnlCuda Device --- vypocet na GPU, tnlHostDevice --- vypocet na CPU
+ this->device = TNL::Devices::HostDevice; /////////////// tnlCuda Device --- vypocet na GPU, TNL::Devices::HostDevice --- vypocet na CPU
#ifdef HAVE_CUDA
if(this->device == tnlCudaDevice)
@@ -123,7 +123,7 @@ bool tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::init
cudaMalloc(&(this->tmp_map), this->map_stretched.getSize()*sizeof(double));
cudaMalloc(&(this->runcuda), sizeof(int));
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
int* tmpUC;
cudaMalloc(&(tmpUC), this->work_u.getSize()*sizeof(int));
@@ -131,19 +131,19 @@ bool tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::init
initCUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<1,1>>>(this->cudaSolver, (this->tmpw), (this->runcuda),tmpUC, tmp_map);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
double* tmpu = NULL;
cudaMemcpy(&tmpu, tmpdev,sizeof(double*), cudaMemcpyDeviceToHost);
cudaMemcpy((this->tmpw), this->work_u.getData(), this->work_u.getSize()*sizeof(double), cudaMemcpyHostToDevice);
cudaMemcpy((this->tmp_map), this->map_stretched.getData(), this->map_stretched.getSize()*sizeof(double), cudaMemcpyHostToDevice);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
#endif
- if(this->device == tnlHostDevice)
+ if(this->device == TNL::Devices::HostDevice)
{
VectorType tmp_map;
tmp_map.setSize(this->n * this->n);
@@ -187,21 +187,21 @@ bool tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::init
else if(this->device == tnlCudaDevice)
{
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 threadsPerBlock(this->n, this->n);
dim3 numBlocks(this->gridCols,this->gridRows);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initRunCUDA2D<SchemeTypeHost,SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock,3*this->n*this->n*sizeof(double)>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
#endif
this->currentStep = 1;
- if(this->device == tnlHostDevice)
+ if(this->device == TNL::Devices::HostDevice)
synchronize();
#ifdef HAVE_CUDA
else if(this->device == tnlCudaDevice)
@@ -211,10 +211,10 @@ bool tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::init
synchronizeCUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
synchronize2CUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
#endif
@@ -226,7 +226,7 @@ bool tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::init
template< typename SchemeHost, typename SchemeDevice, typename Device>
void tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::run()
{
- if(this->device == tnlHostDevice)
+ if(this->device == TNL::Devices::HostDevice)
{
while ((this->boundaryConditions.max() > 0 )/* || !end*/)
{
@@ -360,12 +360,12 @@ void tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::run(
dim3 threadsPerBlock(this->n, this->n);
dim3 numBlocks(this->gridCols,this->gridRows);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
bool* tmpb;
cudaMemcpy(&(this->run_host),this->runcuda,sizeof(int), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
int i = 1;
time_diff = 0.0;
@@ -377,7 +377,7 @@ void tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::run(
else
end_cuda = false;
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
start = std::clock();
runCUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock,3*this->n*this->n*sizeof(double)>>>(this->cudaSolver);
cudaDeviceSynchronize();
@@ -386,10 +386,10 @@ void tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::run(
//start = std::clock();
synchronizeCUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
synchronize2CUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//time_diff += (std::clock() - start) / (double)(CLOCKS_PER_SEC);
cudaMemcpy(&run_host, (this->runcuda),sizeof(int), cudaMemcpyDeviceToHost);
@@ -738,7 +738,7 @@ tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::runSubgri
double maxResidue( 1.0 );
tnlGridEntity<MeshType, 2, tnlGridEntityNoStencilStorage > Entity(subMesh);
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
for( int i = 0; i < u.getSize(); i ++ )
{
@@ -758,9 +758,9 @@ tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::runSubgri
{
Entity.setCoordinates(Containers::StaticVector<2,int>(i % subMesh.getDimensions().x(),i / subMesh.getDimensions().x()));
Entity.refresh();
- neighbourEntities.refresh(subMesh,Entity.getIndex());
+ neighborEntities.refresh(subMesh,Entity.getIndex());
if(map[i] != 0.0)
- fu[ i ] = schemeHost.getValue( this->subMesh, i, Containers::StaticVector<2,int>(i % subMesh.getDimensions().x(),i / subMesh.getDimensions().x()), u, time, boundaryCondition,neighbourEntities,map);
+ fu[ i ] = schemeHost.getValue( this->subMesh, i, Containers::StaticVector<2,int>(i % subMesh.getDimensions().x(),i / subMesh.getDimensions().x()), u, time, boundaryCondition,neighborEntities,map);
}
maxResidue = fu. absMax();
@@ -909,10 +909,10 @@ void tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::runS
__syncthreads();
tnlGridEntity<MeshType, 2, tnlGridEntityNoStencilStorage > Entity(subMesh);
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Entity.setCoordinates(Containers::StaticVector<2,int>(i,j));
Entity.refresh();
- neighbourEntities.refresh(subMesh,Entity.getIndex());
+ neighborEntities.refresh(subMesh,Entity.getIndex());
if(map_local[l] == 0.0)
@@ -929,7 +929,7 @@ void tnlParallelMapSolver<2,SchemeHost, SchemeDevice, Device, double, int>::runS
if(computeFU)
{
- fu = schemeHost.getValueDev( this->subMesh, l, Containers::StaticVector<2,int>(i,j), u, time, boundaryCondition, neighbourEntities, map_local);
+ fu = schemeHost.getValueDev( this->subMesh, l, Containers::StaticVector<2,int>(i,j), u, time, boundaryCondition, neighborEntities, map_local);
sharedTau[l]=abs(cfl/fu);
}
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/main.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/main.h
index 3a976dae469a77dc33559cf562f722913de0ef28..178e816aff7c1fe2c46b16529b1b263b24796879 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/main.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/main.h
@@ -42,24 +42,24 @@ int main( int argc, char* argv[] )
//if (parameters.GetParameter <String>("scheme") == "godunov")
//{
tnlDeviceEnum device;
- device = tnlHostDevice;
+ device = TNL::Devices::HostDevice;
const int& dim = parameters.getParameter< int >( "dim" );
if(dim == 2)
{
- typedef parallelGodunovEikonalScheme< tnlGrid<2,double,tnlHost, int>, double, int > SchemeTypeHost;
+ typedef parallelGodunovEikonalScheme< tnlGrid<2,double,TNL::Devices::Host, int>, double, int > SchemeTypeHost;
/*#ifdef HAVE_CUDA
typedef parallelGodunovEikonalScheme< tnlGrid<2,double,tnlCuda, int>, double, int > SchemeTypeDevice;
#endif
#ifndef HAVE_CUDA*/
- typedef parallelGodunovEikonalScheme< tnlGrid<2,double,tnlHost, int>, double, int > SchemeTypeDevice;
+ typedef parallelGodunovEikonalScheme< tnlGrid<2,double,TNL::Devices::Host, int>, double, int > SchemeTypeDevice;
/*#endif*/
- if(device==tnlHostDevice)
+ if(device==TNL::Devices::HostDevice)
{
- typedef tnlHost Device;
+ typedef TNL::Devices::Host Device;
tnlParallelEikonalSolver<2,SchemeTypeHost,SchemeTypeDevice, Device> solver;
@@ -92,17 +92,17 @@ int main( int argc, char* argv[] )
else if(dim == 3)
{
- typedef parallelGodunovEikonalScheme< tnlGrid<3,double,tnlHost, int>, double, int > SchemeTypeHost;
+ typedef parallelGodunovEikonalScheme< tnlGrid<3,double,TNL::Devices::Host, int>, double, int > SchemeTypeHost;
/*#ifdef HAVE_CUDA
typedef parallelGodunovEikonalScheme< tnlGrid<2,double,tnlCuda, int>, double, int > SchemeTypeDevice;
#endif
#ifndef HAVE_CUDA*/
- typedef parallelGodunovEikonalScheme< tnlGrid<3,double,tnlHost, int>, double, int > SchemeTypeDevice;
+ typedef parallelGodunovEikonalScheme< tnlGrid<3,double,TNL::Devices::Host, int>, double, int > SchemeTypeDevice;
/*#endif*/
- if(device==tnlHostDevice)
+ if(device==TNL::Devices::HostDevice)
{
- typedef tnlHost Device;
+ typedef TNL::Devices::Host Device;
tnlParallelEikonalSolver<3,SchemeTypeHost,SchemeTypeDevice, Device> solver;
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver.h
index f79752b7187cf9de4f58299fa9add9ce72f2c407..19cdd949359d4349172af820def49169146c8717 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver.h
@@ -18,10 +18,10 @@
#define TNLPARALLELEIKONALSOLVER_H_
#include <TNL/Config/ParameterContainer.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
#include <TNL/Containers/StaticVector.h>
#include <functions/tnlMeshFunction.h>
-#include <core/tnlHost.h>
+#include <TNL/Devices/Host.h>
#include <mesh/tnlGrid.h>
#include <mesh/grids/tnlGridEntity.h>
#include <limits.h>
@@ -32,7 +32,6 @@
#include <ctime>
#ifdef HAVE_CUDA
-#include <cuda.h>
#include <core/tnlCuda.h>
#endif
@@ -54,13 +53,13 @@ public:
typedef SchemeDevice SchemeTypeDevice;
typedef SchemeHost SchemeTypeHost;
typedef Device DeviceType;
- typedef tnlVector< double, tnlHost, int > VectorType;
- typedef tnlVector< int, tnlHost, int > IntVectorType;
- typedef tnlGrid< 2, double, tnlHost, int > MeshType;
+ typedef TNL::Containers::Vector< double, TNL::Devices::Host, int > VectorType;
+ typedef TNL::Containers::Vector< int, TNL::Devices::Host, int > IntVectorType;
+ typedef tnlGrid< 2, double, TNL::Devices::Host, int > MeshType;
#ifdef HAVE_CUDA
- typedef tnlVector< double, tnlHost, int > VectorTypeCUDA;
- typedef tnlVector< int, tnlHost, int > IntVectorTypeCUDA;
- typedef tnlGrid< 2, double, tnlHost, int > MeshTypeCUDA;
+ typedef TNL::Containers::Vector< double, TNL::Devices::Host, int > VectorTypeCUDA;
+ typedef TNL::Containers::Vector< int, TNL::Devices::Host, int > IntVectorTypeCUDA;
+ typedef tnlGrid< 2, double, TNL::Devices::Host, int > MeshTypeCUDA;
#endif
tnlParallelEikonalSolver();
bool init( const Config::ParameterContainer& parameters );
@@ -161,7 +160,7 @@ public:
//__device__ bool initCUDA( tnlParallelEikonalSolver<SchemeHost, SchemeDevice, Device, double, int >* cudaSolver);
- /*__global__ void initRunCUDA(tnlParallelEikonalSolver<Scheme, double, tnlHost, int >* caller);*/
+ /*__global__ void initRunCUDA(tnlParallelEikonalSolver<Scheme, double, TNL::Devices::Host, int >* caller);*/
#endif
@@ -181,13 +180,13 @@ public:
typedef SchemeDevice SchemeTypeDevice;
typedef SchemeHost SchemeTypeHost;
typedef Device DeviceType;
- typedef tnlVector< double, tnlHost, int > VectorType;
- typedef tnlVector< int, tnlHost, int > IntVectorType;
- typedef tnlGrid< 3, double, tnlHost, int > MeshType;
+ typedef TNL::Containers::Vector< double, TNL::Devices::Host, int > VectorType;
+ typedef TNL::Containers::Vector< int, TNL::Devices::Host, int > IntVectorType;
+ typedef tnlGrid< 3, double, TNL::Devices::Host, int > MeshType;
#ifdef HAVE_CUDA
- typedef tnlVector< double, tnlHost, int > VectorTypeCUDA;
- typedef tnlVector< int, tnlHost, int > IntVectorTypeCUDA;
- typedef tnlGrid< 3, double, tnlHost, int > MeshTypeCUDA;
+ typedef TNL::Containers::Vector< double, TNL::Devices::Host, int > VectorTypeCUDA;
+ typedef TNL::Containers::Vector< int, TNL::Devices::Host, int > IntVectorTypeCUDA;
+ typedef tnlGrid< 3, double, TNL::Devices::Host, int > MeshTypeCUDA;
#endif
tnlParallelEikonalSolver();
bool init( const Config::ParameterContainer& parameters );
@@ -285,7 +284,7 @@ public:
//__device__ bool initCUDA( tnlParallelEikonalSolver<SchemeHost, SchemeDevice, Device, double, int >* cudaSolver);
- /*__global__ void initRunCUDA(tnlParallelEikonalSolver<Scheme, double, tnlHost, int >* caller);*/
+ /*__global__ void initRunCUDA(tnlParallelEikonalSolver<Scheme, double, TNL::Devices::Host, int >* caller);*/
#endif
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver2D_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver2D_impl.h
index 6279fbb805733172ebda8a78c68a128d0d33f92c..8370b069d5c96bfd7921390e46e4979131f15b36 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver2D_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver2D_impl.h
@@ -25,7 +25,7 @@ template< typename SchemeHost, typename SchemeDevice, typename Device>
tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::tnlParallelEikonalSolver()
{
cout << "a" << endl;
- this->device = tnlCudaDevice; /////////////// tnlCuda Device --- vypocet na GPU, tnlHostDevice --- vypocet na CPU
+ this->device = tnlCudaDevice; /////////////// tnlCuda Device --- vypocet na GPU, TNL::Devices::HostDevice --- vypocet na CPU
#ifdef HAVE_CUDA
if(this->device == tnlCudaDevice)
@@ -124,14 +124,14 @@ bool tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
cudaMalloc(&(this->tmpw), this->work_u.getSize()*sizeof(double));
cudaMalloc(&(this->runcuda), sizeof(int));
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
int* tmpUC;
cudaMalloc(&(tmpUC), this->work_u.getSize()*sizeof(int));
cudaMemcpy(tmpUC, this->unusedCell.getData(), this->unusedCell.getSize()*sizeof(int), cudaMemcpyHostToDevice);
initCUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<1,1>>>(this->cudaSolver, (this->tmpw), (this->runcuda),tmpUC);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cout << "s " << endl;
//cudaMalloc(&(cudaSolver->work_u_cuda), this->work_u.getSize()*sizeof(double));
double* tmpu = NULL;
@@ -140,13 +140,13 @@ bool tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
//printf("%p %p \n",tmpu,tmpw);
cudaMemcpy((this->tmpw), this->work_u.getData(), this->work_u.getSize()*sizeof(double), cudaMemcpyHostToDevice);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cout << "s "<< endl;
}
#endif
- if(this->device == tnlHostDevice)
+ if(this->device == TNL::Devices::HostDevice)
{
for(int i = 0; i < this->subgridValues.getSize(); i++)
{
@@ -182,11 +182,11 @@ bool tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
{
// cout << "pre 1 kernel" << endl;
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 threadsPerBlock(this->n, this->n);
dim3 numBlocks(this->gridCols,this->gridRows);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initRunCUDA2D<SchemeTypeHost,SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock,3*this->n*this->n*sizeof(double)>>>(this->cudaSolver);
cudaDeviceSynchronize();
// cout << "post 1 kernel" << endl;
@@ -196,7 +196,7 @@ bool tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
this->currentStep = 1;
- if(this->device == tnlHostDevice)
+ if(this->device == TNL::Devices::HostDevice)
synchronize();
#ifdef HAVE_CUDA
else if(this->device == tnlCudaDevice)
@@ -208,17 +208,17 @@ bool tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
//cudaMemcpy(/*this->work_u.getData()*/ test, (this->tmpw), this->work_u.getSize()*sizeof(double), cudaMemcpyDeviceToHost);
//cout << this->tmpw << " " << test[0] <<" " << test[1] << " " <<test[2] << " " <<test[3] << endl;
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
synchronizeCUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
synchronize2CUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cout << test[0] << " " <<test[1] <<" " << test[2] << " " <<test[3] << endl;
//cudaMemcpy(/*this->work_u.getData()*/ test, (this->tmpw), this->work_u.getSize()*sizeof(double), cudaMemcpyDeviceToHost);
- //checkCudaDevice;
+ //TNL_CHECK_CUDA_DEVICE;
//cout << this->tmpw << " " << test[0] << " " <<test[1] << " " <<test[2] <<" " << test[3] << endl;
//free(test);
@@ -233,7 +233,7 @@ bool tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
template< typename SchemeHost, typename SchemeDevice, typename Device>
void tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::run()
{
- if(this->device == tnlHostDevice)
+ if(this->device == TNL::Devices::HostDevice)
{
bool end = false;
@@ -343,17 +343,17 @@ void tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
dim3 threadsPerBlock(this->n, this->n);
dim3 numBlocks(this->gridCols,this->gridRows);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cudaMalloc(&runcuda,sizeof(bool));
//cudaMemcpy(runcuda, &run_host, sizeof(bool), cudaMemcpyHostToDevice);
//cout << "fn" << endl;
bool* tmpb;
//cudaMemcpy(tmpb, &(cudaSolver->runcuda),sizeof(bool*), cudaMemcpyDeviceToHost);
//cudaDeviceSynchronize();
- //checkCudaDevice;
+ //TNL_CHECK_CUDA_DEVICE;
cudaMemcpy(&(this->run_host),this->runcuda,sizeof(int), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cout << "fn" << endl;
int i = 1;
time_diff = 0.0;
@@ -366,7 +366,7 @@ void tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
end_cuda = false;
//cout << "a" << endl;
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
start = std::clock();
runCUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock,3*this->n*this->n*sizeof(double)>>>(this->cudaSolver);
//cout << "a" << endl;
@@ -376,10 +376,10 @@ void tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
//start = std::clock();
synchronizeCUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
synchronize2CUDA2D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//time_diff += (std::clock() - start) / (double)(CLOCKS_PER_SEC);
@@ -975,7 +975,7 @@ tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::runSu
double maxResidue( 1.0 );
//double lastResidue( 10000.0 );
tnlGridEntity<MeshType, 2, tnlGridEntityNoStencilStorage > Entity(subMesh);
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
while( time < finalTime /*|| maxResidue > subMesh.template getSpaceStepsProducts< 1, 0 >()*/)
{
/****
@@ -986,8 +986,8 @@ tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::runSu
{
Entity.setCoordinates(Containers::StaticVector<2,int>(i % subMesh.getDimensions().x(),i / subMesh.getDimensions().x()));
Entity.refresh();
- neighbourEntities.refresh(subMesh,Entity.getIndex());
- fu[ i ] = schemeHost.getValue( this->subMesh, i, Containers::StaticVector<2,int>(i % subMesh.getDimensions().x(),i / subMesh.getDimensions().x()), u, time, boundaryCondition,neighbourEntities);
+ neighborEntities.refresh(subMesh,Entity.getIndex());
+ fu[ i ] = schemeHost.getValue( this->subMesh, i, Containers::StaticVector<2,int>(i % subMesh.getDimensions().x(),i / subMesh.getDimensions().x()), u, time, boundaryCondition,neighborEntities);
}
maxResidue = fu. absMax();
@@ -1188,16 +1188,16 @@ void tnlParallelEikonalSolver<2,SchemeHost, SchemeDevice, Device, double, int>::
// if( time + currentTau > finalTime ) currentTau = finalTime - time;
tnlGridEntity<MeshType, 2, tnlGridEntityNoStencilStorage > Entity(subMesh);
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 2, tnlGridEntityNoStencilStorage >,2> neighborEntities(Entity);
Entity.setCoordinates(Containers::StaticVector<2,int>(i,j));
Entity.refresh();
- neighbourEntities.refresh(subMesh,Entity.getIndex());
+ neighborEntities.refresh(subMesh,Entity.getIndex());
while( time < finalTime )
{
if(computeFU)
- fu = schemeHost.getValueDev( this->subMesh, l, Containers::StaticVector<2,int>(i,j)/*this->subMesh.getCellCoordinates(l)*/, u, time, boundaryCondition, neighbourEntities);
+ fu = schemeHost.getValueDev( this->subMesh, l, Containers::StaticVector<2,int>(i,j)/*this->subMesh.getCellCoordinates(l)*/, u, time, boundaryCondition, neighborEntities);
sharedTau[l]=abs(cfl/fu);
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver3D_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver3D_impl.h
index 12d9003099643c923fdd6f4be18f2b07b35ebf3d..b0871824cbf67f7ef18d276903ec84d8eeb0a109 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver3D_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi-parallel/tnlParallelEikonalSolver3D_impl.h
@@ -25,7 +25,7 @@ template< typename SchemeHost, typename SchemeDevice, typename Device>
tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::tnlParallelEikonalSolver()
{
cout << "a" << endl;
- this->device = tnlHostDevice; /////////////// tnlCuda Device --- vypocet na GPU, tnlHostDevice --- vypocet na CPU
+ this->device = TNL::Devices::HostDevice; /////////////// tnlCuda Device --- vypocet na GPU, TNL::Devices::HostDevice --- vypocet na CPU
#ifdef HAVE_CUDA
if(this->device == tnlCudaDevice)
@@ -124,14 +124,14 @@ bool tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
cudaMalloc(&(this->tmpw), this->work_u.getSize()*sizeof(double));
cudaMalloc(&(this->runcuda), sizeof(int));
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
int* tmpUC;
cudaMalloc(&(tmpUC), this->work_u.getSize()*sizeof(int));
cudaMemcpy(tmpUC, this->unusedCell.getData(), this->unusedCell.getSize()*sizeof(int), cudaMemcpyHostToDevice);
initCUDA3D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<1,1>>>(this->cudaSolver, (this->tmpw), (this->runcuda),tmpUC);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cout << "s " << endl;
//cudaMalloc(&(cudaSolver->work_u_cuda), this->work_u.getSize()*sizeof(double));
double* tmpu = NULL;
@@ -140,13 +140,13 @@ bool tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
//printf("%p %p \n",tmpu,tmpw);
cudaMemcpy((this->tmpw), this->work_u.getData(), this->work_u.getSize()*sizeof(double), cudaMemcpyHostToDevice);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cout << "s "<< endl;
}
#endif
- if(this->device == tnlHostDevice)
+ if(this->device == TNL::Devices::HostDevice)
{
#ifdef HAVE_OPENMP
#pragma omp parallel for num_threads(4) schedule(dynamic)
@@ -195,11 +195,11 @@ bool tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
{
// cout << "pre 1 kernel" << endl;
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 threadsPerBlock(this->n, this->n, this->n);
dim3 numBlocks(this->gridCols,this->gridRows,this->gridLevels);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
initRunCUDA3D<SchemeTypeHost,SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock,2*this->n*this->n*this->n*sizeof(double)>>>(this->cudaSolver);
cudaDeviceSynchronize();
// cout << "post 1 kernel" << endl;
@@ -209,7 +209,7 @@ bool tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
this->currentStep = 1;
- if(this->device == tnlHostDevice)
+ if(this->device == TNL::Devices::HostDevice)
synchronize();
#ifdef HAVE_CUDA
else if(this->device == tnlCudaDevice)
@@ -221,17 +221,17 @@ bool tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
//cudaMemcpy(/*this->work_u.getData()*/ test, (this->tmpw), this->work_u.getSize()*sizeof(double), cudaMemcpyDeviceToHost);
//cout << this->tmpw << " " << test[0] <<" " << test[1] << " " <<test[2] << " " <<test[3] << endl;
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
synchronizeCUDA3D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cout << cudaGetErrorString(cudaDeviceSynchronize()) << endl;
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
synchronize2CUDA3D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cout << test[0] << " " <<test[1] <<" " << test[2] << " " <<test[3] << endl;
//cudaMemcpy(/*this->work_u.getData()*/ test, (this->tmpw), this->work_u.getSize()*sizeof(double), cudaMemcpyDeviceToHost);
- //checkCudaDevice;
+ //TNL_CHECK_CUDA_DEVICE;
//cout << this->tmpw << " " << test[0] << " " <<test[1] << " " <<test[2] <<" " << test[3] << endl;
//free(test);
@@ -246,7 +246,7 @@ bool tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
template< typename SchemeHost, typename SchemeDevice, typename Device>
void tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::run()
{
- if(this->device == tnlHostDevice)
+ if(this->device == TNL::Devices::HostDevice)
{
bool end = false;
@@ -381,17 +381,17 @@ void tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
dim3 threadsPerBlock(this->n, this->n, this->n);
dim3 numBlocks(this->gridCols,this->gridRows,this->gridLevels);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cudaMalloc(&runcuda,sizeof(bool));
//cudaMemcpy(runcuda, &run_host, sizeof(bool), cudaMemcpyHostToDevice);
//cout << "fn" << endl;
bool* tmpb;
//cudaMemcpy(tmpb, &(cudaSolver->runcuda),sizeof(bool*), cudaMemcpyDeviceToHost);
//cudaDeviceSynchronize();
- //checkCudaDevice;
+ //TNL_CHECK_CUDA_DEVICE;
cudaMemcpy(&(this->run_host),this->runcuda,sizeof(int), cudaMemcpyDeviceToHost);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//cout << "fn" << endl;
int i = 1;
time_diff = 0.0;
@@ -404,7 +404,7 @@ void tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
end_cuda = false;
//cout << "a" << endl;
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
start = std::clock();
runCUDA3D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock,2*this->n*this->n*this->n*sizeof(double)>>>(this->cudaSolver);
//cout << "a" << endl;
@@ -414,10 +414,10 @@ void tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
//start = std::clock();
synchronizeCUDA3D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,threadsPerBlock>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
synchronize2CUDA3D<SchemeTypeHost, SchemeTypeDevice, DeviceType><<<numBlocks,1>>>(this->cudaSolver);
cudaDeviceSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//time_diff += (std::clock() - start) / (double)(CLOCKS_PER_SEC);
@@ -934,7 +934,7 @@ tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::runSu
double maxResidue( 1.0 );
//double lastResidue( 10000.0 );
tnlGridEntity<MeshType, 3, tnlGridEntityNoStencilStorage > Entity(subMesh);
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities(Entity);
while( time < finalTime /*|| maxResidue > subMesh.template getSpaceStepsProducts< 1, 0, 0 >()*/)
{
/****
@@ -952,9 +952,9 @@ tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::runSu
// cout <<"c" << coords << endl;
Entity.refresh();
// cout << "d" <<endl;
- neighbourEntities.refresh(subMesh,Entity.getIndex());
+ neighborEntities.refresh(subMesh,Entity.getIndex());
// cout << "e" <<endl;
- fu[ i ] = schemeHost.getValue( this->subMesh, i, coords,u, time, boundaryCondition, neighbourEntities );
+ fu[ i ] = schemeHost.getValue( this->subMesh, i, coords,u, time, boundaryCondition, neighborEntities );
// cout << "f" <<endl;
}
maxResidue = fu. absMax();
@@ -1202,10 +1202,10 @@ void tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
if( boundaryCondition == 0 ) finalTime *= 2.0;
tnlGridEntity<MeshType, 3, tnlGridEntityNoStencilStorage > Entity(subMesh);
- tnlNeighbourGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighbourEntities(Entity);
+ tnlNeighborGridEntityGetter<tnlGridEntity< MeshType, 3, tnlGridEntityNoStencilStorage >,3> neighborEntities(Entity);
Entity.setCoordinates(Containers::StaticVector<3,int>(i,j,k));
Entity.refresh();
- neighbourEntities.refresh(subMesh,Entity.getIndex());
+ neighborEntities.refresh(subMesh,Entity.getIndex());
while( time < finalTime )
@@ -1214,7 +1214,7 @@ void tnlParallelEikonalSolver<3,SchemeHost, SchemeDevice, Device, double, int>::
if(computeFU)
{
- fu = schemeHost.getValueDev( this->subMesh, l, Containers::StaticVector<3,int>(i,j,k), u, time, boundaryCondition, neighbourEntities);
+ fu = schemeHost.getValueDev( this->subMesh, l, Containers::StaticVector<3,int>(i,j,k), u, time, boundaryCondition, neighborEntities);
if(abs(fu) > 0.0)
sharedTau[l]=abs(cfl/fu);
}
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/CMakeLists.txt b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/HamiltonJacobiProblem.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/HamiltonJacobiProblem.h
index a4a187cee17f9d8f8c43537347208d1caf10bc56..0523f7e90e1ccd0cef6683ffcff34b324305eee7 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/HamiltonJacobiProblem.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/HamiltonJacobiProblem.h
@@ -20,7 +20,7 @@
#include <solvers/preconditioners/tnlDummyPreconditioner.h>
#include <solvers/tnlSolverMonitor.h>
#include <core/tnlLogger.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
#include <solvers/pde/tnlExplicitUpdater.h>
#include <solvers/pde/tnlLinearSystemAssembler.h>
#include <functions/tnlMeshFunction.h>
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/tnlDirectEikonalMethodsBase_impl.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/tnlDirectEikonalMethodsBase_impl.h
index c2983fac2c12e1b7791bd86973a712a5fcc6c948..b924f72331e7126f3d8accc7392e9e7554827890 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/tnlDirectEikonalMethodsBase_impl.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/tnlDirectEikonalMethodsBase_impl.h
@@ -30,10 +30,10 @@ initInterface( const MeshFunctionType& input,
const RealType& c = input( cell );
if( ! cell.isBoundaryEntity() )
{
- const auto& neighbours = cell.getNeighbourEntities();
+ const auto& neighbors = cell.getNeighborEntities();
//const IndexType& c = cell.getIndex();
- const IndexType e = neighbours.template getEntityIndex< 1 >();
- const IndexType w = neighbours.template getEntityIndex< -1 >();
+ const IndexType e = neighbors.template getEntityIndex< 1 >();
+ const IndexType w = neighbors.template getEntityIndex< -1 >();
if( c * input[ e ] <= 0 || c * input[ w ] <= 0 )
{
@@ -84,11 +84,11 @@ initInterface( const MeshFunctionType& input,
const RealType& c = input( cell );
if( ! cell.isBoundaryEntity() )
{
- auto neighbours = cell.getNeighbourEntities();
- const IndexType e = neighbours.template getEntityIndex< 1, 0 >();
- const IndexType w = neighbours.template getEntityIndex< -1, 0 >();
- const IndexType n = neighbours.template getEntityIndex< 0, 1 >();
- const IndexType s = neighbours.template getEntityIndex< 0, -1 >();
+ auto neighbors = cell.getNeighborEntities();
+ const IndexType e = neighbors.template getEntityIndex< 1, 0 >();
+ const IndexType w = neighbors.template getEntityIndex< -1, 0 >();
+ const IndexType n = neighbors.template getEntityIndex< 0, 1 >();
+ const IndexType s = neighbors.template getEntityIndex< 0, -1 >();
if( c * input[ e ] <= 0 || c * input[ w ] <= 0 ||
c * input[ n ] <= 0 || c * input[ s ] <= 0 )
{
@@ -113,7 +113,7 @@ tnlDirectEikonalMethodsBase< tnlGrid< 2, Real, Device, Index > >::
updateCell( MeshFunctionType& u,
const MeshEntity& cell )
{
- const auto& neighbourEntities = cell.template getNeighbourEntities< 2 >();
+ const auto& neighborEntities = cell.template getNeighborEntities< 2 >();
const MeshType& mesh = cell.getMesh();
const RealType& h = mesh.getSpaceSteps().x();
@@ -121,23 +121,23 @@ updateCell( MeshFunctionType& u,
Real a, b, tmp;
if( cell.getCoordinates().x() == 0 )
- a = u[ neighbourEntities.template getEntityIndex< 1, 0 >() ];
+ a = u[ neighborEntities.template getEntityIndex< 1, 0 >() ];
else if( cell.getCoordinates().x() == mesh.getDimensions().x() - 1 )
- a = u[ neighbourEntities.template getEntityIndex< -1, 0 >() ];
+ a = u[ neighborEntities.template getEntityIndex< -1, 0 >() ];
else
{
- a = ArgAbsMin( u[ neighbourEntities.template getEntityIndex< -1, 0 >() ],
- u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] );
+ a = ArgAbsMin( u[ neighborEntities.template getEntityIndex< -1, 0 >() ],
+ u[ neighborEntities.template getEntityIndex< 1, 0 >() ] );
}
if( cell.getCoordinates().y() == 0 )
- b = u[ neighbourEntities.template getEntityIndex< 0, 1 >()];
+ b = u[ neighborEntities.template getEntityIndex< 0, 1 >()];
else if( cell.getCoordinates().y() == mesh.getDimensions().y() - 1 )
- b = u[ neighbourEntities.template getEntityIndex< 0, -1 >() ];
+ b = u[ neighborEntities.template getEntityIndex< 0, -1 >() ];
else
{
- b = ArgAbsMin( u[ neighbourEntities.template getEntityIndex< 0, -1 >() ],
- u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] );
+ b = ArgAbsMin( u[ neighborEntities.template getEntityIndex< 0, -1 >() ],
+ u[ neighborEntities.template getEntityIndex< 0, 1 >() ] );
}
if( fabs( a ) == tnlTypeInfo< Real >::getMaxValue() &&
@@ -195,14 +195,14 @@ initInterface( const MeshFunctionType& input,
const RealType& c = input( cell );
if( ! cell.isBoundaryEntity() )
{
- auto neighbours = cell.getNeighbourEntities();
+ auto neighbors = cell.getNeighborEntities();
//const IndexType& c = cell.getIndex();
- const IndexType e = neighbours.template getEntityIndex< 1, 0, 0 >();
- const IndexType w = neighbours.template getEntityIndex< -1, 0, 0 >();
- const IndexType n = neighbours.template getEntityIndex< 0, 1, 0 >();
- const IndexType s = neighbours.template getEntityIndex< 0, -1, 0 >();
- const IndexType t = neighbours.template getEntityIndex< 0, 0, 1 >();
- const IndexType b = neighbours.template getEntityIndex< 0, 0, -1 >();
+ const IndexType e = neighbors.template getEntityIndex< 1, 0, 0 >();
+ const IndexType w = neighbors.template getEntityIndex< -1, 0, 0 >();
+ const IndexType n = neighbors.template getEntityIndex< 0, 1, 0 >();
+ const IndexType s = neighbors.template getEntityIndex< 0, -1, 0 >();
+ const IndexType t = neighbors.template getEntityIndex< 0, 0, 1 >();
+ const IndexType b = neighbors.template getEntityIndex< 0, 0, -1 >();
if( c * input[ e ] <= 0 || c * input[ w ] <= 0 ||
c * input[ n ] <= 0 || c * input[ s ] <= 0 ||
diff --git a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/tnlFastSweepingMethod.h b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/tnlFastSweepingMethod.h
index 671448e5bfb282d3f9b84a6c3c71e99e8c403432..59ddc2a690b5e6aecb62d5cf4c0997b7c567ff82 100644
--- a/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/tnlFastSweepingMethod.h
+++ b/src/TNL/Experimental/Hamilton-Jacobi/Solvers/hamilton-jacobi/tnlFastSweepingMethod.h
@@ -30,13 +30,13 @@ template< typename Real,
class tnlFastSweepingMethod< tnlGrid< 1, Real, Device, Index >, Anisotropy >
: public tnlDirectEikonalMethodsBase< tnlGrid< 1, Real, Device, Index > >
{
- static_assert( std::is_same< Device, tnlHost >::value, "The fast sweeping method works only on CPU." );
+ static_assert( std::is_same< Device, TNL::Devices::Host >::value, "The fast sweeping method works only on CPU." );
public:
typedef tnlGrid< 1, Real, Device, Index > MeshType;
typedef Real RealType;
- typedef tnlHost DeviceType;
+ typedef TNL::Devices::Host DeviceType;
typedef Index IndexType;
typedef Anisotropy AnisotropyType;
typedef tnlDirectEikonalMethodsBase< tnlGrid< 1, Real, Device, Index > > BaseType;
@@ -67,13 +67,13 @@ template< typename Real,
class tnlFastSweepingMethod< tnlGrid< 2, Real, Device, Index >, Anisotropy >
: public tnlDirectEikonalMethodsBase< tnlGrid< 2, Real, Device, Index > >
{
- static_assert( std::is_same< Device, tnlHost >::value, "The fast sweeping method works only on CPU." );
+ static_assert( std::is_same< Device, TNL::Devices::Host >::value, "The fast sweeping method works only on CPU." );
public:
typedef tnlGrid< 2, Real, Device, Index > MeshType;
typedef Real RealType;
- typedef tnlHost DeviceType;
+ typedef TNL::Devices::Host DeviceType;
typedef Index IndexType;
typedef Anisotropy AnisotropyType;
typedef tnlDirectEikonalMethodsBase< tnlGrid< 2, Real, Device, Index > > BaseType;
@@ -104,13 +104,13 @@ template< typename Real,
class tnlFastSweepingMethod< tnlGrid< 3, Real, Device, Index >, Anisotropy >
: public tnlDirectEikonalMethodsBase< tnlGrid< 3, Real, Device, Index > >
{
- static_assert( std::is_same< Device, tnlHost >::value, "The fast sweeping method works only on CPU." );
+ static_assert( std::is_same< Device, TNL::Devices::Host >::value, "The fast sweeping method works only on CPU." );
public:
typedef tnlGrid< 3, Real, Device, Index > MeshType;
typedef Real RealType;
- typedef tnlHost DeviceType;
+ typedef TNL::Devices::Host DeviceType;
typedef Index IndexType;
typedef Anisotropy AnisotropyType;
typedef tnlDirectEikonalMethodsBase< tnlGrid< 3, Real, Device, Index > > BaseType;
diff --git a/src/TNL/File.cpp b/src/TNL/File.cpp
index d6e16b17ac89c0d0cd5c9b3f262f47b7eb2ad28c..885173cd40a320f67d149f5b1be6193012059600 100644
--- a/src/TNL/File.cpp
+++ b/src/TNL/File.cpp
@@ -15,7 +15,7 @@ namespace TNL {
int File :: verbose = 0;
File :: File()
-: mode( tnlUndefinedMode ),
+: mode( IOMode::undefined ),
file( NULL ),
fileOK( false ),
writtenElements( 0 ),
@@ -23,26 +23,37 @@ File :: File()
{
}
+File :: ~File()
+{
+ // destroying a file without closing is a memory leak
+ // (an open file descriptor is left behind, on Linux there is typically
+ // only a limited number of descriptors available to each process)
+ close();
+}
+
bool File :: open( const String& fileName,
- const tnlIOMode mode )
+ const IOMode mode )
{
+ // close the existing file to avoid memory leaks
+ this->close();
+
this->fileName = fileName;
if( verbose )
{
std::cout << "Opening file " << fileName;
- if( mode == tnlReadMode )
+ if( mode == IOMode::read )
std::cout << " for reading... " << std::endl;
else
std::cout << " for writing ... " << std::endl;
}
- if( mode == tnlReadMode )
- file = fopen( fileName. getString(), "r" );
- if( mode == tnlWriteMode )
- file = fopen( fileName. getString(), "w" );
+ if( mode == IOMode::read )
+ file = std::fopen( fileName.getString(), "rb" );
+ if( mode == IOMode::write )
+ file = std::fopen( fileName.getString(), "wb" );
if( file == NULL )
{
std::cerr << "I am not able to open the file " << fileName << ". ";
- perror( "" );
+ std::perror( "" );
return false;
}
this->fileOK = true;
@@ -55,24 +66,25 @@ bool File :: close()
if( verbose )
std::cout << "Closing the file " << getFileName() << " ... " << std::endl;
- if( fclose( file ) != 0 )
+ if( file && std::fclose( file ) != 0 )
{
std::cerr << "I was not able to close the file " << fileName << " properly!" << std::endl;
return false;
}
+ // reset all attributes
+ mode = IOMode::undefined;
+ file = NULL;
+ fileOK = false;
+ fileName = "";
readElements = writtenElements = 0;
return true;
-};
+}
bool fileExists( const String& fileName )
{
std::fstream file;
- file.open( fileName. getString(), std::ios::in );
- bool result( true );
- if( ! file )
- result = false;
- file.close();
- return result;
-};
+ file.open( fileName.getString(), std::ios::in );
+ return ! file.fail();
+}
} // namespace TNL
diff --git a/src/TNL/File.h b/src/TNL/File.h
index 1eea42eae61b67867bfeb7218b0a1801932f2c14..78861f5690a86901b170e0056830fe59250faf6e 100644
--- a/src/TNL/File.h
+++ b/src/TNL/File.h
@@ -12,22 +12,22 @@
#include <iostream>
#include <fstream>
-#include <stdio.h>
-#include <stdlib.h>
-#ifdef HAVE_CUDA
- #include <cuda_runtime.h>
-#endif
+#include <cstdio>
#include <TNL/Assert.h>
#include <TNL/String.h>
#include <TNL/Devices/Host.h>
#include <TNL/Devices/Cuda.h>
+#include <TNL/Devices/MIC.h>
namespace TNL {
-enum tnlIOMode { tnlUndefinedMode = 0,
- tnlReadMode = 1,
- tnlWriteMode = 2 };
+enum class IOMode
+{
+ undefined = 0,
+ read = 1,
+ write = 2
+};
/* When we need to transfer data between the GPU and the CPU we use
* 5 MB buffer. This size should ensure good performance -- see.
@@ -41,57 +41,43 @@ const size_t tnlFileGPUvsCPUTransferBufferSize = 5 * 2<<20;
*/
class File
{
- tnlIOMode mode;
+ IOMode mode;
- FILE* file;
+ std::FILE* file;
bool fileOK;
String fileName;
- size_t writtenElements;
+ std::size_t writtenElements;
- size_t readElements;
+ std::size_t readElements;
public:
File();
+ ~File();
+
bool open( const String& fileName,
- const tnlIOMode mode );
+ const IOMode mode );
- const String& getFileName() const
+ const String& getFileName() const
{
- return this->fileName;
+ return this->fileName;
}
- long int getReadElements() const
- {
- return this->readElements;
- }
-
- long int getWrittenElements() const
- {
- return this->writtenElements;
- }
-
- // TODO: this does not work for constant types
-#ifdef HAVE_NOT_CXX11
- template< typename Type, typename Device, typename Index >
- bool read( Type* buffer,
- const Index& elements );
-
- template< typename Type, typename Device >
- bool read( Type* buffer );
+ long int getReadElements() const
+ {
+ return this->readElements;
+ }
- template< typename Type, typename Device, typename Index >
- bool write( const Type* buffer,
- const Index elements );
+ long int getWrittenElements() const
+ {
+ return this->writtenElements;
+ }
- template< typename Type, typename Device >
- bool write( const Type* buffer );
-#else
template< typename Type, typename Device = Devices::Host, typename Index = int >
bool read( Type* buffer,
const Index& elements );
@@ -106,12 +92,52 @@ class File
template< typename Type, typename Device = Devices::Host >
bool write( const Type* buffer );
-#endif
-
- bool close();
-
- static int verbose;
-
+ bool close();
+
+ static int verbose;
+
+protected:
+ template< typename Type,
+ typename Device,
+ typename = typename std::enable_if< std::is_same< Device, Devices::Host >::value >::type >
+ bool read_impl( Type* buffer,
+ const std::size_t& elements );
+
+ template< typename Type,
+ typename Device,
+ typename = typename std::enable_if< std::is_same< Device, Devices::Cuda >::value >::type,
+ typename = void >
+ bool read_impl( Type* buffer,
+ const std::size_t& elements );
+
+ template< typename Type,
+ typename Device,
+ typename = typename std::enable_if< std::is_same< Device, Devices::MIC >::value >::type,
+ typename = void,
+ typename = void >
+ bool read_impl( Type* buffer,
+ const std::size_t& elements );
+
+ template< typename Type,
+ typename Device,
+ typename = typename std::enable_if< std::is_same< Device, Devices::Host >::value >::type >
+ bool write_impl( const Type* buffer,
+ const std::size_t& elements );
+
+ template< typename Type,
+ typename Device,
+ typename = typename std::enable_if< std::is_same< Device, Devices::Cuda >::value >::type,
+ typename = void >
+ bool write_impl( const Type* buffer,
+ const std::size_t& elements );
+
+ template< typename Type,
+ typename Device,
+ typename = typename std::enable_if< std::is_same< Device, Devices::MIC >::value >::type,
+ typename = void,
+ typename = void >
+ bool write_impl( const Type* buffer,
+ const std::size_t& elements );
};
bool fileExists( const String& fileName );
diff --git a/src/TNL/File_impl.h b/src/TNL/File_impl.h
index a6b746f4880f7ebea1e52aee689c7d4b6fd21fcb..932399045f9e2b9990a6272788bbcd8a6f685aea 100644
--- a/src/TNL/File_impl.h
+++ b/src/TNL/File_impl.h
@@ -8,33 +8,38 @@
/* See Copyright Notice in tnl/Copyright */
-#pragma once
+#pragma once
+
+#include <type_traits>
+
+#include <TNL/File.h>
+#include <TNL/Exceptions/CudaSupportMissing.h>
+#include <TNL/Exceptions/MICSupportMissing.h>
namespace TNL {
+
template< typename Type, typename Device >
-bool File :: read( Type* buffer )
+bool File::read( Type* buffer )
{
return read< Type, Device, int >( buffer, 1 );
-};
+}
template< typename Type, typename Device >
-bool File :: write( const Type* buffer )
+bool File::write( const Type* buffer )
{
return write< Type, Device, int >( buffer, 1 );
-};
-
+}
template< typename Type, typename Device, typename Index >
-bool File :: read( Type* buffer,
- const Index& _elements )
+bool File::read( Type* buffer,
+ const Index& _elements )
{
- TNL_ASSERT( _elements >= 0,
- std::cerr << " elements = " << _elements << std::endl; );
+ TNL_ASSERT_GE( _elements, 0, "Number of elements to read must be non-negative." );
- // convert _elements from Index to size_t, which is *unsigned* type
+ // convert _elements from Index to std::size_t, which is *unsigned* type
// (expected by fread etc)
- size_t elements = (size_t) _elements;
+ std::size_t elements = (std::size_t) _elements;
if( ! elements )
return true;
@@ -43,95 +48,141 @@ bool File :: read( Type* buffer,
std::cerr << "File " << fileName << " was not properly opened. " << std::endl;
return false;
}
- if( mode != tnlReadMode )
+ if( mode != IOMode::read )
{
std::cerr << "File " << fileName << " was not opened for reading. " << std::endl;
return false;
}
+
+ return read_impl< Type, Device >( buffer, elements );
+}
+
+// Host
+template< typename Type,
+ typename Device,
+ typename >
+bool File::read_impl( Type* buffer,
+ const std::size_t& elements )
+{
this->readElements = 0;
- const size_t host_buffer_size = std::min( tnlFileGPUvsCPUTransferBufferSize / sizeof( Type ),
- elements );
- void* host_buffer( 0 );
- if( std::is_same< Device, Devices::Host >::value )
+ if( std::fread( buffer,
+ sizeof( Type ),
+ elements,
+ file ) != elements )
{
- if( fread( buffer,
- sizeof( Type ),
- elements,
- file ) != elements )
+ std::cerr << "I am not able to read the data from the file " << fileName << "." << std::endl;
+ std::perror( "Fread ended with the error code" );
+ return false;
+ }
+ this->readElements = elements;
+ return true;
+}
+
+// Cuda
+template< typename Type,
+ typename Device,
+ typename, typename >
+bool File::read_impl( Type* buffer,
+ const std::size_t& elements )
+{
+#ifdef HAVE_CUDA
+ this->readElements = 0;
+ const std::size_t host_buffer_size = std::min( tnlFileGPUvsCPUTransferBufferSize / sizeof( Type ), elements );
+ using BaseType = typename std::remove_cv< Type >::type;
+ BaseType* host_buffer = new BaseType[ host_buffer_size ];
+
+ while( readElements < elements )
+ {
+ std::size_t transfer = std::min( elements - readElements, host_buffer_size );
+ std::size_t transfered = std::fread( host_buffer, sizeof( Type ), transfer, file );
+ if( transfered != transfer )
{
std::cerr << "I am not able to read the data from the file " << fileName << "." << std::endl;
- perror( "Fread ended with the error code" );
+ std::cerr << transfered << " bytes were transfered. " << std::endl;
+ std::perror( "Fread ended with the error code" );
+ delete[] host_buffer;
return false;
}
- this->readElements = elements;
- return true;
+
+ cudaMemcpy( ( void* ) & ( buffer[ readElements ] ),
+ host_buffer,
+ transfer * sizeof( Type ),
+ cudaMemcpyHostToDevice );
+ if( ! TNL_CHECK_CUDA_DEVICE )
+ {
+ std::cerr << "Transfer of data from the CUDA device to the file " << this->fileName
+ << " failed." << std::endl;
+ delete[] host_buffer;
+ return false;
+ }
+ this->readElements += transfer;
}
- if( std::is_same< Device, Devices::Cuda >::value )
+ delete[] host_buffer;
+ return true;
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
+}
+
+// MIC
+template< typename Type,
+ typename Device,
+ typename, typename, typename >
+bool File::read_impl( Type* buffer,
+ const std::size_t& elements )
+{
+#ifdef HAVE_MIC
+ this->readElements = 0;
+ const std::size_t host_buffer_size = std::min( tnlFileGPUvsCPUTransferBufferSize / sizeof( Type ), elements );
+ Type * host_buffer = (Type *)malloc( sizeof( Type ) * host_buffer_size );
+ readElements = 0;
+ if( ! host_buffer )
{
-#ifdef HAVE_CUDA
- /*!***
- * Here we cannot use
- *
- * host_buffer = new Type[ host_buffer_size ];
- *
- * because it does not work for constant types like
- * T = const bool.
- */
- host_buffer = malloc( sizeof( Type ) * host_buffer_size );
- readElements = 0;
- if( ! host_buffer )
+ std::cerr << "I am sorry but I cannot allocate supporting buffer on the host for writing data from the GPU to the file "
+ << this->getFileName() << "." << std::endl;
+ return false;
+ }
+
+ while( readElements < elements )
+ {
+ int transfer = std::min( elements - readElements , host_buffer_size );
+ size_t transfered = fread( host_buffer, sizeof( Type ), transfer, file );
+ if( transfered != transfer )
{
- std::cerr << "I am sorry but I cannot allocate supporting buffer on the host for writing data from the GPU to the file "
- << this->getFileName() << "." << std::endl;
+ std::cerr << "I am not able to read the data from the file " << fileName << "." << std::endl;
+ std::cerr << transfered << " bytes were transfered. " << std::endl;
+ perror( "Fread ended with the error code" );
return false;
-
}
-
- while( readElements < elements )
+ Devices::MICHider<Type> device_buff;
+ device_buff.pointer=buffer;
+ #pragma offload target(mic) in(device_buff,readElements) in(host_buffer:length(transfer))
{
- size_t transfer = std::min( elements - readElements, host_buffer_size );
- size_t transfered = fread( host_buffer, sizeof( Type ), transfer, file );
- if( transfered != transfer )
- {
- std::cerr << "I am not able to read the data from the file " << fileName << "." << std::endl;
- std::cerr << transfered << " bytes were transfered. " << std::endl;
- perror( "Fread ended with the error code" );
- return false;
- }
-
- cudaMemcpy( ( void* ) & ( buffer[ readElements ] ),
- host_buffer,
- transfer * sizeof( Type ),
- cudaMemcpyHostToDevice );
- if( ! checkCudaDevice )
- {
- std::cerr << "Transfer of data from the CUDA device to the file " << this->fileName
- << " failed." << std::endl;
- free( host_buffer );
- return false;
- }
- readElements += transfer;
+ /*
+ for(int i=0;i<transfer;i++)
+ device_buff.pointer[readElements+i]=host_buffer[i];
+ */
+ memcpy(&(device_buff.pointer[readElements]),host_buffer, transfer*sizeof(Type) );
}
- free( host_buffer );
- return true;
-#else
- CudaSupportMissingMessage;;
- return false;
-#endif
+
+ readElements += transfer;
}
+ free( host_buffer );
return true;
-};
+#else
+ throw Exceptions::MICSupportMissing();
+#endif
+}
template< class Type, typename Device, typename Index >
-bool File :: write( const Type* buffer,
- const Index _elements )
+bool File::write( const Type* buffer,
+ const Index _elements )
{
- TNL_ASSERT( _elements >= 0,
- std::cerr << " elements = " << _elements << std::endl; );
+ TNL_ASSERT_GE( _elements, 0, "Number of elements to write must be non-negative." );
- // convert _elements from Index to size_t, which is *unsigned* type
+ // convert _elements from Index to std::size_t, which is *unsigned* type
// (expected by fread etc)
- size_t elements = (size_t) _elements;
+ std::size_t elements = (std::size_t) _elements;
if( ! elements )
return true;
@@ -140,85 +191,134 @@ bool File :: write( const Type* buffer,
std::cerr << "File " << fileName << " was not properly opened. " << std::endl;
return false;
}
- if( mode != tnlWriteMode )
+ if( mode != IOMode::write )
{
std::cerr << "File " << fileName << " was not opened for writing. " << std::endl;
return false;
}
- Type* buf = const_cast< Type* >( buffer );
- void* host_buffer( 0 );
+ return write_impl< Type, Device >( buffer, elements );
+}
+
+// Host
+template< typename Type,
+ typename Device,
+ typename >
+bool File::write_impl( const Type* buffer,
+ const std::size_t& elements )
+{
+ this->writtenElements = 0;
+ if( std::fwrite( buffer,
+ sizeof( Type ),
+ elements,
+ this->file ) != elements )
+ {
+ std::cerr << "I am not able to write the data to the file " << fileName << "." << std::endl;
+ std::perror( "Fwrite ended with the error code" );
+ return false;
+ }
+ this->writtenElements = elements;
+ return true;
+}
+
+// Cuda
+template< typename Type,
+ typename Device,
+ typename, typename >
+bool File::write_impl( const Type* buffer,
+ const std::size_t& elements )
+{
+#ifdef HAVE_CUDA
this->writtenElements = 0;
- const size_t host_buffer_size = std::min( tnlFileGPUvsCPUTransferBufferSize / sizeof( Type ),
+ const std::size_t host_buffer_size = std::min( tnlFileGPUvsCPUTransferBufferSize / sizeof( Type ),
elements );
- if( std::is_same< Device, Devices::Host >::value )
+ using BaseType = typename std::remove_cv< Type >::type;
+ BaseType* host_buffer = new BaseType[ host_buffer_size ];
+
+ while( this->writtenElements < elements )
{
- if( fwrite( buf,
- sizeof( Type ),
- elements,
- this->file ) != elements )
+ std::size_t transfer = std::min( elements - this->writtenElements, host_buffer_size );
+ cudaMemcpy( host_buffer,
+ ( void* ) & ( buffer[ this->writtenElements ] ),
+ transfer * sizeof( Type ),
+ cudaMemcpyDeviceToHost );
+ if( ! TNL_CHECK_CUDA_DEVICE )
+ {
+ std::cerr << "Transfer of data from the file " << this->fileName
+ << " to the CUDA device failed." << std::endl;
+ delete[] host_buffer;
+ return false;
+ }
+ if( std::fwrite( host_buffer,
+ sizeof( Type ),
+ transfer,
+ this->file ) != transfer )
{
std::cerr << "I am not able to write the data to the file " << fileName << "." << std::endl;
- perror( "Fwrite ended with the error code" );
+ std::perror( "Fwrite ended with the error code" );
+ delete[] host_buffer;
return false;
}
- this->writtenElements = elements;
- return true;
+ this->writtenElements += transfer;
}
- if( std::is_same< Device, Devices::Cuda >::value )
+ delete[] host_buffer;
+ return true;
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
+}
+
+// MIC
+template< typename Type,
+ typename Device,
+ typename, typename, typename >
+bool File::write_impl( const Type* buffer,
+ const std::size_t& elements )
+{
+#ifdef HAVE_MIC
+ this->writtenElements = 0;
+ const std::size_t host_buffer_size = std::min( tnlFileGPUvsCPUTransferBufferSize / sizeof( Type ),
+ elements );
+ Type * host_buffer = (Type *)malloc( sizeof( Type ) * host_buffer_size );
+ if( ! host_buffer )
{
-#ifdef HAVE_CUDA
- /*!***
- * Here we cannot use
- *
- * host_buffer = new Type[ host_buffer_size ];
- *
- * because it does not work for constant types like
- * T = const bool.
+ std::cerr << "I am sorry but I cannot allocate supporting buffer on the host for writing data from the GPU to the file "
+ << this->getFileName() << "." << std::endl;
+ return false;
+ }
+
+ while( this->writtenElements < elements )
+ {
+ std::size_t transfer = std::min( elements - this->writtenElements, host_buffer_size );
+
+ Devices::MICHider<const Type> device_buff;
+ device_buff.pointer=buffer;
+ #pragma offload target(mic) in(device_buff,writtenElements) out(host_buffer:length(transfer))
+ {
+ //THIS SHOULD WORK... BUT NOT WHY?
+ /*for(int i=0;i<transfer;i++)
+ host_buffer[i]=device_buff.pointer[writtenElements+i];
*/
- host_buffer = malloc( sizeof( Type ) * host_buffer_size );
- if( ! host_buffer )
- {
- std::cerr << "I am sorry but I cannot allocate supporting buffer on the host for writing data from the GPU to the file "
- << this->getFileName() << "." << std::endl;
- return false;
- }
-
- while( this->writtenElements < elements )
- {
- size_t transfer = std::min( elements - this->writtenElements, host_buffer_size );
- cudaMemcpy( host_buffer,
- ( void* ) & ( buffer[ this->writtenElements ] ),
- transfer * sizeof( Type ),
- cudaMemcpyDeviceToHost );
- if( ! checkCudaDevice )
- {
- std::cerr << "Transfer of data from the file " << this->fileName
- << " to the CUDA device failed." << std::endl;
- free( host_buffer );
- return false;
- }
- if( fwrite( host_buffer,
- sizeof( Type ),
- transfer,
- this->file ) != transfer )
- {
- std::cerr << "I am not able to write the data to the file " << fileName << "." << std::endl;
- perror( "Fwrite ended with the error code" );
- return false;
- }
- this->writtenElements += transfer;
- }
- free( host_buffer );
- return true;
-#else
- CudaSupportMissingMessage;;
+
+ memcpy(host_buffer,&(device_buff.pointer[writtenElements]), transfer*sizeof(Type) );
+ }
+
+ if( fwrite( host_buffer,
+ sizeof( Type ),
+ transfer,
+ this->file ) != transfer )
+ {
+ std::cerr << "I am not able to write the data to the file " << fileName << "." << std::endl;
+ perror( "Fwrite ended with the error code" );
return false;
-#endif
+ }
+ this->writtenElements += transfer;
}
+ free( host_buffer );
return true;
-};
+#else
+ throw Exceptions::MICSupportMissing();
+#endif
+}
} // namespace TNL
-
-
diff --git a/src/TNL/Functions/Analytic/Blob.h b/src/TNL/Functions/Analytic/Blob.h
index 57786beeebf6e9fbfaebed4895e86957f19b9027..e12a27393c7077f71fe57137a6fce3a2abc00d0c 100644
--- a/src/TNL/Functions/Analytic/Blob.h
+++ b/src/TNL/Functions/Analytic/Blob.h
@@ -54,15 +54,9 @@ class Blob< 1, Real > : public BlobBase< Real, 1 >
Blob();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -85,15 +79,9 @@ class Blob< 2, Real > : public BlobBase< Real, 2 >
Blob();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -117,15 +105,9 @@ class Blob< 3, Real > : public BlobBase< Real, 3 >
Blob();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/CMakeLists.txt b/src/TNL/Functions/Analytic/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Functions/Analytic/Constant.h b/src/TNL/Functions/Analytic/Constant.h
index c0c27c19f412d3fda6db952bf34f0d39c9eedb8d..927d5a421320cd91eeaaa759add37126f3242c0e 100644
--- a/src/TNL/Functions/Analytic/Constant.h
+++ b/src/TNL/Functions/Analytic/Constant.h
@@ -27,6 +27,7 @@ class Constant : public Domain< dimensions, NonspaceDomain >
typedef Real RealType;
typedef Containers::StaticVector< dimensions, RealType > PointType;
+ __cuda_callable__
Constant();
static void configSetup( Config::ConfigDescription& config,
@@ -39,15 +40,9 @@ class Constant : public Domain< dimensions, NonspaceDomain >
const RealType& getConstant() const;
- #ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
- #else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
- #endif
__cuda_callable__ inline
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/Constant_impl.h b/src/TNL/Functions/Analytic/Constant_impl.h
index 21311046be06cba4ce7f36ad4e8839d37482e459..bbd46853dc3cbe98448819876ffc8fc17940cc1a 100644
--- a/src/TNL/Functions/Analytic/Constant_impl.h
+++ b/src/TNL/Functions/Analytic/Constant_impl.h
@@ -16,6 +16,7 @@ namespace Analytic {
template< int Dimension,
typename Real >
+__cuda_callable__
Constant< Dimension, Real >::
Constant()
: constant( 0.0 )
diff --git a/src/TNL/Functions/Analytic/Cylinder.h b/src/TNL/Functions/Analytic/Cylinder.h
index 831b9fa18f1862949bb794318bc72eebb01f33c6..fb3f0542ceda6b3c1b334c300419549d5217bf2b 100644
--- a/src/TNL/Functions/Analytic/Cylinder.h
+++ b/src/TNL/Functions/Analytic/Cylinder.h
@@ -58,17 +58,10 @@ class Cylinder< 1, Real > : public CylinderBase< Real, 1 >
Cylinder();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder,
- typename Point >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0,
typename Point = PointType >
-#endif
__cuda_callable__
RealType getPartialDerivative( const Point& v,
const Real& time = 0.0 ) const;
@@ -92,17 +85,10 @@ class Cylinder< 2, Real > : public CylinderBase< Real, 2 >
Cylinder();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder,
- typename Point >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0,
typename Point = PointType >
-#endif
__cuda_callable__
RealType getPartialDerivative( const Point& v,
const Real& time = 0.0 ) const;
@@ -126,17 +112,10 @@ class Cylinder< 3, Real > : public CylinderBase< Real, 3 >
Cylinder();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder,
- typename Point >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0,
typename Point = PointType >
-#endif
__cuda_callable__
RealType getPartialDerivative( const Point& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/ExpBump.h b/src/TNL/Functions/Analytic/ExpBump.h
index ce82270ccc70cc5b0044a7719ed4eec3d7234e3a..36b07c9a27d549532fb055c01c075cf5e30aa8a8 100644
--- a/src/TNL/Functions/Analytic/ExpBump.h
+++ b/src/TNL/Functions/Analytic/ExpBump.h
@@ -62,15 +62,9 @@ class ExpBump< 1, Real > : public ExpBumpBase< 1, Real >
ExpBump();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -92,15 +86,9 @@ class ExpBump< 2, Real > : public ExpBumpBase< 2, Real >
ExpBump();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__ inline
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -123,15 +111,9 @@ class ExpBump< 3, Real > : public ExpBumpBase< 3, Real >
ExpBump();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/Flowerpot.h b/src/TNL/Functions/Analytic/Flowerpot.h
index 2d33a4a9c989b8d0265576edbcc76ca4305418a8..5a42c5f94249aa69c320959a9713f13f88beec56 100644
--- a/src/TNL/Functions/Analytic/Flowerpot.h
+++ b/src/TNL/Functions/Analytic/Flowerpot.h
@@ -58,17 +58,10 @@ class Flowerpot< 1, Real > : public FlowerpotBase< Real, 1 >
Flowerpot();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder,
- typename Point >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0,
typename Point = PointType >
-#endif
__cuda_callable__
RealType getPartialDerivative( const Point& v,
const Real& time = 0.0 ) const;
@@ -92,17 +85,10 @@ class Flowerpot< 2, Real > : public FlowerpotBase< Real, 2 >
Flowerpot();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder,
- typename Point >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0,
typename Point = PointType >
-#endif
__cuda_callable__
RealType getPartialDerivative( const Point& v,
const Real& time = 0.0 ) const;
@@ -126,17 +112,10 @@ class Flowerpot< 3, Real > : public FlowerpotBase< Real, 3 >
Flowerpot();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder,
- typename Point >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0,
typename Point = PointType >
-#endif
__cuda_callable__
RealType getPartialDerivative( const Point& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/Paraboloid.h b/src/TNL/Functions/Analytic/Paraboloid.h
index dd15623785837cbac2d35388ba5af5cb6dc6fbb0..7bbdcca53402994574d45e07526e5af0839a1ec7 100644
--- a/src/TNL/Functions/Analytic/Paraboloid.h
+++ b/src/TNL/Functions/Analytic/Paraboloid.h
@@ -67,15 +67,9 @@ class Paraboloid< 1, Real > : public ParaboloidBase< 1, Real >
typedef Real RealType;
typedef Containers::StaticVector< 1, RealType > PointType;
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -97,15 +91,9 @@ class Paraboloid< 2, Real > : public ParaboloidBase< 2, Real >
typedef Real RealType;
typedef Containers::StaticVector< 2, RealType > PointType;
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -129,15 +117,9 @@ class Paraboloid< 3, Real > : public ParaboloidBase< 3, Real >
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/ParaboloidSDF.h b/src/TNL/Functions/Analytic/ParaboloidSDF.h
index 20c6662aacb74f482c979cb2e60713109dab000a..60a8cedfd754c1ff350c659475b24bd5d5dedaff 100644
--- a/src/TNL/Functions/Analytic/ParaboloidSDF.h
+++ b/src/TNL/Functions/Analytic/ParaboloidSDF.h
@@ -67,15 +67,9 @@ class ParaboloidSDF< 1, Real > : public ParaboloidSDFBase< 1, Real >
typedef Real RealType;
typedef Containers::StaticVector< 1, RealType > PointType;
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -94,15 +88,9 @@ class ParaboloidSDF< 2, Real > : public ParaboloidSDFBase< 2, Real >
typedef Real RealType;
typedef Containers::StaticVector< 2, RealType > PointType;
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -123,15 +111,9 @@ class ParaboloidSDF< 3, Real > : public ParaboloidSDFBase< 3, Real >
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/PseudoSquare.h b/src/TNL/Functions/Analytic/PseudoSquare.h
index 2c027a739367a44fc7cadf141add4abd61f5b8cd..ea4a5ae84e7e306560e67c74f40075cd3cc5a883 100644
--- a/src/TNL/Functions/Analytic/PseudoSquare.h
+++ b/src/TNL/Functions/Analytic/PseudoSquare.h
@@ -54,15 +54,9 @@ class PseudoSquare< 1, Real > : public PseudoSquareBase< Real, 1 >
PseudoSquare();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -85,15 +79,9 @@ class PseudoSquare< 2, Real > : public PseudoSquareBase< Real, 2 >
PseudoSquare();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -116,15 +104,9 @@ class PseudoSquare< 3, Real > : public PseudoSquareBase< Real, 3 >
PseudoSquare();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/SinBumps.h b/src/TNL/Functions/Analytic/SinBumps.h
index 9af5ef739d1e72f629daa4cf6a2649500c5899b7..19b176cc8d48578db7ad3c314b89b30eb66ec760 100644
--- a/src/TNL/Functions/Analytic/SinBumps.h
+++ b/src/TNL/Functions/Analytic/SinBumps.h
@@ -72,15 +72,9 @@ class SinBumps< 1, Real > : public SinBumpsBase< Containers::StaticVector< 1, R
bool setup( const Config::ParameterContainer& parameters,
const String& prefix = "" );
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -105,15 +99,9 @@ class SinBumps< 2, Real > : public SinBumpsBase< Containers::StaticVector< 2, Re
bool setup( const Config::ParameterContainer& parameters,
const String& prefix = "" );
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -137,15 +125,9 @@ class SinBumps< 3, Real > : public SinBumpsBase< Containers::StaticVector< 3, Re
bool setup( const Config::ParameterContainer& parameters,
const String& prefix = "" );
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/SinBumpsSDF.h b/src/TNL/Functions/Analytic/SinBumpsSDF.h
index 6f4fe5979889d772814f13407a4e36d46b10387a..e2e242eb4a3bea1fbd784d0923f7f95d4dc0d532 100644
--- a/src/TNL/Functions/Analytic/SinBumpsSDF.h
+++ b/src/TNL/Functions/Analytic/SinBumpsSDF.h
@@ -70,15 +70,9 @@ class SinBumpsSDF< 1, Real > : public SinBumpsSDFBase< Containers::StaticVector
bool setup( const Config::ParameterContainer& parameters,
const String& prefix = "" );
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -103,15 +97,9 @@ class SinBumpsSDF< 2, Real > : public SinBumpsSDFBase< Containers::StaticVector<
bool setup( const Config::ParameterContainer& parameters,
const String& prefix = "" );
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -135,15 +123,9 @@ class SinBumpsSDF< 3, Real > : public SinBumpsSDFBase< Containers::StaticVector<
bool setup( const Config::ParameterContainer& parameters,
const String& prefix = "" );
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/SinWave.h b/src/TNL/Functions/Analytic/SinWave.h
index f4b955e2e815ee089df5b31b9efb9bf55245bf06..635125818381d383f9db36f313498887f8f18ac3 100644
--- a/src/TNL/Functions/Analytic/SinWave.h
+++ b/src/TNL/Functions/Analytic/SinWave.h
@@ -68,15 +68,9 @@ class SinWave< 1, Real > : public SinWaveBase< 1, Real >
typedef Real RealType;
typedef Containers::StaticVector< 1, RealType > PointType;
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -95,15 +89,9 @@ class SinWave< 2, Real > : public SinWaveBase< 2, Real >
typedef Real RealType;
typedef Containers::StaticVector< 2, RealType > PointType;
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -124,15 +112,9 @@ class SinWave< 3, Real > : public SinWaveBase< 3, Real >
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/SinWaveSDF.h b/src/TNL/Functions/Analytic/SinWaveSDF.h
index b186693f37a6e656d3e01791bd7a69c25a61bdfb..e4db3f3a5832749e2da34d9e516c75d07c595196 100644
--- a/src/TNL/Functions/Analytic/SinWaveSDF.h
+++ b/src/TNL/Functions/Analytic/SinWaveSDF.h
@@ -66,15 +66,9 @@ class SinWaveSDF< 1, Real > : public SinWaveSDFBase< 1, Real >
typedef Real RealType;
typedef Containers::StaticVector< 1, RealType > PointType;
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -93,15 +87,9 @@ class SinWaveSDF< 2, Real > : public SinWaveSDFBase< 2, Real >
typedef Real RealType;
typedef Containers::StaticVector< 2, RealType > PointType;
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
@@ -122,15 +110,9 @@ class SinWaveSDF< 3, Real > : public SinWaveSDFBase< 3, Real >
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
-#endif
__cuda_callable__
RealType getPartialDerivative( const PointType& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/SinWaveSDF_impl.h b/src/TNL/Functions/Analytic/SinWaveSDF_impl.h
index d7ac240901502c9d3fcfb391da969954dc40609d..d104ca32210896e9103dbd4205670da4202c1940 100644
--- a/src/TNL/Functions/Analytic/SinWaveSDF_impl.h
+++ b/src/TNL/Functions/Analytic/SinWaveSDF_impl.h
@@ -13,8 +13,8 @@
#include <TNL/Functions/Analytic/SinWaveSDF.h>
namespace TNL {
- namespace Functions {
- namespace Analytic {
+namespace Functions {
+namespace Analytic {
template< int dimensions, typename Real >
SinWaveSDFBase< dimensions, Real >::SinWaveSDFBase()
@@ -115,7 +115,7 @@ getPartialDerivative( const PointType& v,
const RealType distance = ::sqrt( x * x ) + this->phase * this->waveLength / (2.0*M_PI);
if( XDiffOrder == 0 )
return this->sinWaveFunctionSDF( distance );
- TNL_ASSERT( false, std::cerr << "TODO: implement this" );
+ TNL_ASSERT_TRUE( false, "TODO: implement this" );
return 0.0;
}
@@ -138,7 +138,7 @@ getPartialDerivative( const PointType& v,
const RealType distance = ::sqrt( x * x + y * y ) + this->phase * this->waveLength / (2.0*M_PI);
if( XDiffOrder == 0 && YDiffOrder == 0)
return this->sinWaveFunctionSDF( distance );
- TNL_ASSERT( false, std::cerr << "TODO: implement this" );
+ TNL_ASSERT_TRUE( false, "TODO: implement this" );
return 0.0;
}
@@ -158,10 +158,10 @@ getPartialDerivative( const PointType& v,
const RealType distance = ::sqrt( x * x + y * y + z * z ) + this->phase * this->waveLength / (2.0*M_PI);
if( XDiffOrder == 0 && YDiffOrder == 0 && ZDiffOrder == 0 )
return this->sinWaveFunctionSDF( distance );
- TNL_ASSERT( false, std::cerr << "TODO: implement this" );
+ TNL_ASSERT_TRUE( false, "TODO: implement this" );
return 0.0;
}
- } // namespace Analytic
- } // namespace Functions
+} // namespace Analytic
+} // namespace Functions
} // namespace TNL
diff --git a/src/TNL/Functions/Analytic/Twins.h b/src/TNL/Functions/Analytic/Twins.h
index 31ecdd6e4cc8b481f885669ee07bc6aad809d89c..c882ec4eb133c326195151b9d6db07098bec3735 100644
--- a/src/TNL/Functions/Analytic/Twins.h
+++ b/src/TNL/Functions/Analytic/Twins.h
@@ -50,17 +50,10 @@ class Twins< 1, Real > : public TwinsBase< Real, 1 >
Twins();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder,
- typename Point >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0,
typename Point = PointType >
-#endif
__cuda_callable__
RealType getPartialDerivative( const Point& v,
const Real& time = 0.0 ) const;
@@ -84,17 +77,10 @@ class Twins< 2, Real > : public TwinsBase< Real, 2 >
Twins();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder,
- typename Point >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0,
typename Point = PointType >
-#endif
__cuda_callable__
RealType getPartialDerivative( const Point& v,
const Real& time = 0.0 ) const;
@@ -118,17 +104,10 @@ class Twins< 3, Real > : public TwinsBase< Real, 3 >
Twins();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder,
- typename Point >
-#else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0,
typename Point = PointType >
-#endif
__cuda_callable__
RealType getPartialDerivative( const Point& v,
const Real& time = 0.0 ) const;
diff --git a/src/TNL/Functions/Analytic/VectorNorm.h b/src/TNL/Functions/Analytic/VectorNorm.h
index 447b31973815ddbdee4ff99c09c518d7fd75b753..b54513705980249db0037044b00dabeca045bce6 100644
--- a/src/TNL/Functions/Analytic/VectorNorm.h
+++ b/src/TNL/Functions/Analytic/VectorNorm.h
@@ -211,7 +211,7 @@ class VectorNorm< 2, Real > : public VectorNormBase< 2, Real >
return ( std::pow( std::pow( TNL::abs( x ), this->power ) * this->anisotropy.x() +
std::pow( TNL::abs( y ), this->power ) * this->anisotropy.y(), 1.0 / this-> power ) - this->radius ) * this->multiplicator;
}
- TNL_ASSERT( false, std::cerr << "Not implemented yet." << std::endl );
+ TNL_ASSERT_TRUE( false, "Not implemented yet." );
return 0.0;
}
@@ -262,7 +262,7 @@ class VectorNorm< 3, Real > : public VectorNormBase< 3, Real >
std::pow( TNL::abs( y ), this->power ) * this->anisotropy.y() +
std::pow( TNL::abs( z ), this->power ) * this->anisotropy.z(), 1.0 / this-> power ) - this->radius ) * this->multiplicator;
}
- TNL_ASSERT( false, std::cerr << "Not implemented yet." << std::endl );
+ TNL_ASSERT_TRUE( false, "Not implemented yet." );
return 0.0;
}
diff --git a/src/TNL/Functions/CMakeLists.txt b/src/TNL/Functions/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Functions/FunctionAdapter.h b/src/TNL/Functions/FunctionAdapter.h
index 2128ce811d135c28262b8b8b0c684d7016cfcaba..425996f5282e26cdf345c9543a5bb24c50ef2575 100644
--- a/src/TNL/Functions/FunctionAdapter.h
+++ b/src/TNL/Functions/FunctionAdapter.h
@@ -29,7 +29,7 @@ class FunctionAdapter
typedef Function FunctionType;
typedef Mesh MeshType;
typedef typename FunctionType::RealType RealType;
- typedef typename MeshType::IndexType IndexType;
+ typedef typename MeshType::GlobalIndexType IndexType;
//typedef typename FunctionType::PointType PointType;
template< typename MeshPointer >
@@ -64,7 +64,7 @@ class FunctionAdapter< Mesh, Function, SpaceDomain >
typedef Function FunctionType;
typedef Mesh MeshType;
typedef typename FunctionType::RealType RealType;
- typedef typename MeshType::IndexType IndexType;
+ typedef typename MeshType::GlobalIndexType IndexType;
typedef typename FunctionType::PointType PointType;
template< typename MeshPointer >
@@ -101,7 +101,7 @@ class FunctionAdapter< Mesh, Function, NonspaceDomain >
typedef Function FunctionType;
typedef Mesh MeshType;
typedef typename FunctionType::RealType RealType;
- typedef typename MeshType::IndexType IndexType;
+ typedef typename MeshType::GlobalIndexType IndexType;
typedef typename FunctionType::PointType PointType;
template< typename MeshPointer >
@@ -137,7 +137,7 @@ class FunctionAdapter< Mesh, Function, MeshFunction >
typedef Function FunctionType;
typedef Mesh MeshType;
typedef typename FunctionType::RealType RealType;
- typedef typename MeshType::IndexType IndexType;
+ typedef typename MeshType::GlobalIndexType IndexType;
template< typename EntityType >
__cuda_callable__ inline
@@ -161,7 +161,7 @@ class FunctionAdapter< Mesh, Function, SpaceDomain >
typedef Function FunctionType;
typedef Mesh MeshType;
typedef typename FunctionType::RealType RealType;
- typedef typename MeshType::IndexType IndexType;
+ typedef typename MeshType::GlobalIndexType IndexType;
typedef typename FunctionType::PointType PointType;
template< typename EntityType >
@@ -186,7 +186,7 @@ class FunctionAdapter< Mesh, Function, SpaceDomain >
typedef Function FunctionType;
typedef Mesh MeshType;
typedef typename FunctionType::RealType RealType;
- typedef typename MeshType::IndexType IndexType;
+ typedef typename MeshType::GlobalIndexType IndexType;
typedef typename FunctionType::PointType PointType;
template< typename EntityType >
diff --git a/src/TNL/Functions/MeshFunction.h b/src/TNL/Functions/MeshFunction.h
index b7744cb3d37327707a9324e6eeb14f4e77da9d47..2c8d41a564c3af16ad406fbf9dca5cf95f3e98e5 100644
--- a/src/TNL/Functions/MeshFunction.h
+++ b/src/TNL/Functions/MeshFunction.h
@@ -20,11 +20,11 @@ namespace TNL {
namespace Functions {
template< typename Mesh,
- int MeshEntityDimension = Mesh::meshDimension,
+ int MeshEntityDimension = Mesh::getMeshDimension(),
typename Real = typename Mesh::RealType >
class MeshFunction :
public Object,
- public Domain< Mesh::meshDimension, MeshDomain >
+ public Domain< Mesh::getMeshDimension(), MeshDomain >
{
//static_assert( Mesh::DeviceType::DeviceType == Vector::DeviceType::DeviceType,
// "Both mesh and vector of a mesh function must reside on the same device.");
@@ -32,7 +32,7 @@ class MeshFunction :
typedef Mesh MeshType;
typedef typename MeshType::DeviceType DeviceType;
- typedef typename MeshType::IndexType IndexType;
+ typedef typename MeshType::GlobalIndexType IndexType;
typedef SharedPointer< MeshType > MeshPointer;
typedef Real RealType;
typedef Containers::Vector< RealType, DeviceType, IndexType > VectorType;
@@ -40,7 +40,7 @@ class MeshFunction :
static constexpr int getEntitiesDimension() { return MeshEntityDimension; }
- static constexpr int getMeshDimensions() { return MeshType::getMeshDimension(); }
+ static constexpr int getMeshDimension() { return MeshType::getMeshDimension(); }
MeshFunction();
@@ -94,7 +94,7 @@ class MeshFunction :
const MeshPointer& getMeshPointer() const;
- __cuda_callable__ static IndexType getDofs( const MeshPointer& meshPointer );
+ static IndexType getDofs( const MeshPointer& meshPointer );
__cuda_callable__ const VectorType& getData() const;
diff --git a/src/TNL/Functions/MeshFunctionEvaluator.h b/src/TNL/Functions/MeshFunctionEvaluator.h
index ad08b901404b1110c81943ffd753984e9ebe9aaa..f4f544d1769523087c396b5691a1be1526bc6d5f 100644
--- a/src/TNL/Functions/MeshFunctionEvaluator.h
+++ b/src/TNL/Functions/MeshFunctionEvaluator.h
@@ -10,7 +10,6 @@
#pragma once
-#include <TNL/Meshes/Grid.h>
#include <TNL/Functions/MeshFunction.h>
#include <TNL/Functions/OperatorFunction.h>
#include <TNL/Functions/FunctionAdapter.h>
@@ -162,4 +161,3 @@ class MeshFunctionEvaluatorAdditionEntitiesProcessor
} // namespace TNL
#include <TNL/Functions/MeshFunctionEvaluator_impl.h>
-
diff --git a/src/TNL/Functions/MeshFunctionEvaluator_impl.h b/src/TNL/Functions/MeshFunctionEvaluator_impl.h
index 408e0882fc1e354ae87d89e1859fc61150fd0049..540ae7077b73baab728c215c425b6bedfc478f1e 100644
--- a/src/TNL/Functions/MeshFunctionEvaluator_impl.h
+++ b/src/TNL/Functions/MeshFunctionEvaluator_impl.h
@@ -115,7 +115,7 @@ evaluateEntities( OutMeshFunctionPointer& meshFunction,
static_assert( std::is_same< typename std::decay< typename OutMeshFunctionPointer::ObjectType >::type, OutMeshFunction >::value, "expected a smart pointer" );
static_assert( std::is_same< typename std::decay< typename InFunctionPointer::ObjectType >::type, InFunction >::value, "expected a smart pointer" );
- typedef typename MeshType::template MeshEntity< OutMeshFunction::getEntitiesDimension() > MeshEntityType;
+ typedef typename MeshType::template EntityType< OutMeshFunction::getEntitiesDimension() > MeshEntityType;
typedef Functions::MeshFunctionEvaluatorAssignmentEntitiesProcessor< MeshType, TraverserUserData > AssignmentEntitiesProcessor;
typedef Functions::MeshFunctionEvaluatorAdditionEntitiesProcessor< MeshType, TraverserUserData > AdditionEntitiesProcessor;
//typedef typename OutMeshFunction::MeshPointer OutMeshPointer;
@@ -171,4 +171,3 @@ evaluateEntities( OutMeshFunctionPointer& meshFunction,
} // namespace Functions
} // namespace TNL
-
diff --git a/src/TNL/Functions/MeshFunctionNormGetter.h b/src/TNL/Functions/MeshFunctionNormGetter.h
index 6f7e127c173b0dc0b67f51f800ced2f9d333cc6a..50e39c6de767fa79828d7e4bc9c488356c1d5be6 100644
--- a/src/TNL/Functions/MeshFunctionNormGetter.h
+++ b/src/TNL/Functions/MeshFunctionNormGetter.h
@@ -134,7 +134,7 @@ class MeshFunctionNormGetter< MeshFunction< Meshes::Grid< Dimension, MeshReal, D
}
if( EntityDimension > 0 )
{
- TNL_ASSERT( false, std::cerr << "Not implemented yet." << std::endl );
+ TNL_ASSERT_TRUE( false, "Not implemented yet." );
}
if( p == 1.0 )
diff --git a/src/TNL/Functions/MeshFunctionVTKWriter_impl.h b/src/TNL/Functions/MeshFunctionVTKWriter_impl.h
index 56f1d4cdd6ba3180df071209fae4e27b52fb1e92..606eac9ce348e2260109845af0bb05cea064a3e8 100644
--- a/src/TNL/Functions/MeshFunctionVTKWriter_impl.h
+++ b/src/TNL/Functions/MeshFunctionVTKWriter_impl.h
@@ -279,8 +279,8 @@ write( const MeshFunctionType& function,
std::ostream& str,
const double& scale )
{
- typedef typename MeshType::template MeshEntity< 0 > Vertex;
- typedef typename MeshType::template MeshEntity< 1 > Face;
+ typedef typename MeshType::template EntityType< 0 > Vertex;
+ typedef typename MeshType::template EntityType< 1 > Face;
writeHeader(function, str);
const MeshType& mesh = function.getMesh();
@@ -369,7 +369,7 @@ write( const MeshFunctionType& function,
std::ostream& str,
const double& scale )
{
- typedef typename MeshType::template MeshEntity< 0 > Vertex;
+ typedef typename MeshType::template EntityType< 0 > Vertex;
writeHeader(function, str);
const MeshType& mesh = function.getMesh();
diff --git a/src/TNL/Functions/MeshFunction_impl.h b/src/TNL/Functions/MeshFunction_impl.h
index 2ab5a06457527cb9bad9b89df70ed59e0f941175..7164170f1ab51a20bcdeaeb9011cb97bb97b4590 100644
--- a/src/TNL/Functions/MeshFunction_impl.h
+++ b/src/TNL/Functions/MeshFunction_impl.h
@@ -36,10 +36,10 @@ MeshFunction< Mesh, MeshEntityDimension, Real >::
MeshFunction( const MeshPointer& meshPointer )
: meshPointer( meshPointer )
{
- this->data.setSize( meshPointer->template getEntitiesCount< typename Mesh::template MeshEntity< MeshEntityDimension > >() );
- TNL_ASSERT( this->data.getSize() == this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >(),
- std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
- << "this->mesh->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() = " << this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() );
+ this->data.setSize( getMesh().template getEntitiesCount< typename Mesh::template EntityType< MeshEntityDimension > >() );
+ TNL_ASSERT( this->data.getSize() == this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >(),
+ std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
+ << "this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() = " << this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() );
}
template< typename Mesh,
@@ -62,10 +62,10 @@ MeshFunction( const MeshPointer& meshPointer,
const IndexType& offset )
: meshPointer( meshPointer )
{
- this->data.bind( data, offset, meshPointer->template getEntitiesCount< typename Mesh::template MeshEntity< MeshEntityDimension > >() );
- TNL_ASSERT( this->data.getSize() == this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >(),
- std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
- << "this->mesh->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() = " << this->meshPointer->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() );
+ this->data.bind( data, offset, getMesh().template getEntitiesCount< typename Mesh::template EntityType< MeshEntityDimension > >() );
+ TNL_ASSERT( this->data.getSize() == this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >(),
+ std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
+ << "this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() = " << this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() );
}
@@ -79,10 +79,10 @@ MeshFunction( const MeshPointer& meshPointer,
const IndexType& offset )
: meshPointer( meshPointer )
{
- this->data.bind( *data, offset, meshPointer->template getEntitiesCount< typename Mesh::template MeshEntity< MeshEntityDimension > >() );
- TNL_ASSERT( this->data.getSize() == this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >(),
- std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
- << "this->mesh->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() = " << this->meshPointer->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() );
+ this->data.bind( *data, offset, getMesh().template getEntitiesCount< typename Mesh::template EntityType< MeshEntityDimension > >() );
+ TNL_ASSERT( this->data.getSize() == this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >(),
+ std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
+ << "this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() = " << this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() );
}
template< typename Mesh,
@@ -191,10 +191,10 @@ bind( const MeshPointer& meshPointer,
const IndexType& offset )
{
this->meshPointer = meshPointer;
- this->data.bind( data, offset, meshPointer->template getEntitiesCount< typename Mesh::template MeshEntity< MeshEntityDimension > >() );
- TNL_ASSERT( this->data.getSize() == this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >(),
- std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
- << "this->mesh->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() = " << this->meshPointer->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() );
+ this->data.bind( data, offset, getMesh().template getEntitiesCount< typename Mesh::template EntityType< MeshEntityDimension > >() );
+ TNL_ASSERT( this->data.getSize() == this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >(),
+ std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
+ << "this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() = " << this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() );
}
template< typename Mesh,
@@ -208,10 +208,10 @@ bind( const MeshPointer& meshPointer,
const IndexType& offset )
{
this->meshPointer = meshPointer;
- this->data.bind( *data, offset, meshPointer->template getEntitiesCount< typename Mesh::template MeshEntity< MeshEntityDimension > >() );
- TNL_ASSERT( this->data.getSize() == this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >(),
- std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
- << "this->mesh->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() = " << this->meshPointer->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() );
+ this->data.bind( *data, offset, getMesh().template getEntitiesCount< typename Mesh::template EntityType< MeshEntityDimension > >() );
+ TNL_ASSERT( this->data.getSize() == this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >(),
+ std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
+ << "this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() = " << this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() );
}
@@ -223,10 +223,10 @@ MeshFunction< Mesh, MeshEntityDimension, Real >::
setMesh( const MeshPointer& meshPointer )
{
this->meshPointer = meshPointer;
- this->data.setSize( meshPointer->template getEntitiesCount< typename Mesh::template MeshEntity< MeshEntityDimension > >() );
- TNL_ASSERT( this->data.getSize() == this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >(),
- std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
- << "this->mesh->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() = " << this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() );
+ this->data.setSize( getMesh().template getEntitiesCount< typename Mesh::template EntityType< MeshEntityDimension > >() );
+ TNL_ASSERT( this->data.getSize() == this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >(),
+ std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
+ << "this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() = " << this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() );
}
template< typename Mesh,
@@ -254,7 +254,6 @@ getMeshPointer() const
template< typename Mesh,
int MeshEntityDimension,
typename Real >
-__cuda_callable__
typename MeshFunction< Mesh, MeshEntityDimension, Real >::IndexType
MeshFunction< Mesh, MeshEntityDimension, Real >::
getDofs( const MeshPointer& meshPointer )
@@ -312,7 +311,7 @@ typename Functions::MeshFunction< Mesh, MeshEntityDimension, Real >::RealType
MeshFunction< Mesh, MeshEntityDimension, Real >::
getValue( const EntityType& meshEntity ) const
{
- static_assert( EntityType::entityDimension == MeshEntityDimension, "Calling with wrong EntityType -- entity dimension do not match." );
+ static_assert( EntityType::getEntityDimension() == MeshEntityDimension, "Calling with wrong EntityType -- entity dimensions do not match." );
return this->data.getValue( meshEntity.getIndex() );
}
@@ -325,7 +324,7 @@ MeshFunction< Mesh, MeshEntityDimension, Real >::
setValue( const EntityType& meshEntity,
const RealType& value )
{
- static_assert( EntityType::entityDimension == MeshEntityDimension, "Calling with wrong EntityType -- entity dimension do not match." );
+ static_assert( EntityType::getEntityDimension() == MeshEntityDimension, "Calling with wrong EntityType -- entity dimensions do not match." );
this->data.setValue( meshEntity.getIndex(), value );
}
@@ -339,7 +338,7 @@ MeshFunction< Mesh, MeshEntityDimension, Real >::
operator()( const EntityType& meshEntity,
const RealType& time )
{
- static_assert( EntityType::entityDimension == MeshEntityDimension, "Calling with wrong EntityType -- entity dimension do not match." );
+ static_assert( EntityType::getEntityDimension() == MeshEntityDimension, "Calling with wrong EntityType -- entity dimensions do not match." );
return this->data[ meshEntity.getIndex() ];
}
@@ -353,7 +352,7 @@ MeshFunction< Mesh, MeshEntityDimension, Real >::
operator()( const EntityType& meshEntity,
const RealType& time ) const
{
- static_assert( EntityType::entityDimension == MeshEntityDimension, "Calling with wrong EntityType -- entity dimension do not match." );
+ static_assert( EntityType::getEntityDimension() == MeshEntityDimension, "Calling with wrong EntityType -- entity dimensions do not match." );
return this->data[ meshEntity.getIndex() ];
}
@@ -448,9 +447,9 @@ bool
MeshFunction< Mesh, MeshEntityDimension, Real >::
save( File& file ) const
{
- TNL_ASSERT( this->data.getSize() == this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >(),
- std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
- << "this->mesh->template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() = " << this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >() );
+ TNL_ASSERT( this->data.getSize() == this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >(),
+ std::cerr << "this->data.getSize() = " << this->data.getSize() << std::endl
+ << "this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() = " << this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >() );
if( ! Object::save( file ) )
return false;
return this->data.save( file );
@@ -467,7 +466,7 @@ load( File& file )
return false;
if( ! this->data.load( file ) )
return false;
- const IndexType meshSize = this->meshPointer.getData().template getEntitiesCount< typename MeshType::template MeshEntity< MeshEntityDimension > >();
+ const IndexType meshSize = this->getMesh().template getEntitiesCount< typename MeshType::template EntityType< MeshEntityDimension > >();
if( this->data.getSize() != meshSize )
{
std::cerr << "Size of the data loaded to the mesh function (" << this->data.getSize() << ") does not fit with the mesh size (" << meshSize << ")." << std::endl;
diff --git a/src/TNL/Functions/OperatorFunction.h b/src/TNL/Functions/OperatorFunction.h
index 4ae8907daaee25d9c8d3305e0c447f44691f1420..c5dbde42363c27f2d58d53018d7efb41096b76ad 100644
--- a/src/TNL/Functions/OperatorFunction.h
+++ b/src/TNL/Functions/OperatorFunction.h
@@ -87,13 +87,13 @@ class OperatorFunction< Operator, MeshFunctionT, void, true, IsAnalytic >
const MeshType& getMesh() const
{
- TNL_ASSERT( this->preimageFunction, std::cerr << "The preimage function was not set." << std::endl );
+ TNL_ASSERT_TRUE( this->preimageFunction, "The preimage function was not set." );
return this->preimageFunction->getMesh();
};
const MeshPointer& getMeshPointer() const
{
- TNL_ASSERT( this->preimageFunction, std::cerr << "The preimage function was not set." << std::endl );
+ TNL_ASSERT_TRUE( this->preimageFunction, "The preimage function was not set." );
return this->preimageFunction->getMeshPointer();
};
@@ -114,7 +114,7 @@ class OperatorFunction< Operator, MeshFunctionT, void, true, IsAnalytic >
const MeshEntity& meshEntity,
const RealType& time = 0.0 ) const
{
- TNL_ASSERT( this->preimageFunction, std::cerr << "The preimage function was not set." << std::endl );
+ TNL_ASSERT_TRUE( this->preimageFunction, "The preimage function was not set." );
return operator_( *preimageFunction, meshEntity, time );
}
@@ -298,13 +298,13 @@ class OperatorFunction< Operator, Function, BoundaryConditions, false, IsAnalyti
const PreimageFunctionType& getPreimageFunction() const
{
- TNL_ASSERT( this->preimageFunction, );
+ TNL_ASSERT_TRUE( this->preimageFunction, "The preimage function was not set." );
return *this->preimageFunction;
};
PreimageFunctionType& getPreimageFunction()
{
- TNL_ASSERT( this->preimageFunction, );
+ TNL_ASSERT_TRUE( this->preimageFunction, "The preimage function was not set." );
return *this->preimageFunction;
};
diff --git a/src/TNL/Functions/TestFunction.h b/src/TNL/Functions/TestFunction.h
index d01c9df2c547f11f8b059a6f0ec6b8b42cc166c2..0f97ff21af51b74656afa4e9f2d554f026f20290 100644
--- a/src/TNL/Functions/TestFunction.h
+++ b/src/TNL/Functions/TestFunction.h
@@ -65,15 +65,9 @@ class TestFunction : public Domain< FunctionDimension, SpaceDomain >
const TestFunction& operator = ( const TestFunction& function );
- #ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
- #else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
- #endif
__cuda_callable__
Real getPartialDerivative( const PointType& vertex,
const Real& time = 0 ) const;
@@ -86,27 +80,12 @@ class TestFunction : public Domain< FunctionDimension, SpaceDomain >
}
- #ifdef HAVE_NOT_CXX11
- template< int XDiffOrder,
- int YDiffOrder,
- int ZDiffOrder >
- #else
template< int XDiffOrder = 0,
int YDiffOrder = 0,
int ZDiffOrder = 0 >
- #endif
__cuda_callable__
Real getTimeDerivative( const PointType& vertex,
const Real& time = 0 ) const;
- #ifdef HAVE_NOT_CXX11
- template< typename Point >
- __cuda_callable__
- Real getTimeDerivative( const Point& vertex,
- const Real& time = 0 ) const
- {
- return this->getTimeDerivative< 0, 0, 0, Point >( vertex, time );
- }
- #endif
std::ostream& print( std::ostream& str ) const;
diff --git a/src/TNL/Functions/TestFunction_impl.h b/src/TNL/Functions/TestFunction_impl.h
index 2d31c65102801c4cb082419bf65141bdc7df6494..34ae43529b92cadb3f23ca4f57176255c9d09d45 100644
--- a/src/TNL/Functions/TestFunction_impl.h
+++ b/src/TNL/Functions/TestFunction_impl.h
@@ -131,7 +131,7 @@ setupFunction( const Config::ParameterContainer& parameters,
{
this->function = Devices::Cuda::passToDevice( *auxFunction );
delete auxFunction;
- if( ! checkCudaDevice )
+ if( ! TNL_CHECK_CUDA_DEVICE )
return false;
}
return true;
@@ -161,7 +161,7 @@ setupOperator( const Config::ParameterContainer& parameters,
{
this->operator_ = Devices::Cuda::passToDevice( *auxOperator );
delete auxOperator;
- if( ! checkCudaDevice )
+ if( ! TNL_CHECK_CUDA_DEVICE )
return false;
}
return true;
@@ -602,66 +602,66 @@ getTimeDerivative( const PointType& vertex,
case constant:
{
typedef Constant< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
}
case paraboloid:
{
typedef Paraboloid< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
}
case expBump:
{
typedef ExpBump< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
}
case sinBumps:
{
typedef SinBumps< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
}
case sinWave:
{
typedef SinWave< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
break;
}
case cylinder:
{
typedef Cylinder< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
break;
}
case flowerpot:
{
typedef Flowerpot< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
break;
}
case twins:
{
typedef Twins< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
break;
}
case pseudoSquare:
{
typedef PseudoSquare< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
break;
}
case blob:
{
typedef Blob< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
break;
}
@@ -670,19 +670,19 @@ getTimeDerivative( const PointType& vertex,
case paraboloidSDF:
{
typedef ParaboloidSDF< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
}
case sinBumpsSDF:
{
typedef SinBumpsSDF< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
}
case sinWaveSDF:
{
typedef SinWaveSDF< Dimension, Real > FunctionType;
- return scale * ( ( FunctionType* ) function )->
+ return scale * ( ( FunctionType* ) function )->template
getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
}
default:
diff --git a/src/TNL/Functions/VectorField.h b/src/TNL/Functions/VectorField.h
index ad31e09d8d571b96571b98e6216765819a59c559..2fe085e750c931d5d81cd1dbdf2c49fd9af38ad9 100644
--- a/src/TNL/Functions/VectorField.h
+++ b/src/TNL/Functions/VectorField.h
@@ -72,11 +72,11 @@ class VectorField
template< int Size,
typename Mesh,
- int MeshEntityDimensions,
+ int MeshEntityDimension,
typename Real >
-class VectorField< Size, MeshFunction< Mesh, MeshEntityDimensions, Real > >
-: public Functions::Domain< MeshFunction< Mesh, MeshEntityDimensions, Real >::getDomainDimension(),
- MeshFunction< Mesh, MeshEntityDimensions, Real >::getDomainType() >,
+class VectorField< Size, MeshFunction< Mesh, MeshEntityDimension, Real > >
+: public Functions::Domain< MeshFunction< Mesh, MeshEntityDimension, Real >::getDomainDimension(),
+ MeshFunction< Mesh, MeshEntityDimension, Real >::getDomainType() >,
public Object
{
public:
@@ -84,11 +84,11 @@ class VectorField< Size, MeshFunction< Mesh, MeshEntityDimensions, Real > >
typedef Mesh MeshType;
typedef Real RealType;
typedef SharedPointer< MeshType > MeshPointer;
- typedef MeshFunction< MeshType, MeshEntityDimensions, RealType > FunctionType;
+ typedef MeshFunction< MeshType, MeshEntityDimension, RealType > FunctionType;
typedef SharedPointer< FunctionType > FunctionPointer;
typedef typename MeshType::DeviceType DeviceType;
typedef typename MeshType::IndexType IndexType;
- typedef VectorField< Size, MeshFunction< Mesh, MeshEntityDimensions, RealType > > ThisType;
+ typedef VectorField< Size, MeshFunction< Mesh, MeshEntityDimension, RealType > > ThisType;
typedef Containers::StaticVector< Size, RealType > VectorType;
static void configSetup( Config::ConfigDescription& config,
@@ -182,6 +182,7 @@ class VectorField< Size, MeshFunction< Mesh, MeshEntityDimensions, Real > >
{
VectorType v;
for( int i = 0; i < Size; i++ )
+ // FIXME: the dereferencing operator of FunctionPointer is not __cuda_callable__
v[ i ] = ( *this->vectorField[ i ] )[ index ];
return v;
}
@@ -192,6 +193,7 @@ class VectorField< Size, MeshFunction< Mesh, MeshEntityDimensions, Real > >
{
VectorType v;
for( int i = 0; i < Size; i++ )
+ // FIXME: the dereferencing operator of FunctionPointer is not __cuda_callable__
v[ i ] = ( *this->vectorField[ i ] )( meshEntity );
return v;
}
@@ -260,14 +262,14 @@ class VectorField< Size, MeshFunction< Mesh, MeshEntityDimensions, Real > >
};
-template< int Dimensions,
+template< int Dimension,
typename Function >
-std::ostream& operator << ( std::ostream& str, const VectorField< Dimensions, Function >& f )
+std::ostream& operator << ( std::ostream& str, const VectorField< Dimension, Function >& f )
{
- for( int i = 0; i < Dimensions; i++ )
+ for( int i = 0; i < Dimension; i++ )
{
str << "[ " << f[ i ] << " ]";
- if( i < Dimensions - 1 )
+ if( i < Dimension - 1 )
str << ", ";
}
return str;
diff --git a/src/TNL/Math.h b/src/TNL/Math.h
index 1ff7cc923c8eed40b6676bed16e8371137b88c04..ca15bb876d6da3accedb65f7b9c729a0b6d7cda9 100644
--- a/src/TNL/Math.h
+++ b/src/TNL/Math.h
@@ -12,159 +12,108 @@
#include <cmath>
#include <type_traits>
+#include <algorithm>
-#include <TNL/Devices/Cuda.h>
-
-#ifdef HAVE_CUDA
-#include <cuda.h>
-#endif
+#include <TNL/Devices/CudaCallable.h>
namespace TNL {
template< typename T1, typename T2 >
using enable_if_same_base = std::enable_if< std::is_same< typename std::decay< T1 >::type, T2 >::value, T2 >;
+template< typename T1, typename T2 >
+using both_integral_or_floating = typename std::conditional<
+ ( std::is_integral< T1 >::value && std::is_integral< T2 >::value ) ||
+ ( std::is_floating_point< T1 >::value && std::is_floating_point< T2 >::value ),
+ std::true_type,
+ std::false_type >::type;
+
+// 1. If both types are integral or floating-point, the larger type is selected.
+// 2. If one type is integral and the other floating-point, the floating-point type is selected.
+// This is necessary only due to the limitations of nvcc. Note that clang and gcc
+// can handle automatic promotion using a single-type template, exactly like
+// std::min and std::max are implemented in STL.
+template< typename T1, typename T2 >
+using larger_type = typename std::conditional<
+ ( both_integral_or_floating< T1, T2 >::value && sizeof(T1) >= sizeof(T2) ) ||
+ std::is_floating_point<T1>::value,
+ T1, T2 >::type;
+
/***
* This function returns minimum of two numbers.
- * Specializations use the functions defined in the CUDA's math_functions.h
- * in CUDA device code and STL functions otherwise.
+ * GPU device code uses the functions defined in the CUDA's math_functions.h,
+ * MIC uses trivial override and host uses the STL functions.
*/
-template< typename Type1, typename Type2 >
+template< typename T1, typename T2, typename ResultType = larger_type< T1, T2 > >
__cuda_callable__ inline
-Type1 min( const Type1& a, const Type2& b )
+ResultType min( const T1& a, const T2& b )
{
+#if defined(__CUDA_ARCH__)
+ return ::min( (ResultType) a, (ResultType) b );
+#elif defined(__MIC__)
return a < b ? a : b;
-};
-
-// specialization for int
-template< class T >
-__cuda_callable__ inline
-typename enable_if_same_base< T, int >::type
-min( const T& a, const T& b )
-{
-#ifdef __CUDA_ARCH__
- return ::min( a, b );
-#else
- return std::min( a, b );
-#endif
-}
-
-// specialization for float
-template< class T >
-__cuda_callable__ inline
-typename enable_if_same_base< T, float >::type
-min( const T& a, const T& b )
-{
-#ifdef __CUDA_ARCH__
- return ::fminf( a, b );
#else
- return std::fmin( a, b );
-#endif
-}
-
-// specialization for double
-template< class T >
-__cuda_callable__ inline
-typename enable_if_same_base< T, double >::type
-min( const T& a, const T& b )
-{
-#ifdef __CUDA_ARCH__
- return ::fmin( a, b );
-#else
- return std::fmin( a, b );
+ return std::min( (ResultType) a, (ResultType) b );
#endif
}
/***
* This function returns maximum of two numbers.
- * Specializations use the functions defined in the CUDA's math_functions.h
- * in CUDA device code and STL functions otherwise.
+ * GPU device code uses the functions defined in the CUDA's math_functions.h,
+ * MIC uses trivial override and host uses the STL functions.
*/
-template< typename Type1, typename Type2 >
+template< typename T1, typename T2, typename ResultType = larger_type< T1, T2 > >
__cuda_callable__
-Type1 max( const Type1& a, const Type2& b )
+ResultType max( const T1& a, const T2& b )
{
+#if defined(__CUDA_ARCH__)
+ return ::max( (ResultType) a, (ResultType) b );
+#elif defined(__MIC__)
return a > b ? a : b;
-};
-
-// specialization for int
-template< class T >
-__cuda_callable__ inline
-typename enable_if_same_base< T, int >::type
-max( const T& a, const T& b )
-{
-#ifdef __CUDA_ARCH__
- return ::max( a, b );
#else
- return std::max( a, b );
-#endif
-}
-
-// specialization for float
-template< class T >
-__cuda_callable__ inline
-typename enable_if_same_base< T, float >::type
-max( const T& a, const T& b )
-{
-#ifdef __CUDA_ARCH__
- return ::fmaxf( a, b );
-#else
- return std::fmax( a, b );
-#endif
-}
-
-// specialization for double
-template< class T >
-__cuda_callable__ inline
-typename enable_if_same_base< T, double >::type
-max( const T& a, const T& b )
-{
-#ifdef __CUDA_ARCH__
- return ::fmax( a, b );
-#else
- return std::fmax( a, b );
+ return std::max( (ResultType) a, (ResultType) b );
#endif
}
/***
* This function returns absolute value of given number.
- * Specializations use the functions defined in the CUDA's math_functions.h
- * in CUDA device code and STL functions otherwise.
*/
template< class T >
__cuda_callable__ inline
-typename std::enable_if< ! std::is_arithmetic< T >::value, T >::type
-abs( const T& n )
+T abs( const T& n )
{
+#if defined(__MIC__)
if( n < ( T ) 0 )
return -n;
return n;
+#else
+ return std::abs( n );
+#endif
}
-// specialization for any arithmetic type (e.g. int, float, double)
-template< class T >
+
+template< typename T1, typename T2, typename ResultType = larger_type< T1, T2 > >
__cuda_callable__ inline
-typename std::enable_if< std::is_arithmetic< T >::value, T >::type
-abs( const T& n )
+ResultType pow( const T1& base, const T2& exp )
{
-#ifdef __CUDA_ARCH__
- return ::abs( n );
+#if defined(__CUDA_ARCH__) || defined(__MIC__)
+ return ::pow( (ResultType) base, (ResultType) exp );
#else
- return std::abs( n );
+ return std::pow( (ResultType) base, (ResultType) exp );
#endif
}
-template< class T >
+template< typename T >
__cuda_callable__ inline
-T pow( const T& base, const T& exp )
+T sqrt( const T& value )
{
-#ifdef __CUDA_ARCH__
- return ::pow( base, exp );
+#if defined(__CUDA_ARCH__) || defined(__MIC__)
+ return ::sqrt( value );
#else
- return std::pow( base, exp );
+ return std::sqrt( value );
#endif
}
@@ -176,7 +125,7 @@ void swap( Type& a, Type& b )
Type tmp( a );
a = b;
b = tmp;
-};
+}
template< class T >
__cuda_callable__
@@ -185,7 +134,7 @@ T sign( const T& a )
if( a < ( T ) 0 ) return ( T ) -1;
if( a == ( T ) 0 ) return ( T ) 0;
return ( T ) 1;
-};
+}
template< typename Real >
__cuda_callable__
@@ -220,4 +169,3 @@ inline bool isPow2( long int x )
}
} // namespace TNL
-
diff --git a/src/TNL/Matrices/CMakeLists.txt b/src/TNL/Matrices/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Matrices/CSR.h b/src/TNL/Matrices/CSR.h
index bc06d7c2c0872744befb483db48c5b657a7d2f9e..e3d6253e8a09d27404f78801ef44c7ee2699ba81 100644
--- a/src/TNL/Matrices/CSR.h
+++ b/src/TNL/Matrices/CSR.h
@@ -30,7 +30,16 @@ class CSRDeviceDependentCode;
template< typename Real, typename Device = Devices::Host, typename Index = int >
class CSR : public Sparse< Real, Device, Index >
{
- public:
+private:
+ // convenient template alias for controlling the selection of copy-assignment operator
+ template< typename Device2 >
+ using Enabler = std::enable_if< ! std::is_same< Device2, Device >::value >;
+
+ // friend class will be needed for templated assignment operators
+ template< typename Real2, typename Device2, typename Index2 >
+ friend class CSR;
+
+public:
typedef Real RealType;
typedef Device DeviceType;
@@ -52,15 +61,22 @@ class CSR : public Sparse< Real, Device, Index >
String getTypeVirtual() const;
- bool setDimensions( const IndexType rows,
+ static String getSerializationType();
+
+ virtual String getSerializationTypeVirtual() const;
+
+ void setDimensions( const IndexType rows,
const IndexType columns );
- bool setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
+ void setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
IndexType getRowLength( const IndexType row ) const;
+ __cuda_callable__
+ IndexType getRowLengthFast( const IndexType row ) const;
+
template< typename Real2, typename Device2, typename Index2 >
- bool setLike( const CSR< Real2, Device2, Index2 >& matrix );
+ void setLike( const CSR< Real2, Device2, Index2 >& matrix );
void reset();
@@ -154,6 +170,14 @@ class CSR : public Sparse< Real, Device, Index >
Vector& x,
const RealType& omega = 1.0 ) const;
+ // copy assignment
+ CSR& operator=( const CSR& matrix );
+
+ // cross-device copy assignment
+ template< typename Real2, typename Device2, typename Index2,
+ typename = typename Enabler< Device2 >::type >
+ CSR& operator=( const CSR< Real2, Device2, Index2 >& matrix );
+
bool save( File& file ) const;
bool load( File& file );
@@ -201,37 +225,43 @@ class CSR : public Sparse< Real, Device, Index >
// The following getters allow us to interface TNL with external C-like
// libraries such as UMFPACK or SuperLU, which need the raw data.
- Index* getRowPointers()
+ const Containers::Vector< Index, Device, Index >&
+ getRowPointers() const
{
- return this->rowPointers.getData();
+ return this->rowPointers;
}
- const Index* getRowPointers() const
+ Containers::Vector< Index, Device, Index >&
+ getRowPointers()
{
- return this->rowPointers.getData();
+ return this->rowPointers;
}
- Index* getColumnIndexes()
+ const Containers::Vector< Index, Device, Index >&
+ getColumnIndexes() const
{
- return this->columnIndexes.getData();
+ return this->columnIndexes;
}
- const Index* getColumnIndexes() const
+ Containers::Vector< Index, Device, Index >&
+ getColumnIndexes()
{
- return this->columnIndexes.getData();
+ return this->columnIndexes;
}
- Real* getValues()
+ const Containers::Vector< Real, Device, Index >&
+ getValues() const
{
- return this->values.getData();
+ return this->values;
}
- const Real* getValues() const
+ Containers::Vector< Real, Device, Index >&
+ getValues()
{
- return this->values.getData();
+ return this->values;
}
- protected:
+protected:
Containers::Vector< Index, Device, Index > rowPointers;
@@ -248,4 +278,3 @@ class CSR : public Sparse< Real, Device, Index >
} // namespace TNL
#include <TNL/Matrices/CSR_impl.h>
-
diff --git a/src/TNL/Matrices/CSR_impl.h b/src/TNL/Matrices/CSR_impl.h
index 6084df88a8afb4c7bb2389e356e438df71059b4c..0b47633ea9a78b906822bc1e0466e196846dd774 100644
--- a/src/TNL/Matrices/CSR_impl.h
+++ b/src/TNL/Matrices/CSR_impl.h
@@ -12,7 +12,6 @@
#include <TNL/Matrices/CSR.h>
#include <TNL/Containers/Vector.h>
-#include <TNL/Containers/SharedVector.h>
#include <TNL/Math.h>
#ifdef HAVE_CUSPARSE
@@ -61,30 +60,46 @@ String CSR< Real, Device, Index >::getTypeVirtual() const
template< typename Real,
typename Device,
typename Index >
-bool CSR< Real, Device, Index >::setDimensions( const IndexType rows,
- const IndexType columns )
+String CSR< Real, Device, Index >::getSerializationType()
{
- if( ! Sparse< Real, Device, Index >::setDimensions( rows, columns ) ||
- ! this->rowPointers.setSize( this->rows + 1 ) )
- return false;
+ return HostType::getType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+String CSR< Real, Device, Index >::getSerializationTypeVirtual() const
+{
+ return this->getSerializationType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+void CSR< Real, Device, Index >::setDimensions( const IndexType rows,
+ const IndexType columns )
+{
+ Sparse< Real, Device, Index >::setDimensions( rows, columns );
+ this->rowPointers.setSize( this->rows + 1 );
this->rowPointers.setValue( 0 );
- return true;
}
template< typename Real,
typename Device,
typename Index >
-bool CSR< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
+void CSR< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
{
+ TNL_ASSERT_GT( this->getRows(), 0, "cannot set row lengths of an empty matrix" );
+ TNL_ASSERT_GT( this->getColumns(), 0, "cannot set row lengths of an empty matrix" );
+ TNL_ASSERT_EQ( this->getRows(), rowLengths.getSize(), "wrong size of the rowLengths vector" );
+
/****
* Compute the rows pointers. The last one is
* the end of the last row and so it says the
* necessary length of the vectors this->values
* and this->columnIndexes.
*/
- TNL_ASSERT( this->getRows() > 0, );
- TNL_ASSERT( this->getColumns() > 0, );
- Containers::SharedVector< IndexType, DeviceType, IndexType > rowPtrs;
+ Containers::Vector< IndexType, DeviceType, IndexType > rowPtrs;
rowPtrs.bind( this->rowPointers.getData(), this->getRows() );
rowPtrs = rowLengths;
this->rowPointers.setElement( this->rows, 0 );
@@ -94,17 +109,24 @@ bool CSR< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLen
/****
* Allocate values and column indexes
*/
- if( ! this->values.setSize( this->rowPointers.getElement( this->rows ) ) ||
- ! this->columnIndexes.setSize( this->rowPointers.getElement( this->rows ) ) )
- return false;
+ this->values.setSize( this->rowPointers.getElement( this->rows ) );
+ this->columnIndexes.setSize( this->rowPointers.getElement( this->rows ) );
this->columnIndexes.setValue( this->columns );
- return true;
}
template< typename Real,
typename Device,
typename Index >
Index CSR< Real, Device, Index >::getRowLength( const IndexType row ) const
+{
+ return this->rowPointers.getElement( row + 1 ) - this->rowPointers.getElement( row );
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+__cuda_callable__
+Index CSR< Real, Device, Index >::getRowLengthFast( const IndexType row ) const
{
return this->rowPointers[ row + 1 ] - this->rowPointers[ row ];
}
@@ -115,12 +137,10 @@ template< typename Real,
template< typename Real2,
typename Device2,
typename Index2 >
-bool CSR< Real, Device, Index >::setLike( const CSR< Real2, Device2, Index2 >& matrix )
+void CSR< Real, Device, Index >::setLike( const CSR< Real2, Device2, Index2 >& matrix )
{
- if( ! Sparse< Real, Device, Index >::setLike( matrix ) ||
- ! this->rowPointers.setLike( matrix.rowPointers ) )
- return false;
- return true;
+ Sparse< Real, Device, Index >::setLike( matrix );
+ this->rowPointers.setLike( matrix.rowPointers );
}
template< typename Real,
@@ -271,7 +291,7 @@ bool CSR< Real, Device, Index > :: setRowFast( const IndexType row,
for( IndexType i = 0; i < elements; i++ )
{
- printf( "Setting element row: %d column: %d value: %f \n", row, columnIndexes[ i ], values[ i ] );
+ //printf( "Setting element row: %d column: %d value: %f \n", row, columnIndexes[ i ], values[ i ] );
this->columnIndexes[ elementPointer ] = columnIndexes[ i ];
this->values[ elementPointer ] = values[ i ];
elementPointer++;
@@ -508,6 +528,36 @@ bool CSR< Real, Device, Index >::performSORIteration( const Vector& b,
}
+// copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+CSR< Real, Device, Index >&
+CSR< Real, Device, Index >::operator=( const CSR& matrix )
+{
+ this->setLike( matrix );
+ this->values = matrix.values;
+ this->columnIndexes = matrix.columnIndexes;
+ this->rowPointers = matrix.rowPointers;
+ return *this;
+}
+
+// cross-device copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+ template< typename Real2, typename Device2, typename Index2, typename >
+CSR< Real, Device, Index >&
+CSR< Real, Device, Index >::operator=( const CSR< Real2, Device2, Index2 >& matrix )
+{
+ this->setLike( matrix );
+ this->values = matrix.values;
+ this->columnIndexes = matrix.columnIndexes;
+ this->rowPointers = matrix.rowPointers;
+ return *this;
+}
+
+
template< typename Real,
typename Device,
typename Index >
@@ -731,6 +781,38 @@ class CSRDeviceDependentCode< Devices::Host >
};
+#ifdef HAVE_MIC
+template<>
+class CSRDeviceDependentCode< Devices::MIC >
+{
+ public:
+
+ typedef Devices::MIC Device;
+
+ template< typename Real,
+ typename Index,
+ typename InVector,
+ typename OutVector >
+ static void vectorProduct( const CSR< Real, Device, Index >& matrix,
+ const InVector& inVector,
+ OutVector& outVector )
+ {
+ cout <<"Not Implemented YET tnlCSRMatrixDeviceDependentCode for MIC" <<endl;
+ };
+ /* const Index rows = matrix.getRows();
+ const tnlCSRMatrix< Real, Device, Index >* matrixPtr = &matrix;
+ const InVector* inVectorPtr = &inVector;
+ OutVector* outVectorPtr = &outVector;
+#ifdef HAVE_OPENMP
+#pragma omp parallel for firstprivate( matrixPtr, inVectorPtr, outVectorPtr ), schedule(static ), if( tnlHost::isOMPEnabled() )
+#endif
+ for( Index row = 0; row < rows; row ++ )
+ ( *outVectorPtr )[ row ] = matrixPtr->rowVectorProduct( row, *inVectorPtr );
+ }*/
+
+};
+#endif
+
#ifdef HAVE_CUDA
template< typename Real,
typename Index,
@@ -773,7 +855,7 @@ void CSRVectorProductCuda( const CSR< Real, Devices::Cuda, Index >& matrix,
Matrix* kernel_this = Devices::Cuda::passToDevice( matrix );
InVector* kernel_inVector = Devices::Cuda::passToDevice( inVector );
OutVector* kernel_outVector = Devices::Cuda::passToDevice( outVector );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( Devices::Cuda::getMaxGridSize() );
const IndexType cudaBlocks = roundUpDivision( matrix.getRows(), cudaBlockSize.x );
const IndexType cudaGrids = roundUpDivision( cudaBlocks, Devices::Cuda::getMaxGridSize() );
@@ -826,11 +908,11 @@ void CSRVectorProductCuda( const CSR< Real, Devices::Cuda, Index >& matrix,
gridIdx );
}
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
Devices::Cuda::freeFromDevice( kernel_this );
Devices::Cuda::freeFromDevice( kernel_inVector );
Devices::Cuda::freeFromDevice( kernel_outVector );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
diff --git a/src/TNL/Matrices/ChunkedEllpack.h b/src/TNL/Matrices/ChunkedEllpack.h
index 299f85f5d6b95f4833153a4fc403c92c9603570b..ba14092163815ca30e7f296705d00013ed84e124 100644
--- a/src/TNL/Matrices/ChunkedEllpack.h
+++ b/src/TNL/Matrices/ChunkedEllpack.h
@@ -62,8 +62,16 @@ __global__ void ChunkedEllpackVectorProductCudaKernel( const ChunkedEllpack< Rea
template< typename Real, typename Device, typename Index >
class ChunkedEllpack : public Sparse< Real, Device, Index >
{
- public:
+private:
+ // convenient template alias for controlling the selection of copy-assignment operator
+ template< typename Device2 >
+ using Enabler = std::enable_if< ! std::is_same< Device2, Device >::value >;
+ // friend class will be needed for templated assignment operators
+ template< typename Real2, typename Device2, typename Index2 >
+ friend class ChunkedEllpack;
+
+public:
typedef Real RealType;
typedef Device DeviceType;
typedef Index IndexType;
@@ -74,6 +82,7 @@ class ChunkedEllpack : public Sparse< Real, Device, Index >
typedef ChunkedEllpack< Real, Devices::Cuda, Index > CudaType;
typedef Sparse< Real, Device, Index > BaseType;
typedef typename BaseType::MatrixRow MatrixRow;
+ typedef SparseRow< const RealType, const IndexType > ConstMatrixRow;
ChunkedEllpack();
@@ -81,15 +90,22 @@ class ChunkedEllpack : public Sparse< Real, Device, Index >
String getTypeVirtual() const;
- bool setDimensions( const IndexType rows,
+ static String getSerializationType();
+
+ virtual String getSerializationTypeVirtual() const;
+
+ void setDimensions( const IndexType rows,
const IndexType columns );
- bool setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
+ void setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
IndexType getRowLength( const IndexType row ) const;
+ __cuda_callable__
+ IndexType getRowLengthFast( const IndexType row ) const;
+
template< typename Real2, typename Device2, typename Index2 >
- bool setLike( const ChunkedEllpack< Real2, Device2, Index2 >& matrix );
+ void setLike( const ChunkedEllpack< Real2, Device2, Index2 >& matrix );
void reset();
@@ -178,7 +194,7 @@ class ChunkedEllpack : public Sparse< Real, Device, Index >
MatrixRow getRow( const IndexType rowIndex );
__cuda_callable__
- const MatrixRow getRow( const IndexType rowIndex ) const;
+ ConstMatrixRow getRow( const IndexType rowIndex ) const;
template< typename Vector >
__cuda_callable__
@@ -214,6 +230,14 @@ class ChunkedEllpack : public Sparse< Real, Device, Index >
Vector& x,
const RealType& omega = 1.0 ) const;
+ // copy assignment
+ ChunkedEllpack& operator=( const ChunkedEllpack& matrix );
+
+ // cross-device copy assignment
+ template< typename Real2, typename Device2, typename Index2,
+ typename = typename Enabler< Device2 >::type >
+ ChunkedEllpack& operator=( const ChunkedEllpack< Real2, Device2, Index2 >& matrix );
+
bool save( File& file ) const;
bool load( File& file );
@@ -227,8 +251,7 @@ class ChunkedEllpack : public Sparse< Real, Device, Index >
void printStructure( std::ostream& str,
const String& = "" ) const;
- protected:
-
+protected:
void resolveSliceSizes( const Containers::Vector< Index, Devices::Host, Index >& rowLengths );
diff --git a/src/TNL/Matrices/ChunkedEllpack_impl.h b/src/TNL/Matrices/ChunkedEllpack_impl.h
index 6addfa7a3241f29d2eec07b2fc8b9ece99b917f7..5c5c71543c82c10cc2da55728fd7352d83b943b0 100644
--- a/src/TNL/Matrices/ChunkedEllpack_impl.h
+++ b/src/TNL/Matrices/ChunkedEllpack_impl.h
@@ -14,10 +14,6 @@
#include <TNL/Containers/Vector.h>
#include <TNL/Math.h>
-#ifdef HAVE_CUDA
-#include <cuda.h>
-#endif
-
namespace TNL {
namespace Matrices {
@@ -62,25 +58,38 @@ String ChunkedEllpack< Real, Device, Index >::getTypeVirtual() const
template< typename Real,
typename Device,
typename Index >
-bool ChunkedEllpack< Real, Device, Index >::setDimensions( const IndexType rows,
- const IndexType columns )
+String ChunkedEllpack< Real, Device, Index >::getSerializationType()
+{
+ return getType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+String ChunkedEllpack< Real, Device, Index >::getSerializationTypeVirtual() const
+{
+ return this->getSerializationType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+void ChunkedEllpack< Real, Device, Index >::setDimensions( const IndexType rows,
+ const IndexType columns )
{
TNL_ASSERT( rows > 0 && columns > 0,
std::cerr << "rows = " << rows
<< " columns = " << columns << std::endl );
- if( ! Sparse< Real, Device, Index >::setDimensions( rows, columns ) )
- return false;
+ Sparse< Real, Device, Index >::setDimensions( rows, columns );
/****
* Allocate slice info array. Note that there cannot be
* more slices than rows.
*/
- if( ! this->slices.setSize( this->rows ) ||
- ! this->rowToChunkMapping.setSize( this-> rows ) ||
- ! this->rowToSliceMapping.setSize( this->rows ) ||
- ! this->rowPointers.setSize( this->rows + 1 ) )
- return false;
- return true;
+ this->slices.setSize( this->rows );
+ this->rowToChunkMapping.setSize( this-> rows );
+ this->rowToSliceMapping.setSize( this->rows );
+ this->rowPointers.setSize( this->rows + 1 );
}
template< typename Real,
@@ -206,10 +215,11 @@ bool ChunkedEllpack< Real, Device, Index >::setSlice( const CompressedRowLengths
template< typename Real,
typename Device,
typename Index >
-bool ChunkedEllpack< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
+void ChunkedEllpack< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
{
- TNL_ASSERT( this->getRows() > 0, );
- TNL_ASSERT( this->getColumns() > 0, );
+ TNL_ASSERT_GT( this->getRows(), 0, "cannot set row lengths of an empty matrix" );
+ TNL_ASSERT_GT( this->getColumns(), 0, "cannot set row lengths of an empty matrix" );
+ TNL_ASSERT_EQ( this->getRows(), rowLengths.getSize(), "wrong size of the rowLengths vector" );
IndexType elementsToAllocation( 0 );
@@ -245,7 +255,7 @@ bool ChunkedEllpack< Real, Device, Index >::setCompressedRowLengths( const Compr
elementsToAllocation = hostMatrix.values.getSize();
}
this->maxRowLength = rowLengths.max();
- return Sparse< Real, Device, Index >::allocateMatrixElements( elementsToAllocation );
+ Sparse< Real, Device, Index >::allocateMatrixElements( elementsToAllocation );
}
template< typename Real,
@@ -253,9 +263,21 @@ template< typename Real,
typename Index >
Index ChunkedEllpack< Real, Device, Index >::getRowLength( const IndexType row ) const
{
- const IndexType& sliceIndex = rowToSliceMapping[ row ];
+ const IndexType& sliceIndex = rowToSliceMapping.getElement( row );
TNL_ASSERT( sliceIndex < this->rows, );
const IndexType& chunkSize = slices.getElement( sliceIndex ).chunkSize;
+ return rowPointers.getElement( row + 1 ) - rowPointers.getElement( row );
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+__cuda_callable__
+Index ChunkedEllpack< Real, Device, Index >::getRowLengthFast( const IndexType row ) const
+{
+ const IndexType& sliceIndex = rowToSliceMapping[ row ];
+ TNL_ASSERT( sliceIndex < this->rows, );
+ const IndexType& chunkSize = slices[ sliceIndex ].chunkSize;
return rowPointers[ row + 1 ] - rowPointers[ row ];
}
@@ -265,16 +287,14 @@ template< typename Real,
template< typename Real2,
typename Device2,
typename Index2 >
-bool ChunkedEllpack< Real, Device, Index >::setLike( const ChunkedEllpack< Real2, Device2, Index2 >& matrix )
+void ChunkedEllpack< Real, Device, Index >::setLike( const ChunkedEllpack< Real2, Device2, Index2 >& matrix )
{
this->chunksInSlice = matrix.chunksInSlice;
this->desiredChunkSize = matrix.desiredChunkSize;
- if( ! Sparse< Real, Device, Index >::setLike( matrix ) ||
- ! this->rowToChunkMapping.setLike( matrix.rowToChunkMapping ) ||
- ! this->rowToSliceMapping.setLike( matrix.rowToSliceMapping ) ||
- ! this->slices.setLike( matrix.slices ) )
- return false;
- return true;
+ Sparse< Real, Device, Index >::setLike( matrix );
+ this->rowToChunkMapping.setLike( matrix.rowToChunkMapping );
+ this->rowToSliceMapping.setLike( matrix.rowToSliceMapping );
+ this->slices.setLike( matrix.slices );
}
template< typename Real,
@@ -916,7 +936,7 @@ getRow( const IndexType rowIndex )
{
const IndexType rowOffset = this->rowPointers[ rowIndex ];
const IndexType rowLength = this->rowPointers[ rowIndex + 1 ] - rowOffset;
- return MatrixRow( &this->columns[ rowOffset ],
+ return MatrixRow( &this->columnIndexes[ rowOffset ],
&this->values[ rowOffset ],
rowLength,
1 );
@@ -926,13 +946,13 @@ template< typename Real,
typename Device,
typename Index >
__cuda_callable__
-const typename ChunkedEllpack< Real, Device, Index >::MatrixRow
+typename ChunkedEllpack< Real, Device, Index >::ConstMatrixRow
ChunkedEllpack< Real, Device, Index >::
getRow( const IndexType rowIndex ) const
{
const IndexType rowOffset = this->rowPointers[ rowIndex ];
const IndexType rowLength = this->rowPointers[ rowIndex + 1 ] - rowOffset;
- return MatrixRow( &this->columns[ rowOffset ],
+ return MatrixRow( &this->columnIndexes[ rowOffset ],
&this->values[ rowOffset ],
rowLength,
1 );
@@ -1171,6 +1191,46 @@ bool ChunkedEllpack< Real, Device, Index >::performSORIteration( const Vector& b
}
+// copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+ChunkedEllpack< Real, Device, Index >&
+ChunkedEllpack< Real, Device, Index >::operator=( const ChunkedEllpack& matrix )
+{
+ this->setLike( matrix );
+ this->values = matrix.values;
+ this->columnIndexes = matrix.columnIndexes;
+ this->chunksInSlice = matrix.chunksInSlice;
+ this->desiredChunkSize = matrix.desiredChunkSize;
+ this->rowToChunkMapping = matrix.rowToChunkMapping;
+ this->rowToSliceMapping = matrix.rowToSliceMapping;
+ this->rowPointers = matrix.rowPointers;
+ this->slices = matrix.slices;
+ this->numberOfSlices = matrix.numberOfSlices;
+ return *this;
+}
+
+// cross-device copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+ template< typename Real2, typename Device2, typename Index2, typename >
+ChunkedEllpack< Real, Device, Index >&
+ChunkedEllpack< Real, Device, Index >::operator=( const ChunkedEllpack< Real2, Device2, Index2 >& matrix )
+{
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::Cuda >::value,
+ "unknown device" );
+ static_assert( std::is_same< Device2, Devices::Host >::value || std::is_same< Device2, Devices::Cuda >::value,
+ "unknown device" );
+
+ this->setLike( matrix );
+
+ std::cerr << "Cross-device assignment for the ChunkedEllpack format is not implemented yet." << std::endl;
+ throw 1;
+}
+
+
template< typename Real,
typename Device,
typename Index >
@@ -1395,7 +1455,7 @@ class ChunkedEllpackDeviceDependentCode< Devices::Cuda >
Devices::Cuda::freeFromDevice( kernel_this );
Devices::Cuda::freeFromDevice( kernel_inVector );
Devices::Cuda::freeFromDevice( kernel_outVector );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
diff --git a/src/TNL/Matrices/Dense.h b/src/TNL/Matrices/Dense.h
index 10145c1b9240238feff4a826832021dabc204b18..3904f5c059b210ae9b61aa0f1455d4a2ca762964 100644
--- a/src/TNL/Matrices/Dense.h
+++ b/src/TNL/Matrices/Dense.h
@@ -26,8 +26,16 @@ template< typename Real = double,
typename Index = int >
class Dense : public Matrix< Real, Device, Index >
{
- public:
+private:
+ // convenient template alias for controlling the selection of copy-assignment operator
+ template< typename Device2 >
+ using Enabler = std::enable_if< ! std::is_same< Device2, Device >::value >;
+ // friend class will be needed for templated assignment operators
+ template< typename Real2, typename Device2, typename Index2 >
+ friend class Dense;
+
+public:
typedef Real RealType;
typedef Device DeviceType;
typedef Index IndexType;
@@ -45,16 +53,20 @@ class Dense : public Matrix< Real, Device, Index >
String getTypeVirtual() const;
- bool setDimensions( const IndexType rows,
+ static String getSerializationType();
+
+ virtual String getSerializationTypeVirtual() const;
+
+ void setDimensions( const IndexType rows,
const IndexType columns );
template< typename Real2, typename Device2, typename Index2 >
- bool setLike( const Dense< Real2, Device2, Index2 >& matrix );
+ void setLike( const Dense< Real2, Device2, Index2 >& matrix );
/****
* This method is only for the compatibility with the sparse matrices.
*/
- bool setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
+ void setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
/****
* Returns maximal number of the nonzero matrix elements that can be stored
@@ -62,6 +74,9 @@ class Dense : public Matrix< Real, Device, Index >
*/
IndexType getRowLength( const IndexType row ) const;
+ __cuda_callable__
+ IndexType getRowLengthFast( const IndexType row ) const;
+
IndexType getMaxRowLength() const;
IndexType getNumberOfMatrixElements() const;
@@ -152,29 +167,15 @@ class Dense : public Matrix< Real, Device, Index >
const RealType& matrixMultiplicator = 1.0,
const RealType& thisMatrixMultiplicator = 1.0 );
-#ifdef HAVE_NOT_CXX11
- template< typename Matrix1, typename Matrix2, int tileDim >
- void getMatrixProduct( const Matrix1& matrix1,
- const Matrix2& matrix2,
- const RealType& matrix1Multiplicator = 1.0,
- const RealType& matrix2Multiplicator = 1.0 );
-#else
template< typename Matrix1, typename Matrix2, int tileDim = 32 >
void getMatrixProduct( const Matrix1& matrix1,
const Matrix2& matrix2,
const RealType& matrix1Multiplicator = 1.0,
const RealType& matrix2Multiplicator = 1.0 );
-#endif
-#ifdef HAVE_NOT_CXX11
- template< typename Matrix, int tileDim >
- void getTransposition( const Matrix& matrix,
- const RealType& matrixMultiplicator = 1.0 );
-#else
template< typename Matrix, int tileDim = 32 >
void getTransposition( const Matrix& matrix,
const RealType& matrixMultiplicator = 1.0 );
-#endif
template< typename Vector >
void performSORIteration( const Vector& b,
@@ -182,6 +183,14 @@ class Dense : public Matrix< Real, Device, Index >
Vector& x,
const RealType& omega = 1.0 ) const;
+ // copy assignment
+ Dense& operator=( const Dense& matrix );
+
+ // cross-device copy assignment
+ template< typename Real2, typename Device2, typename Index2,
+ typename = typename Enabler< Device2 >::type >
+ Dense& operator=( const Dense< Real2, Device2, Index2 >& matrix );
+
bool save( const String& fileName ) const;
bool load( const String& fileName );
@@ -192,7 +201,7 @@ class Dense : public Matrix< Real, Device, Index >
void print( std::ostream& str ) const;
- protected:
+protected:
__cuda_callable__
IndexType getElementIndex( const IndexType row,
@@ -200,11 +209,9 @@ class Dense : public Matrix< Real, Device, Index >
typedef DenseDeviceDependentCode< DeviceType > DeviceDependentCode;
friend class DenseDeviceDependentCode< DeviceType >;
-
};
} // namespace Matrices
} // namespace TNL
#include <TNL/Matrices/Dense_impl.h>
-
diff --git a/src/TNL/Matrices/Dense_impl.h b/src/TNL/Matrices/Dense_impl.h
index aed56985ee61cf0b7c4b342e6f5079fd13c50bd6..bb146105eb4d903dd51b105bc5db49da74c4de44 100644
--- a/src/TNL/Matrices/Dense_impl.h
+++ b/src/TNL/Matrices/Dense_impl.h
@@ -49,14 +49,28 @@ String Dense< Real, Device, Index >::getTypeVirtual() const
template< typename Real,
typename Device,
typename Index >
-bool Dense< Real, Device, Index >::setDimensions( const IndexType rows,
- const IndexType columns )
+String Dense< Real, Device, Index >::getSerializationType()
{
- if( ! Matrix< Real, Device, Index >::setDimensions( rows, columns ) ||
- ! this->values.setSize( rows * columns ) )
- return false;
+ return getType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+String Dense< Real, Device, Index >::getSerializationTypeVirtual() const
+{
+ return this->getSerializationType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+void Dense< Real, Device, Index >::setDimensions( const IndexType rows,
+ const IndexType columns )
+{
+ Matrix< Real, Device, Index >::setDimensions( rows, columns );
+ this->values.setSize( rows * columns );
this->values.setValue( 0.0 );
- return true;
}
template< typename Real,
@@ -65,17 +79,16 @@ template< typename Real,
template< typename Real2,
typename Device2,
typename Index2 >
-bool Dense< Real, Device, Index >::setLike( const Dense< Real2, Device2, Index2 >& matrix )
+void Dense< Real, Device, Index >::setLike( const Dense< Real2, Device2, Index2 >& matrix )
{
- return this->setDimensions( matrix.getRows(), matrix.getColumns() );
+ this->setDimensions( matrix.getRows(), matrix.getColumns() );
}
template< typename Real,
typename Device,
typename Index >
-bool Dense< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
+void Dense< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
{
- return true;
}
template< typename Real,
@@ -86,6 +99,15 @@ Index Dense< Real, Device, Index >::getRowLength( const IndexType row ) const
return this->getColumns();
}
+template< typename Real,
+ typename Device,
+ typename Index >
+__cuda_callable__
+Index Dense< Real, Device, Index >::getRowLengthFast( const IndexType row ) const
+{
+ return this->getColumns();
+}
+
template< typename Real,
typename Device,
typename Index >
@@ -818,7 +840,7 @@ void Dense< Real, Device, Index >::getTransposition( const Matrix& matrix,
gridIdx_x,
gridIdx_y );
}
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
Devices::Cuda::freeFromDevice( this_device );
Devices::Cuda::freeFromDevice( matrix_device );
@@ -846,6 +868,39 @@ void Dense< Real, Device, Index >::performSORIteration( const Vector& b,
x[ row ] = ( 1.0 - omega ) * x[ row ] + omega / diagonalValue * ( b[ row ] - sum );
}
+
+// copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+Dense< Real, Device, Index >&
+Dense< Real, Device, Index >::operator=( const Dense& matrix )
+{
+ this->setLike( matrix );
+ this->values = matrix.values;
+ return *this;
+}
+
+// cross-device copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+ template< typename Real2, typename Device2, typename Index2, typename >
+Dense< Real, Device, Index >&
+Dense< Real, Device, Index >::operator=( const Dense< Real2, Device2, Index2 >& matrix )
+{
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::Cuda >::value,
+ "unknown device" );
+ static_assert( std::is_same< Device2, Devices::Host >::value || std::is_same< Device2, Devices::Cuda >::value,
+ "unknown device" );
+
+ this->setLike( matrix );
+
+ std::cerr << "Cross-device assignment for the Dense format is not implemented yet." << std::endl;
+ throw 1;
+}
+
+
template< typename Real,
typename Device,
typename Index >
diff --git a/src/TNL/Matrices/Ellpack.h b/src/TNL/Matrices/Ellpack.h
index 0082d414dcc545875c1d8473526ec5fbfbd5cd0b..b284f02adb5bd20d7311fc0d4016e20eda0509de 100644
--- a/src/TNL/Matrices/Ellpack.h
+++ b/src/TNL/Matrices/Ellpack.h
@@ -22,8 +22,16 @@ class EllpackDeviceDependentCode;
template< typename Real, typename Device = Devices::Host, typename Index = int >
class Ellpack : public Sparse< Real, Device, Index >
{
- public:
+private:
+ // convenient template alias for controlling the selection of copy-assignment operator
+ template< typename Device2 >
+ using Enabler = std::enable_if< ! std::is_same< Device2, Device >::value >;
+ // friend class will be needed for templated assignment operators
+ template< typename Real2, typename Device2, typename Index2 >
+ friend class Ellpack;
+
+public:
typedef Real RealType;
typedef Device DeviceType;
typedef Index IndexType;
@@ -43,17 +51,24 @@ class Ellpack : public Sparse< Real, Device, Index >
String getTypeVirtual() const;
- bool setDimensions( const IndexType rows,
+ static String getSerializationType();
+
+ virtual String getSerializationTypeVirtual() const;
+
+ void setDimensions( const IndexType rows,
const IndexType columns );
- bool setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
+ void setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
- bool setConstantCompressedRowLengths( const IndexType& rowLengths );
+ void setConstantCompressedRowLengths( const IndexType& rowLengths );
IndexType getRowLength( const IndexType row ) const;
+ __cuda_callable__
+ IndexType getRowLengthFast( const IndexType row ) const;
+
template< typename Real2, typename Device2, typename Index2 >
- bool setLike( const Ellpack< Real2, Device2, Index2 >& matrix );
+ void setLike( const Ellpack< Real2, Device2, Index2 >& matrix );
void reset();
@@ -63,10 +78,6 @@ class Ellpack : public Sparse< Real, Device, Index >
template< typename Real2, typename Device2, typename Index2 >
bool operator != ( const Ellpack< Real2, Device2, Index2 >& matrix ) const;
- /*template< typename Matrix >
- bool copyFrom( const Matrix& matrix,
- const CompressedRowLengthsVector& rowLengths );*/
-
__cuda_callable__
bool setElementFast( const IndexType row,
const IndexType column,
@@ -156,6 +167,14 @@ class Ellpack : public Sparse< Real, Device, Index >
Vector& x,
const RealType& omega = 1.0 ) const;
+ // copy assignment
+ Ellpack& operator=( const Ellpack& matrix );
+
+ // cross-device copy assignment
+ template< typename Real2, typename Device2, typename Index2,
+ typename = typename Enabler< Device2 >::type >
+ Ellpack& operator=( const Ellpack< Real2, Device2, Index2 >& matrix );
+
bool save( File& file ) const;
bool load( File& file );
@@ -166,9 +185,9 @@ class Ellpack : public Sparse< Real, Device, Index >
void print( std::ostream& str ) const;
- protected:
+protected:
- bool allocateElements();
+ void allocateElements();
IndexType rowLengths, alignedRows;
diff --git a/src/TNL/Matrices/Ellpack_impl.h b/src/TNL/Matrices/Ellpack_impl.h
index 6a98fd86b280a6a30c94fc6bf2fa641f064cb0b8..4055515b35e3687a3a01688a95b0858da4b1cbc7 100644
--- a/src/TNL/Matrices/Ellpack_impl.h
+++ b/src/TNL/Matrices/Ellpack_impl.h
@@ -50,8 +50,24 @@ String Ellpack< Real, Device, Index >::getTypeVirtual() const
template< typename Real,
typename Device,
typename Index >
-bool Ellpack< Real, Device, Index >::setDimensions( const IndexType rows,
- const IndexType columns )
+String Ellpack< Real, Device, Index >::getSerializationType()
+{
+ return getType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+String Ellpack< Real, Device, Index >::getSerializationTypeVirtual() const
+{
+ return this->getSerializationType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+void Ellpack< Real, Device, Index >::setDimensions( const IndexType rows,
+ const IndexType columns )
{
TNL_ASSERT( rows > 0 && columns > 0,
std::cerr << "rows = " << rows
@@ -62,33 +78,32 @@ bool Ellpack< Real, Device, Index >::setDimensions( const IndexType rows,
this->alignedRows = roundToMultiple( columns, Devices::Cuda::getWarpSize() );
else this->alignedRows = rows;
if( this->rowLengths != 0 )
- return allocateElements();
- return true;
+ allocateElements();
}
template< typename Real,
typename Device,
typename Index >
-bool Ellpack< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
+void Ellpack< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
{
- TNL_ASSERT( this->getRows() > 0, );
- TNL_ASSERT( this->getColumns() > 0, );
- TNL_ASSERT( rowLengths.getSize() > 0, );
+ TNL_ASSERT_GT( this->getRows(), 0, "cannot set row lengths of an empty matrix" );
+ TNL_ASSERT_GT( this->getColumns(), 0, "cannot set row lengths of an empty matrix" );
+ TNL_ASSERT_EQ( this->getRows(), rowLengths.getSize(), "wrong size of the rowLengths vector" );
+
this->rowLengths = this->maxRowLength = rowLengths.max();
- return allocateElements();
+ allocateElements();
}
template< typename Real,
typename Device,
typename Index >
-bool Ellpack< Real, Device, Index >::setConstantCompressedRowLengths( const IndexType& rowLengths )
+void Ellpack< Real, Device, Index >::setConstantCompressedRowLengths( const IndexType& rowLengths )
{
TNL_ASSERT( rowLengths > 0,
std::cerr << " rowLengths = " << rowLengths );
this->rowLengths = rowLengths;
if( this->rows > 0 )
- return allocateElements();
- return true;
+ allocateElements();
}
template< typename Real,
@@ -99,19 +114,26 @@ Index Ellpack< Real, Device, Index >::getRowLength( const IndexType row ) const
return this->rowLengths;
}
+template< typename Real,
+ typename Device,
+ typename Index >
+__cuda_callable__
+Index Ellpack< Real, Device, Index >::getRowLengthFast( const IndexType row ) const
+{
+ return this->rowLengths;
+}
+
template< typename Real,
typename Device,
typename Index >
template< typename Real2,
typename Device2,
typename Index2 >
-bool Ellpack< Real, Device, Index >::setLike( const Ellpack< Real2, Device2, Index2 >& matrix )
+void Ellpack< Real, Device, Index >::setLike( const Ellpack< Real2, Device2, Index2 >& matrix )
{
- if( ! Sparse< Real, Device, Index >::setLike( matrix ) )
- return false;
+ Sparse< Real, Device, Index >::setLike( matrix );
this->rowLengths = matrix.rowLengths;
this->alignedRows = matrix.alignedRows;
- return true;
}
template< typename Real,
@@ -153,16 +175,6 @@ bool Ellpack< Real, Device, Index >::operator != ( const Ellpack< Real2, Device2
return ! ( ( *this ) == matrix );
}
-/*template< typename Real,
- typename Device,
- typename Index >
- template< typename Matrix >
-bool Ellpack< Real, Device, Index >::copyFrom( const Matrix& matrix,
- const CompressedRowLengthsVector& rowLengths )
-{
- return Matrix< RealType, DeviceType, IndexType >::copyFrom( matrix, rowLengths );
-}*/
-
template< typename Real,
typename Device,
typename Index >
@@ -564,17 +576,95 @@ bool Ellpack< Real, Device, Index > :: performSORIteration( const Vector& b,
}
+// copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+Ellpack< Real, Device, Index >&
+Ellpack< Real, Device, Index >::operator=( const Ellpack& matrix )
+{
+ this->setLike( matrix );
+ this->values = matrix.values;
+ this->columnIndexes = matrix.columnIndexes;
+ return *this;
+}
+
+// cross-device copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+ template< typename Real2, typename Device2, typename Index2, typename >
+Ellpack< Real, Device, Index >&
+Ellpack< Real, Device, Index >::operator=( const Ellpack< Real2, Device2, Index2 >& matrix )
+{
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::Cuda >::value,
+ "unknown device" );
+ static_assert( std::is_same< Device2, Devices::Host >::value || std::is_same< Device2, Devices::Cuda >::value,
+ "unknown device" );
+
+ // setLike does not work here due to different alignment on Cuda and Host
+ this->rowLengths = matrix.rowLengths;
+ this->setDimensions( matrix.getRows(), matrix.getColumns() );
+
+ const int blockSize = 32;
+ const int blocks = roundUpDivision( this->getRows(), blockSize );
+
+ // host -> cuda
+ if( std::is_same< Device, Devices::Cuda >::value ) {
+ typename ValuesVector::HostType tmpValues;
+ typename ColumnIndexesVector::HostType tmpColumnIndexes;
+ tmpValues.setLike( this->values );
+ tmpColumnIndexes.setLike( this->columnIndexes );
+
+#ifdef HAVE_OPENMP
+#pragma omp parallel for if( Devices::Host::isOMPEnabled() )
+#endif
+ for( Index b = 0; b < blocks; b++ ) {
+ const Index offset = b * blockSize;
+ for( Index j = 0; j < rowLengths; j++ )
+ for( Index i = 0; i < blockSize && offset + i < this->getRows(); i++ ) {
+ tmpValues[ offset + j * alignedRows + i ] = matrix.values[ ( offset + i ) * rowLengths + j ];
+ tmpColumnIndexes[ offset + j * alignedRows + i ] = matrix.columnIndexes[ ( offset + i ) * rowLengths + j ];
+ }
+ }
+
+ this->values = tmpValues;
+ this->columnIndexes = tmpColumnIndexes;
+ }
+
+ // cuda -> host
+ if( std::is_same< Device, Devices::Host >::value ) {
+ ValuesVector tmpValues;
+ ColumnIndexesVector tmpColumnIndexes;
+ tmpValues.setLike( matrix.values );
+ tmpColumnIndexes.setLike( matrix.columnIndexes );
+ tmpValues = matrix.values;
+ tmpColumnIndexes = matrix.columnIndexes;
+
+#ifdef HAVE_OPENMP
+#pragma omp parallel for if( Devices::Host::isOMPEnabled() )
+#endif
+ for( Index b = 0; b < blocks; b++ ) {
+ const Index offset = b * rowLengths;
+ for( Index i = 0; i < blockSize && b * blockSize + i < this->getRows(); i++ )
+ for( Index j = 0; j < rowLengths; j++ ) {
+ this->values[ offset + i * rowLengths + j ] = tmpValues[ b * blockSize + j * matrix.alignedRows + i ];
+ this->columnIndexes[ offset + i * rowLengths + j ] = tmpColumnIndexes[ b * blockSize + j * matrix.alignedRows + i ];
+ }
+ }
+ }
+
+ return *this;
+}
+
+
template< typename Real,
typename Device,
typename Index >
bool Ellpack< Real, Device, Index >::save( File& file ) const
{
if( ! Sparse< Real, Device, Index >::save( file) ) return false;
-#ifdef HAVE_NOT_CXX11
- if( ! file.write< IndexType, Devices::Host, IndexType >( &this->rowLengths, 1 ) ) return false;
-#else
if( ! file.write( &this->rowLengths ) ) return false;
-#endif
return true;
}
@@ -584,11 +674,7 @@ template< typename Real,
bool Ellpack< Real, Device, Index >::load( File& file )
{
if( ! Sparse< Real, Device, Index >::load( file) ) return false;
-#ifdef HAVE_NOT_CXX11
- if( ! file.read< IndexType, Devices::Host, IndexType >( &this->rowLengths, 1 ) ) return false;
-#else
if( ! file.read( &this->rowLengths ) ) return false;
-#endif
return true;
}
@@ -634,11 +720,9 @@ void Ellpack< Real, Device, Index >::print( std::ostream& str ) const
template< typename Real,
typename Device,
typename Index >
-bool Ellpack< Real, Device, Index >::allocateElements()
+void Ellpack< Real, Device, Index >::allocateElements()
{
- if( ! Sparse< Real, Device, Index >::allocateMatrixElements( this->alignedRows * this->rowLengths ) )
- return false;
- return true;
+ Sparse< Real, Device, Index >::allocateMatrixElements( this->alignedRows * this->rowLengths );
}
template<>
@@ -803,12 +887,12 @@ class EllpackDeviceDependentCode< Devices::Cuda >
inVector.getData(),
outVector.getData(),
gridIdx );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
//Devices::Cuda::freeFromDevice( kernel_this );
//Devices::Cuda::freeFromDevice( kernel_inVector );
//Devices::Cuda::freeFromDevice( kernel_outVector );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cudaThreadSynchronize();
#endif
diff --git a/src/TNL/Matrices/Matrix.h b/src/TNL/Matrices/Matrix.h
index f8117165c4355068dc8cc083370f34165f74e4c2..2af4d0a890c9acf52e98019f23d0aeeb282d77cb 100644
--- a/src/TNL/Matrices/Matrix.h
+++ b/src/TNL/Matrices/Matrix.h
@@ -22,8 +22,7 @@ template< typename Real = double,
typename Index = int >
class Matrix : public virtual Object
{
- public:
-
+public:
typedef Real RealType;
typedef Device DeviceType;
typedef Index IndexType;
@@ -32,17 +31,17 @@ class Matrix : public virtual Object
Matrix();
- virtual bool setDimensions( const IndexType rows,
+ virtual void setDimensions( const IndexType rows,
const IndexType columns );
- virtual bool setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths ) = 0;
+ virtual void setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths ) = 0;
virtual IndexType getRowLength( const IndexType row ) const = 0;
virtual void getCompressedRowLengths( Containers::Vector< IndexType, DeviceType, IndexType >& rowLengths ) const;
template< typename Real2, typename Device2, typename Index2 >
- bool setLike( const Matrix< Real2, Device2, Index2 >& matrix );
+ void setLike( const Matrix< Real2, Device2, Index2 >& matrix );
virtual IndexType getNumberOfMatrixElements() const = 0;
@@ -85,30 +84,22 @@ class Matrix : public virtual Object
virtual Real getElement( const IndexType row,
const IndexType column ) const = 0;
- Matrix< RealType, DeviceType, IndexType >& operator = ( const Matrix< RealType, DeviceType, IndexType >& );
-
template< typename Matrix >
bool operator == ( const Matrix& matrix ) const;
template< typename Matrix >
bool operator != ( const Matrix& matrix ) const;
- template< typename Matrix >
- bool copyFrom( const Matrix& matrix,
- const CompressedRowLengthsVector& rowLengths );
-
virtual bool save( File& file ) const;
virtual bool load( File& file );
virtual void print( std::ostream& str ) const;
- protected:
+protected:
IndexType rows, columns;
- public: // TODO: remove this
-
ValuesVector values;
};
@@ -123,8 +114,8 @@ template< typename Matrix,
typename InVector,
typename OutVector >
void MatrixVectorProductCuda( const Matrix& matrix,
- const InVector& inVector,
- OutVector& outVector );
+ const InVector& inVector,
+ OutVector& outVector );
} // namespace Matrices
} // namespace TNL
diff --git a/src/TNL/Matrices/MatrixOperations.h b/src/TNL/Matrices/MatrixOperations.h
index 31e72fb24288a2fa799a536fef52b3a3079090a6..2c0b8d184841545e3429a1a022a315eb047eb808 100644
--- a/src/TNL/Matrices/MatrixOperations.h
+++ b/src/TNL/Matrices/MatrixOperations.h
@@ -17,6 +17,7 @@
* The algorithms should be incorporated into the Matrices::Dense class.
*/
+#include <TNL/Exceptions/CudaSupportMissing.h>
#include <TNL/Devices/Host.h>
#include <TNL/Devices/Cuda.h>
#include <TNL/Math.h>
@@ -36,6 +37,8 @@ public:
* lda >= m is the leading dimension of two-dimensional array used to store matrix A,
* x is a vector of n elements,
* y is a vector of m elements.
+ *
+ * It is assumed that n is much smaller than m.
*/
template< typename RealType,
typename IndexType >
@@ -49,31 +52,179 @@ public:
const RealType& beta,
RealType* y )
{
- TNL_ASSERT( m <= lda, );
+ TNL_ASSERT_GT( m, 0, "m must be positive" );
+ TNL_ASSERT_GT( n, 0, "n must be positive" );
+ TNL_ASSERT_GE( lda, m, "lda must be at least m" );
- if( beta != 0.0 ) {
-#ifdef HAVE_OPENMP
-#pragma omp parallel for if( TNL::Devices::Host::isOMPEnabled() )
-#endif
- for( IndexType j = 0; j < m; j++ ) {
- RealType tmp = 0.0;
- for( int k = 0; k < n; k++ )
- tmp += A[ j + k * lda ] * x[ k ];
- y[ j ] = alpha * tmp + beta * y[ j ];
+ RealType alphax[ n ];
+ for( IndexType k = 0; k < n; k++ )
+ alphax[ k ] = alpha * x[ k ];
+
+ if( n == 1 ) {
+ if( beta != 0.0 ) {
+ #ifdef HAVE_OPENMP
+ #pragma omp parallel for if( TNL::Devices::Host::isOMPEnabled() )
+ #endif
+ for( IndexType j = 0; j < m; j++ )
+ y[ j ] = A[ j ] * alphax[ 0 ] + beta * y[ j ];
+ }
+ else {
+ // the vector y might be uninitialized, and 0.0 * NaN = NaN
+ #ifdef HAVE_OPENMP
+ #pragma omp parallel for if( TNL::Devices::Host::isOMPEnabled() )
+ #endif
+ for( IndexType j = 0; j < m; j++ )
+ y[ j ] = A[ j ] * alphax[ 0 ];
}
}
else {
- // the vector y might be uninitialized, and 0.0 * NaN = NaN
-#ifdef HAVE_OPENMP
-#pragma omp parallel for if( TNL::Devices::Host::isOMPEnabled() )
-#endif
- for( IndexType j = 0; j < m; j++ ) {
- RealType tmp = 0.0;
- for( int k = 0; k < n; k++ )
- tmp += A[ j + k * lda ] * x[ k ];
- y[ j ] = alpha * tmp;
+ // the matrix A should be accessed column-wise so we split the work into small
+ // blocks and each block process by columns, either parallelly or serially
+ constexpr int block_size = 128;
+ const int blocks = m / block_size;
+
+ #ifdef HAVE_OPENMP
+ #pragma omp parallel if( TNL::Devices::Host::isOMPEnabled() && blocks >= 2 )
+ #endif
+ {
+ RealType aux[ block_size ];
+
+ #ifdef HAVE_OPENMP
+ #pragma omp for nowait
+ #endif
+ for( int b = 0; b < blocks; b++ ) {
+ const int block_offset = b * block_size;
+
+ // initialize array for thread-local results
+ for( int j = 0; j < block_size; j++ )
+ aux[ j ] = 0.0;
+
+ // compute aux = A * alphax
+ for( int k = 0; k < n; k++ ) {
+ const int offset = block_offset + k * lda;
+ for( int j = 0; j < block_size; j++ )
+ aux[ j ] += A[ offset + j ] * alphax[ k ];
+ }
+
+ // write result: y = aux + beta * y
+ if( beta != 0.0 ) {
+ for( int j = 0; j < block_size; j++ )
+ y[ block_offset + j ] = aux[ j ] + beta * y[ block_offset + j ];
+ }
+ else {
+ // the vector y might be uninitialized, and 0.0 * NaN = NaN
+ for( IndexType j = 0; j < block_size; j++ )
+ y[ block_offset + j ] = aux[ j ];
+ }
+ }
+
+ // the first thread that reaches here processes the last, incomplete block
+ #ifdef HAVE_OPENMP
+ #pragma omp single nowait
+ #endif
+ {
+ // TODO: unlike the complete blocks, the tail is traversed row-wise
+ if( beta != 0.0 ) {
+ for( IndexType j = blocks * block_size; j < m; j++ ) {
+ RealType tmp = 0.0;
+ for( int k = 0; k < n; k++ )
+ tmp += A[ j + k * lda ] * alphax[ k ];
+ y[ j ] = tmp + beta * y[ j ];
+ }
+ }
+ else {
+ // the vector y might be uninitialized, and 0.0 * NaN = NaN
+ for( IndexType j = blocks * block_size; j < m; j++ ) {
+ RealType tmp = 0.0;
+ for( int k = 0; k < n; k++ )
+ tmp += A[ j + k * lda ] * alphax[ k ];
+ y[ j ] = tmp;
+ }
+ }
+ }
+ }
+ }
+
+ }
+
+ /*
+ * This function performs the matrix-matrix addition
+ * C = alpha * A + beta * B
+ * where:
+ * alpha and beta are scalars,
+ * A, B, C are (m by n) matrices stored in column-major format on Devices::Cuda,
+ * lda, ldb, ldc (all >= m) are the leading dimensions of matrices A, B, C,
+ * respectively.
+ *
+ * It is assumed that n is much smaller than m.
+ */
+ template< typename RealType,
+ typename IndexType >
+ static void
+ geam( const IndexType& m,
+ const IndexType& n,
+ const RealType& alpha,
+ const RealType* A,
+ const IndexType& lda,
+ const RealType& beta,
+ const RealType* B,
+ const IndexType& ldb,
+ RealType* C,
+ const IndexType& ldc )
+ {
+ TNL_ASSERT_GT( m, 0, "m must be positive" );
+ TNL_ASSERT_GT( n, 0, "n must be positive" );
+ TNL_ASSERT_GE( lda, m, "lda must be at least m" );
+ TNL_ASSERT_GE( ldb, m, "lda must be at least m" );
+ TNL_ASSERT_GE( ldc, m, "lda must be at least m" );
+
+ if( n == 1 ) {
+ #ifdef HAVE_OPENMP
+ #pragma omp parallel for if( TNL::Devices::Host::isOMPEnabled() )
+ #endif
+ for( IndexType j = 0; j < m; j++ )
+ C[ j ] = alpha * A[ j ] + beta * B[ j ];
+ }
+ else {
+ // all matrices should be accessed column-wise so we split the work into small
+ // blocks and each block process by columns, either parallelly or serially
+ constexpr int block_size = 128;
+ const int blocks = m / block_size;
+
+ #ifdef HAVE_OPENMP
+ #pragma omp parallel if( TNL::Devices::Host::isOMPEnabled() && blocks >= 2 )
+ #endif
+ {
+ #ifdef HAVE_OPENMP
+ #pragma omp for nowait
+ #endif
+ for( int b = 0; b < blocks; b++ ) {
+ const int block_offset = b * block_size;
+ for( IndexType j = 0; j < n; j++ ) {
+ const IndexType offset_A = j * lda + block_offset;
+ const IndexType offset_B = j * ldb + block_offset;
+ const IndexType offset_C = j * ldc + block_offset;
+ for( int i = 0; i < block_size; i++ )
+ C[ offset_C + i ] = alpha * A[ offset_A + i ] + beta * B[ offset_B + i ];
+ }
+ }
+
+ // the first thread that reaches here processes the last, incomplete block
+ #ifdef HAVE_OPENMP
+ #pragma omp single nowait
+ #endif
+ {
+ for( IndexType j = 0; j < n; j++ ) {
+ const IndexType offset_A = j * lda;
+ const IndexType offset_B = j * ldb;
+ const IndexType offset_C = j * ldc;
+ for( IndexType i = blocks * block_size; i < m; i++ )
+ C[ offset_C + i ] = alpha * A[ offset_A + i ] + beta * B[ offset_B + i ];
+ }
+ }
}
}
+
}
};
@@ -81,12 +232,6 @@ public:
// CUDA kernels
#ifdef HAVE_CUDA
-#if (__CUDA_ARCH__ >= 300 )
- static constexpr int Gemv_minBlocksPerMultiprocessor = 8;
-#else
- static constexpr int Gemv_minBlocksPerMultiprocessor = 4;
-#endif
-
template< typename RealType,
typename IndexType >
__global__ void
@@ -105,7 +250,7 @@ GemvCudaKernel( const IndexType m,
RealType* shx = Devices::Cuda::getSharedMemory< RealType >();
if( threadIdx.x < n )
- shx[ threadIdx.x ] = x[ threadIdx.x ];
+ shx[ threadIdx.x ] = alpha * x[ threadIdx.x ];
__syncthreads();
if( beta != 0.0 ) {
@@ -113,7 +258,7 @@ GemvCudaKernel( const IndexType m,
RealType tmp = 0.0;
for( IndexType k = 0; k < n; k++ )
tmp += A[ elementIdx + k * lda ] * shx[ k ];
- y[ elementIdx ] = alpha * tmp + beta * y[ elementIdx ];
+ y[ elementIdx ] = tmp + beta * y[ elementIdx ];
elementIdx += gridSize;
}
}
@@ -123,11 +268,39 @@ GemvCudaKernel( const IndexType m,
RealType tmp = 0.0;
for( IndexType k = 0; k < n; k++ )
tmp += A[ elementIdx + k * lda ] * shx[ k ];
- y[ elementIdx ] = alpha * tmp;
+ y[ elementIdx ] = tmp;
elementIdx += gridSize;
}
}
}
+
+template< typename RealType,
+ typename IndexType >
+__global__ void
+GeamCudaKernel( const IndexType m,
+ const IndexType n,
+ const RealType alpha,
+ const RealType* A,
+ const IndexType lda,
+ const RealType beta,
+ const RealType* B,
+ const IndexType ldb,
+ RealType* C,
+ const IndexType ldc )
+{
+ IndexType x = blockIdx.x * blockDim.x + threadIdx.x;
+ const IndexType gridSizeX = blockDim.x * gridDim.x;
+ const IndexType y = blockIdx.y * blockDim.y + threadIdx.y;
+ const IndexType offset_A = y * lda;
+ const IndexType offset_B = y * ldb;
+ const IndexType offset_C = y * ldc;
+
+ if( y < n )
+ while( x < m ) {
+ C[ x + offset_C ] = alpha * A[ x + offset_A ] + beta * B[ x + offset_B ];
+ x += gridSizeX;
+ }
+}
#endif
// specialization for CUDA
@@ -144,6 +317,8 @@ public:
* lda >= m is the leading dimension of two-dimensional array used to store matrix A,
* x is a vector of n elements, stored on Devices::Host,
* y is a vector of m elements, stored on Devices::Cuda.
+ *
+ * It is assumed that n is much smaller than m.
*/
template< typename RealType,
typename IndexType >
@@ -167,19 +342,76 @@ public:
if( ! Containers::Algorithms::ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< RealType, RealType, IndexType >( xDevice.getData(), x, n ) )
throw 1;
- dim3 blockSize( 256 );
- dim3 gridSize;
- const IndexType desGridSize = 4 * Gemv_minBlocksPerMultiprocessor
- * Devices::CudaDeviceInfo::getCudaMultiprocessors( Devices::CudaDeviceInfo::getActiveDevice() );
+ // desGridSize = blocksPerMultiprocessor * numberOfMultiprocessors
+ const int desGridSize = 32 * Devices::CudaDeviceInfo::getCudaMultiprocessors( Devices::CudaDeviceInfo::getActiveDevice() );
+ dim3 blockSize, gridSize;
+ blockSize.x = 256;
gridSize.x = min( desGridSize, Devices::Cuda::getNumberOfBlocks( m, blockSize.x ) );
GemvCudaKernel<<< gridSize, blockSize, n * sizeof( RealType ) >>>(
m, n,
alpha, A, lda,
xDevice.getData(), beta, y );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
+ }
+
+ /*
+ * This function performs the matrix-matrix addition
+ * C = alpha * A + beta * B
+ * where:
+ * alpha and beta are scalars,
+ * A, B, C are (m by n) matrices stored in column-major format on Devices::Cuda,
+ * lda, ldb, ldc (all >= m) are the leading dimensions of matrices A, B, C,
+ * respectively.
+ *
+ * It is assumed that n is much smaller than m.
+ */
+ template< typename RealType,
+ typename IndexType >
+ static void
+ geam( const IndexType& m,
+ const IndexType& n,
+ const RealType& alpha,
+ const RealType* A,
+ const IndexType& lda,
+ const RealType& beta,
+ const RealType* B,
+ const IndexType& ldb,
+ RealType* C,
+ const IndexType& ldc )
+ {
+ TNL_ASSERT_GT( m, 0, "m must be positive" );
+ TNL_ASSERT_GT( n, 0, "n must be positive" );
+ TNL_ASSERT_GE( lda, m, "lda must be at least m" );
+ TNL_ASSERT_GE( ldb, m, "lda must be at least m" );
+ TNL_ASSERT_GE( ldc, m, "lda must be at least m" );
+
+#ifdef HAVE_CUDA
+ dim3 blockSize, gridSize;
+
+ // max 16 columns of threads
+ blockSize.y = min( n, 16 );
+ // max 256 threads per block, power of 2
+ blockSize.x = 256;
+ while( blockSize.x * blockSize.y > 256 )
+ blockSize.x /= 2;
+
+ // desGridSize = blocksPerMultiprocessor * numberOfMultiprocessors
+ const int desGridSize = 32 * Devices::CudaDeviceInfo::getCudaMultiprocessors( Devices::CudaDeviceInfo::getActiveDevice() );
+ gridSize.x = min( desGridSize, Devices::Cuda::getNumberOfBlocks( m, blockSize.x ) );
+ gridSize.y = Devices::Cuda::getNumberOfBlocks( n, blockSize.y );
+
+ GeamCudaKernel<<< gridSize, blockSize >>>(
+ m, n,
+ alpha, A, lda,
+ beta, B, ldb,
+ C, ldc );
+ TNL_CHECK_CUDA_DEVICE;
#else
- CudaSupportMissingMessage;
+ throw Exceptions::CudaSupportMissing();
#endif
}
};
diff --git a/src/TNL/Matrices/MatrixReader_impl.h b/src/TNL/Matrices/MatrixReader_impl.h
index 4d0058e5aa6892430f9e84248734c684ad130e8f..f768d748cd62dd5d01923254a5e246c8171b79d1 100644
--- a/src/TNL/Matrices/MatrixReader_impl.h
+++ b/src/TNL/Matrices/MatrixReader_impl.h
@@ -15,6 +15,7 @@
#include <TNL/String.h>
#include <TNL/Containers/Vector.h>
#include <TNL/Timer.h>
+#include <TNL/Matrices/MatrixReader.h>
namespace TNL {
namespace Matrices {
@@ -55,18 +56,13 @@ bool MatrixReader< Matrix >::readMtxFileHostMatrix( std::istream& file,
return false;
- if( ! matrix.setDimensions( rows, columns ) ||
- ! rowLengths.setSize( rows ) )
- {
- std::cerr << "Not enough memory to allocate the sparse or the full matrix for testing." << std::endl;
- return false;
- }
+ matrix.setDimensions( rows, columns );
+ rowLengths.setSize( rows );
if( ! computeCompressedRowLengthsFromMtxFile( file, rowLengths, columns, rows, symmetricMatrix, verbose ) )
return false;
- if( ! matrix.setCompressedRowLengths( rowLengths ) )
- return false;
+ matrix.setCompressedRowLengths( rowLengths );
if( ! readMatrixElementsFromMtxFile( file, matrix, symmetricMatrix, verbose ) )
return false;
@@ -163,7 +159,7 @@ bool MatrixReader< Matrix >::checkMtxHeader( const String& header,
bool& symmetric )
{
Containers::List< String > parsedLine;
- header.parse( parsedLine );
+ header.split( parsedLine );
if( parsedLine.getSize() < 5 )
return false;
if( parsedLine[ 0 ] != "%%MatrixMarket" )
@@ -230,7 +226,7 @@ bool MatrixReader< Matrix >::readMtxHeader( std::istream& file,
}
parsedLine.reset();
- line. parse( parsedLine );
+ line.split( parsedLine );
if( parsedLine. getSize() != 3 )
{
std::cerr << "Wrong number of parameters in the matrix header." << std::endl;
@@ -365,7 +361,7 @@ bool MatrixReader< Matrix >::parseMtxLineWithElement( const String& line,
RealType& value )
{
Containers::List< String > parsedLine;
- line.parse( parsedLine );
+ line.split( parsedLine );
if( parsedLine.getSize() != 3 )
{
std::cerr << "Wrong number of parameters in the matrix row at line:" << line << std::endl;
@@ -410,11 +406,7 @@ class MatrixReaderDeviceDependentCode< Devices::Cuda >
if( ! MatrixReader< HostMatrixType >::readMtxFileHostMatrix( file, hostMatrix, rowLengthsVector, verbose ) )
return false;
- typename Matrix::CompressedRowLengthsVector cudaCompressedRowLengthsVector;
- cudaCompressedRowLengthsVector.setLike( rowLengthsVector );
- cudaCompressedRowLengthsVector = rowLengthsVector;
- if( ! matrix.copyFrom( hostMatrix, cudaCompressedRowLengthsVector ) )
- return false;
+ matrix = hostMatrix;
return true;
}
};
diff --git a/src/TNL/Matrices/MatrixWriter_impl.h b/src/TNL/Matrices/MatrixWriter_impl.h
index 1db1a9df9e63435e9052aa65f4d35527038278bc..40368d0dd9fc157ff90ee0766add1cb64f2acca7 100644
--- a/src/TNL/Matrices/MatrixWriter_impl.h
+++ b/src/TNL/Matrices/MatrixWriter_impl.h
@@ -26,7 +26,7 @@ bool MatrixWriter< Matrix >::writeToGnuplot( std::ostream& str,
{
RealType elementValue = matrix.getElement( row, column );
if( elementValue != ( RealType ) 0.0 )
- str << column << " " << row << " " << elementValue << std::endl;
+ str << column << " " << row << " " << elementValue << "\n";
}
if( verbose )
std::cout << "Drawing the row " << row << " \r" << std::flush;
@@ -86,7 +86,7 @@ bool MatrixWriter< Matrix >::writeEpsBody( std::ostream& str,
{
str << ( column - lastColumn ) * elementSize
<< " " << -( row - lastRow ) * elementSize
- << " translate newpath 0 0 " << elementSize << " " << elementSize << " rectstroke" << std::endl;
+ << " translate newpath 0 0 " << elementSize << " " << elementSize << " rectstroke\n";
lastColumn = column;
lastRow = row;
}
diff --git a/src/TNL/Matrices/Matrix_impl.h b/src/TNL/Matrices/Matrix_impl.h
index fb881d16f2eb01845750026525425345bc4167b5..45593b1552dd501e113b1327e1c33e5bb73a3d16 100644
--- a/src/TNL/Matrices/Matrix_impl.h
+++ b/src/TNL/Matrices/Matrix_impl.h
@@ -28,14 +28,13 @@ Matrix< Real, Device, Index >::Matrix()
template< typename Real,
typename Device,
typename Index >
- bool Matrix< Real, Device, Index >::setDimensions( const IndexType rows,
- const IndexType columns )
+void Matrix< Real, Device, Index >::setDimensions( const IndexType rows,
+ const IndexType columns )
{
TNL_ASSERT( rows > 0 && columns > 0,
- std::cerr << " rows = " << rows << " columns = " << columns );
+ std::cerr << " rows = " << rows << " columns = " << columns );
this->rows = rows;
this->columns = columns;
- return true;
}
template< typename Real,
@@ -54,9 +53,9 @@ template< typename Real,
template< typename Real2,
typename Device2,
typename Index2 >
-bool Matrix< Real, Device, Index >::setLike( const Matrix< Real2, Device2, Index2 >& matrix )
+void Matrix< Real, Device, Index >::setLike( const Matrix< Real2, Device2, Index2 >& matrix )
{
- return setDimensions( matrix.getRows(), matrix.getColumns() );
+ setDimensions( matrix.getRows(), matrix.getColumns() );
}
template< typename Real,
@@ -86,63 +85,6 @@ void Matrix< Real, Device, Index >::reset()
this->columns = 0;
}
-template< typename Real,
- typename Device,
- typename Index >
- template< typename MatrixT >
-bool Matrix< Real, Device, Index >::copyFrom( const MatrixT& matrix,
- const CompressedRowLengthsVector& rowLengths )
-{
- /*tnlStaticTNL_ASSERT( DeviceType::DeviceType == Devices::HostDevice, );
- tnlStaticTNL_ASSERT( DeviceType::DeviceType == Matrix:DeviceType::DeviceType, );*/
-
- this->setLike( matrix );
- if( ! this->setCompressedRowLengths( rowLengths ) )
- return false;
- Containers::Vector< RealType, Devices::Host, IndexType > values;
- Containers::Vector< IndexType, Devices::Host, IndexType > columns;
- if( ! values.setSize( this->getColumns() ) ||
- ! columns.setSize( this->getColumns() ) )
- return false;
- for( IndexType row = 0; row < this->getRows(); row++ )
- {
- TNL_ASSERT( false, );
- // TODO: fix this
- //matrix.getRow( row, columns.getData(), values.getData() );
- this->setRow( row, columns.getData(), values.getData(), rowLengths.getElement( row ) );
- }
- return true;
-}
-
-template< typename Real,
- typename Device,
- typename Index >
-Matrix< Real, Device, Index >& Matrix< Real, Device, Index >::operator = ( const Matrix< RealType, DeviceType, IndexType >& m )
-{
- this->setLike( m );
-
- Containers::Vector< IndexType, DeviceType, IndexType > rowLengths;
- m.getCompressedRowLengths( rowLengths );
- this->setCompressedRowLengths( rowLengths );
-
- Containers::Vector< RealType, DeviceType, IndexType > rowValues;
- Containers::Vector< IndexType, DeviceType, IndexType > rowColumns;
- const IndexType maxRowLength = rowLengths.max();
- rowValues.setSize( maxRowLength );
- rowColumns.setSize( maxRowLength );
- for( IndexType row = 0; row < this->getRows(); row++ )
- {
- m.getRow( row,
- rowColumns.getData(),
- rowValues.getData() );
- this->setRow( row,
- rowColumns.getData(),
- rowValues.getData(),
- m.getRowLength( row ) );
- }
- return *this;
-}
-
template< typename Real,
typename Device,
typename Index >
@@ -173,19 +115,11 @@ template< typename Real,
typename Index >
bool Matrix< Real, Device, Index >::save( File& file ) const
{
-#ifdef HAVE_NOT_CXX11
- if( ! Object::save( file ) ||
- ! file.write< IndexType, Devices::Host, Index >( &this->rows, 1 ) ||
- ! file.write< IndexType, Devices::Host, Index >( &this->columns, 1 ) ||
- ! this->values.save( file ) )
- return false;
-#else
if( ! Object::save( file ) ||
! file.write( &this->rows ) ||
! file.write( &this->columns ) ||
! this->values.save( file ) )
return false;
-#endif
return true;
}
@@ -194,19 +128,11 @@ template< typename Real,
typename Index >
bool Matrix< Real, Device, Index >::load( File& file )
{
-#ifdef HAVE_NOT_CXX11
- if( ! Object::load( file ) ||
- ! file.read< IndexType, Devices::Host, Index >( &this->rows, 1 ) ||
- ! file.read< IndexType, Devices::Host, Index >( &this->columns, 1 ) ||
- ! this->values.load( file ) )
- return false;
-#else
if( ! Object::load( file ) ||
! file.read( &this->rows ) ||
! file.read( &this->columns ) ||
! this->values.load( file ) )
return false;
-#endif
return true;
}
@@ -257,12 +183,12 @@ void MatrixVectorProductCuda( const Matrix& matrix,
kernel_inVector,
kernel_outVector,
gridIdx );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
Devices::Cuda::freeFromDevice( kernel_this );
Devices::Cuda::freeFromDevice( kernel_inVector );
Devices::Cuda::freeFromDevice( kernel_outVector );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
diff --git a/src/TNL/Matrices/Multidiagonal.h b/src/TNL/Matrices/Multidiagonal.h
index 5acd602ac2a7b4c34b3a52d99a34f2297c71c527..28de74b8ac64b9fdf002aa71a4ca6a34a6ad4b26 100644
--- a/src/TNL/Matrices/Multidiagonal.h
+++ b/src/TNL/Matrices/Multidiagonal.h
@@ -23,8 +23,16 @@ class MultidiagonalDeviceDependentCode;
template< typename Real, typename Device = Devices::Host, typename Index = int >
class Multidiagonal : public Matrix< Real, Device, Index >
{
- public:
+private:
+ // convenient template alias for controlling the selection of copy-assignment operator
+ template< typename Device2 >
+ using Enabler = std::enable_if< ! std::is_same< Device2, Device >::value >;
+ // friend class will be needed for templated assignment operators
+ template< typename Real2, typename Device2, typename Index2 >
+ friend class Multidiagonal;
+
+public:
typedef Real RealType;
typedef Device DeviceType;
typedef Index IndexType;
@@ -42,22 +50,29 @@ class Multidiagonal : public Matrix< Real, Device, Index >
String getTypeVirtual() const;
- bool setDimensions( const IndexType rows,
+ static String getSerializationType();
+
+ virtual String getSerializationTypeVirtual() const;
+
+ void setDimensions( const IndexType rows,
const IndexType columns );
- bool setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
+ void setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
IndexType getRowLength( const IndexType row ) const;
+ __cuda_callable__
+ IndexType getRowLengthFast( const IndexType row ) const;
+
IndexType getMaxRowLength() const;
template< typename Vector >
- bool setDiagonals( const Vector& diagonals );
+ void setDiagonals( const Vector& diagonals );
const Containers::Vector< Index, Device, Index >& getDiagonals() const;
template< typename Real2, typename Device2, typename Index2 >
- bool setLike( const Multidiagonal< Real2, Device2, Index2 >& matrix );
+ void setLike( const Multidiagonal< Real2, Device2, Index2 >& matrix );
IndexType getNumberOfMatrixElements() const;
@@ -168,6 +183,14 @@ class Multidiagonal : public Matrix< Real, Device, Index >
Vector& x,
const RealType& omega = 1.0 ) const;
+ // copy assignment
+ Multidiagonal& operator=( const Multidiagonal& matrix );
+
+ // cross-device copy assignment
+ template< typename Real2, typename Device2, typename Index2,
+ typename = typename Enabler< Device2 >::type >
+ Multidiagonal& operator=( const Multidiagonal< Real2, Device2, Index2 >& matrix );
+
bool save( File& file ) const;
bool load( File& file );
@@ -178,7 +201,7 @@ class Multidiagonal : public Matrix< Real, Device, Index >
void print( std::ostream& str ) const;
- protected:
+protected:
bool getElementIndex( const IndexType row,
const IndexType column,
@@ -195,8 +218,6 @@ class Multidiagonal : public Matrix< Real, Device, Index >
typedef MultidiagonalDeviceDependentCode< DeviceType > DeviceDependentCode;
friend class MultidiagonalDeviceDependentCode< DeviceType >;
-
-
};
} // namespace Matrices
diff --git a/src/TNL/Matrices/MultidiagonalMatrixSetter_impl.h b/src/TNL/Matrices/MultidiagonalMatrixSetter_impl.h
index c126162caf8d128aa473cc1896ab75059a6d130d..26f6b2994f1f46440f170e42d32689e46807ab77 100644
--- a/src/TNL/Matrices/MultidiagonalMatrixSetter_impl.h
+++ b/src/TNL/Matrices/MultidiagonalMatrixSetter_impl.h
@@ -29,14 +29,14 @@ setupMatrix( const MeshType& mesh,
matrix.setDimensions( dofs, dofs );
CoordinatesType centerCell( stencilSize );
Containers::Vector< Index, Device, Index > diagonals;
- if( ! diagonals.setSize( 3 ) )
- return false;
+ diagonals.setSize( 3 );
Index centerCellIndex = mesh.getCellIndex( CoordinatesType( stencilSize ) );
diagonals.setElement( 0, mesh.getCellIndex( CoordinatesType( stencilSize - 1 ) ) - centerCellIndex );
diagonals.setElement( 1, 0 );
diagonals.setElement( 2, mesh.getCellIndex( CoordinatesType( stencilSize + 1 ) ) - centerCellIndex );
//cout << "Setting the multidiagonal matrix offsets to: " << diagonals << std::endl;
- return matrix.setDiagonals( diagonals );
+ matrix.setDiagonals( diagonals );
+ return true;
}
template< typename MeshReal,
@@ -55,8 +55,7 @@ setupMatrix( const MeshType& mesh,
matrix.setDimensions( dofs, dofs );
CoordinatesType centerCell( stencilSize );
Containers::Vector< Index, Device, Index > diagonals;
- if( ! diagonals.setSize( 5 ) )
- return false;
+ diagonals.setSize( 5 );
Index centerCellIndex = mesh.getCellIndex( CoordinatesType( stencilSize, stencilSize ) );
diagonals.setElement( 0, mesh.getCellIndex( CoordinatesType( stencilSize, stencilSize - 1 ) ) - centerCellIndex );
diagonals.setElement( 1, mesh.getCellIndex( CoordinatesType( stencilSize - 1, stencilSize ) ) - centerCellIndex );
@@ -64,7 +63,8 @@ setupMatrix( const MeshType& mesh,
diagonals.setElement( 3, mesh.getCellIndex( CoordinatesType( stencilSize + 1, stencilSize ) ) - centerCellIndex );
diagonals.setElement( 4, mesh.getCellIndex( CoordinatesType( stencilSize, stencilSize + 1 ) ) - centerCellIndex );
//cout << "Setting the multidiagonal matrix offsets to: " << diagonals << std::endl;
- return matrix.setDiagonals( diagonals );
+ matrix.setDiagonals( diagonals );
+ return true;
}
template< typename MeshReal,
@@ -83,8 +83,7 @@ setupMatrix( const MeshType& mesh,
matrix.setDimensions( dofs, dofs );
CoordinatesType centerCell( stencilSize );
Containers::Vector< Index, Device, Index > diagonals;
- if( ! diagonals.setSize( 7 ) )
- return false;
+ diagonals.setSize( 7 );
Index centerCellIndex = mesh.getCellIndex( CoordinatesType( stencilSize, stencilSize, stencilSize ) );
diagonals.setElement( 0, mesh.getCellIndex( CoordinatesType( stencilSize, stencilSize, stencilSize - 1 ) ) - centerCellIndex );
diagonals.setElement( 1, mesh.getCellIndex( CoordinatesType( stencilSize, stencilSize - 1, stencilSize ) ) - centerCellIndex );
@@ -94,7 +93,8 @@ setupMatrix( const MeshType& mesh,
diagonals.setElement( 5, mesh.getCellIndex( CoordinatesType( stencilSize, stencilSize + 1, stencilSize ) ) - centerCellIndex );
diagonals.setElement( 6, mesh.getCellIndex( CoordinatesType( stencilSize, stencilSize, stencilSize + 1 ) ) - centerCellIndex );
//cout << "Setting the multidiagonal matrix offsets to: " << diagonals << std::endl;
- return matrix.setDiagonals( diagonals );
+ matrix.setDiagonals( diagonals );
+ return true;
}
} // namespace Matrices
diff --git a/src/TNL/Matrices/Multidiagonal_impl.h b/src/TNL/Matrices/Multidiagonal_impl.h
index 9545a2070c66b7c27a7cf2cb2e7a2e716bc74d64..47d827b93df0904b568c234cc8bfbea602479e46 100644
--- a/src/TNL/Matrices/Multidiagonal_impl.h
+++ b/src/TNL/Matrices/Multidiagonal_impl.h
@@ -50,32 +50,44 @@ String Multidiagonal< Real, Device, Index >::getTypeVirtual() const
template< typename Real,
typename Device,
typename Index >
-bool Multidiagonal< Real, Device, Index >::setDimensions( const IndexType rows,
- const IndexType columns )
+String Multidiagonal< Real, Device, Index >::getSerializationType()
+{
+ return getType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+String Multidiagonal< Real, Device, Index >::getSerializationTypeVirtual() const
+{
+ return this->getSerializationType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+void Multidiagonal< Real, Device, Index >::setDimensions( const IndexType rows,
+ const IndexType columns )
{
TNL_ASSERT( rows > 0 && columns > 0,
std::cerr << "rows = " << rows
<< " columns = " << columns << std::endl );
- if( ! Matrix< Real, Device, Index >::setDimensions( rows, columns ) )
- return false;
+ Matrix< Real, Device, Index >::setDimensions( rows, columns );
if( this->diagonalsShift.getSize() != 0 )
{
- if( ! this->values.setSize( min( this->rows, this->columns ) * this->diagonalsShift.getSize() ) )
- return false;
+ this->values.setSize( min( this->rows, this->columns ) * this->diagonalsShift.getSize() );
this->values.setValue( 0.0 );
}
- return true;
}
template< typename Real,
typename Device,
typename Index >
-bool Multidiagonal< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
+void Multidiagonal< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
{
/****
* TODO: implement some check here similar to the one in the tridiagonal matrix
*/
- return true;
}
template< typename Real,
@@ -93,6 +105,22 @@ Index Multidiagonal< Real, Device, Index >::getRowLength( const IndexType row )
return rowLength;
}
+template< typename Real,
+ typename Device,
+ typename Index >
+__cuda_callable__
+Index Multidiagonal< Real, Device, Index >::getRowLengthFast( const IndexType row ) const
+{
+ IndexType rowLength( 0 );
+ for( IndexType i = 0; i < diagonalsShift.getSize(); i++ )
+ {
+ const IndexType column = row + diagonalsShift[ i ];
+ if( column >= 0 && column < this->getColumns() )
+ rowLength++;
+ }
+ return rowLength;
+}
+
template< typename Real,
typename Device,
typename Index >
@@ -107,7 +135,7 @@ template< typename Real,
typename Device,
typename Index >
template< typename Vector >
-bool Multidiagonal< Real, Device, Index > :: setDiagonals( const Vector& diagonals )
+void Multidiagonal< Real, Device, Index > :: setDiagonals( const Vector& diagonals )
{
TNL_ASSERT( diagonals.getSize() > 0,
std::cerr << "New number of diagonals = " << diagonals.getSize() << std::endl );
@@ -115,11 +143,9 @@ bool Multidiagonal< Real, Device, Index > :: setDiagonals( const Vector& diagon
this->diagonalsShift = diagonals;
if( this->rows != 0 && this->columns != 0 )
{
- if( ! this->values.setSize( min( this->rows, this->columns ) * this->diagonalsShift.getSize() ) )
- return false;
+ this->values.setSize( min( this->rows, this->columns ) * this->diagonalsShift.getSize() );
this->values.setValue( 0.0 );
}
- return true;
}
template< typename Real,
@@ -136,13 +162,10 @@ template< typename Real,
template< typename Real2,
typename Device2,
typename Index2 >
-bool Multidiagonal< Real, Device, Index > :: setLike( const Multidiagonal< Real2, Device2, Index2 >& matrix )
+void Multidiagonal< Real, Device, Index > :: setLike( const Multidiagonal< Real2, Device2, Index2 >& matrix )
{
- if( ! this->setDimensions( matrix.getRows(), matrix.getColumns() ) )
- return false;
- if( ! setDiagonals( matrix.getDiagonals() ) )
- return false;
- return true;
+ this->setDimensions( matrix.getRows(), matrix.getColumns() );
+ setDiagonals( matrix.getDiagonals() );
}
template< typename Real,
@@ -591,6 +614,39 @@ bool Multidiagonal< Real, Device, Index > :: performSORIteration( const Vector&
}
+// copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+Multidiagonal< Real, Device, Index >&
+Multidiagonal< Real, Device, Index >::operator=( const Multidiagonal& matrix )
+{
+ this->setLike( matrix );
+ this->values = matrix.values;
+ this->diagonalsShift = matrix.diagonalsShift;
+ return *this;
+}
+
+// cross-device copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+ template< typename Real2, typename Device2, typename Index2, typename >
+Multidiagonal< Real, Device, Index >&
+Multidiagonal< Real, Device, Index >::operator=( const Multidiagonal< Real2, Device2, Index2 >& matrix )
+{
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::Cuda >::value,
+ "unknown device" );
+ static_assert( std::is_same< Device2, Devices::Host >::value || std::is_same< Device2, Devices::Cuda >::value,
+ "unknown device" );
+
+ this->setLike( matrix );
+
+ std::cerr << "Cross-device assignment for the Multidiagonal format is not implemented yet." << std::endl;
+ throw 1;
+}
+
+
template< typename Real,
typename Device,
typename Index >
diff --git a/src/TNL/Matrices/SlicedEllpack.h b/src/TNL/Matrices/SlicedEllpack.h
index ebae0c936e4a8b95beecc503e771ce1ed5d0b666..0557d26ebbd94ee57ab2bcd51bb029f48cbd30c7 100644
--- a/src/TNL/Matrices/SlicedEllpack.h
+++ b/src/TNL/Matrices/SlicedEllpack.h
@@ -51,17 +51,25 @@ template< typename Real,
int SliceSize >
class SlicedEllpack : public Sparse< Real, Device, Index >
{
- public:
+private:
+ // convenient template alias for controlling the selection of copy-assignment operator
+ template< typename Device2 >
+ using Enabler = std::enable_if< ! std::is_same< Device2, Device >::value >;
+ // friend class will be needed for templated assignment operators
+ template< typename Real2, typename Device2, typename Index2, int SliceSize2 >
+ friend class SlicedEllpack;
+
+public:
typedef Real RealType;
typedef Device DeviceType;
typedef Index IndexType;
typedef typename Sparse< RealType, DeviceType, IndexType >::CompressedRowLengthsVector CompressedRowLengthsVector;
typedef typename Sparse< RealType, DeviceType, IndexType >::ValuesVector ValuesVector;
typedef typename Sparse< RealType, DeviceType, IndexType >::ColumnIndexesVector ColumnIndexesVector;
- typedef SlicedEllpack< Real, Device, Index > ThisType;
- typedef SlicedEllpack< Real, Devices::Host, Index > HostType;
- typedef SlicedEllpack< Real, Devices::Cuda, Index > CudaType;
+ typedef SlicedEllpack< Real, Device, Index, SliceSize > ThisType;
+ typedef SlicedEllpack< Real, Devices::Host, Index, SliceSize > HostType;
+ typedef SlicedEllpack< Real, Devices::Cuda, Index, SliceSize > CudaType;
typedef Sparse< Real, Device, Index > BaseType;
typedef typename BaseType::MatrixRow MatrixRow;
typedef SparseRow< const RealType, const IndexType > ConstMatrixRow;
@@ -73,15 +81,22 @@ class SlicedEllpack : public Sparse< Real, Device, Index >
String getTypeVirtual() const;
- bool setDimensions( const IndexType rows,
+ static String getSerializationType();
+
+ virtual String getSerializationTypeVirtual() const;
+
+ void setDimensions( const IndexType rows,
const IndexType columns );
- bool setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
+ void setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
IndexType getRowLength( const IndexType row ) const;
+ __cuda_callable__
+ IndexType getRowLengthFast( const IndexType row ) const;
+
template< typename Real2, typename Device2, typename Index2 >
- bool setLike( const SlicedEllpack< Real2, Device2, Index2, SliceSize >& matrix );
+ void setLike( const SlicedEllpack< Real2, Device2, Index2, SliceSize >& matrix );
void reset();
@@ -178,6 +193,14 @@ class SlicedEllpack : public Sparse< Real, Device, Index >
Vector& x,
const RealType& omega = 1.0 ) const;
+ // copy assignment
+ SlicedEllpack& operator=( const SlicedEllpack& matrix );
+
+ // cross-device copy assignment
+ template< typename Real2, typename Device2, typename Index2,
+ typename = typename Enabler< Device2 >::type >
+ SlicedEllpack& operator=( const SlicedEllpack< Real2, Device2, Index2, SliceSize >& matrix );
+
bool save( File& file ) const;
bool load( File& file );
@@ -188,7 +211,7 @@ class SlicedEllpack : public Sparse< Real, Device, Index >
void print( std::ostream& str ) const;
- protected:
+protected:
Containers::Vector< Index, Device, Index > slicePointers, sliceCompressedRowLengths;
@@ -201,12 +224,10 @@ class SlicedEllpack : public Sparse< Real, Device, Index >
*/
// TODO: The friend declaration above does not work because of __global__ storage specifier. Therefore we declare the following method as public. Fix this, when possible.
- public:
+public:
__device__ void computeMaximalRowLengthInSlicesCuda( const CompressedRowLengthsVector& rowLengths,
const IndexType sliceIdx );
-
#endif
-
};
} // namespace Matrices
diff --git a/src/TNL/Matrices/SlicedEllpack_impl.h b/src/TNL/Matrices/SlicedEllpack_impl.h
index 097f1b6772ceb4035ceba25732872de122377051..a08aaf19bc479ce4f9fa7c4b99775af0479c78b0 100644
--- a/src/TNL/Matrices/SlicedEllpack_impl.h
+++ b/src/TNL/Matrices/SlicedEllpack_impl.h
@@ -51,34 +51,53 @@ template< typename Real,
typename Device,
typename Index,
int SliceSize >
-bool SlicedEllpack< Real, Device, Index, SliceSize >::setDimensions( const IndexType rows,
- const IndexType columns )
+String SlicedEllpack< Real, Device, Index, SliceSize >::getSerializationType()
+{
+ return getType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index,
+ int SliceSize >
+String SlicedEllpack< Real, Device, Index, SliceSize >::getSerializationTypeVirtual() const
+{
+ return this->getSerializationType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index,
+ int SliceSize >
+void SlicedEllpack< Real, Device, Index, SliceSize >::setDimensions( const IndexType rows,
+ const IndexType columns )
{
TNL_ASSERT( rows > 0 && columns > 0,
std::cerr << "rows = " << rows
<< " columns = " << columns << std::endl );
- return Sparse< Real, Device, Index >::setDimensions( rows, columns );
+ Sparse< Real, Device, Index >::setDimensions( rows, columns );
}
template< typename Real,
typename Device,
typename Index,
int SliceSize >
-bool SlicedEllpack< Real, Device, Index, SliceSize >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
+void SlicedEllpack< Real, Device, Index, SliceSize >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
{
- TNL_ASSERT( this->getRows() > 0, );
- TNL_ASSERT( this->getColumns() > 0, );
+ TNL_ASSERT_GT( this->getRows(), 0, "cannot set row lengths of an empty matrix" );
+ TNL_ASSERT_GT( this->getColumns(), 0, "cannot set row lengths of an empty matrix" );
+ TNL_ASSERT_EQ( this->getRows(), rowLengths.getSize(), "wrong size of the rowLengths vector" );
+
const IndexType slices = roundUpDivision( this->rows, SliceSize );
- if( ! this->sliceCompressedRowLengths.setSize( slices ) ||
- ! this->slicePointers.setSize( slices + 1 ) )
- return false;
+ this->sliceCompressedRowLengths.setSize( slices );
+ this->slicePointers.setSize( slices + 1 );
DeviceDependentCode::computeMaximalRowLengthInSlices( *this, rowLengths );
this->maxRowLength = rowLengths.max();
this->slicePointers.computeExclusivePrefixSum();
- return this->allocateMatrixElements( this->slicePointers.getElement( slices ) );
+ this->allocateMatrixElements( this->slicePointers.getElement( slices ) );
}
template< typename Real,
@@ -91,6 +110,17 @@ Index SlicedEllpack< Real, Device, Index, SliceSize >::getRowLength( const Index
return this->sliceCompressedRowLengths.getElement( slice );
}
+template< typename Real,
+ typename Device,
+ typename Index,
+ int SliceSize >
+__cuda_callable__
+Index SlicedEllpack< Real, Device, Index, SliceSize >::getRowLengthFast( const IndexType row ) const
+{
+ const IndexType slice = row / SliceSize;
+ return this->sliceCompressedRowLengths[ slice ];
+}
+
template< typename Real,
typename Device,
typename Index,
@@ -98,13 +128,11 @@ template< typename Real,
template< typename Real2,
typename Device2,
typename Index2 >
-bool SlicedEllpack< Real, Device, Index, SliceSize >::setLike( const SlicedEllpack< Real2, Device2, Index2, SliceSize >& matrix )
+void SlicedEllpack< Real, Device, Index, SliceSize >::setLike( const SlicedEllpack< Real2, Device2, Index2, SliceSize >& matrix )
{
- if( !Sparse< Real, Device, Index >::setLike( matrix ) ||
- ! this->slicePointers.setLike( matrix.slicePointers ) ||
- ! this->sliceCompressedRowLengths.setLike( matrix.sliceCompressedRowLengths ) )
- return false;
- return true;
+ Sparse< Real, Device, Index >::setLike( matrix );
+ this->slicePointers.setLike( matrix.slicePointers );
+ this->sliceCompressedRowLengths.setLike( matrix.sliceCompressedRowLengths );
}
template< typename Real,
@@ -573,6 +601,95 @@ bool SlicedEllpack< Real, Device, Index, SliceSize >::performSORIteration( const
}
+// copy assignment
+template< typename Real,
+ typename Device,
+ typename Index,
+ int SliceSize >
+SlicedEllpack< Real, Device, Index, SliceSize >&
+SlicedEllpack< Real, Device, Index, SliceSize >::operator=( const SlicedEllpack& matrix )
+{
+ this->setLike( matrix );
+ this->values = matrix.values;
+ this->columnIndexes = matrix.columnIndexes;
+ this->slicePointers = matrix.slicePointers;
+ this->sliceCompressedRowLengths = matrix.sliceCompressedRowLengths;
+ return *this;
+}
+
+// cross-device copy assignment
+template< typename Real,
+ typename Device,
+ typename Index,
+ int SliceSize >
+ template< typename Real2, typename Device2, typename Index2, typename >
+SlicedEllpack< Real, Device, Index, SliceSize >&
+SlicedEllpack< Real, Device, Index, SliceSize >::operator=( const SlicedEllpack< Real2, Device2, Index2, SliceSize >& matrix )
+{
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::Cuda >::value || std::is_same< Device, Devices::MIC >::value,
+ "unknown device" );
+ static_assert( std::is_same< Device2, Devices::Host >::value || std::is_same< Device2, Devices::Cuda >::value || std::is_same< Device2, Devices::MIC >::value,
+ "unknown device" );
+
+ this->setLike( matrix );
+ this->slicePointers = matrix.slicePointers;
+ this->sliceCompressedRowLengths = matrix.sliceCompressedRowLengths;
+
+ // host -> cuda
+ if( std::is_same< Device, Devices::Cuda >::value ) {
+ typename ValuesVector::HostType tmpValues;
+ typename ColumnIndexesVector::HostType tmpColumnIndexes;
+ tmpValues.setLike( matrix.values );
+ tmpColumnIndexes.setLike( matrix.columnIndexes );
+
+#ifdef HAVE_OPENMP
+#pragma omp parallel for if( Devices::Host::isOMPEnabled() )
+#endif
+ for( Index sliceIdx = 0; sliceIdx < matrix.sliceCompressedRowLengths.getSize(); sliceIdx++ ) {
+ const Index rowLength = matrix.sliceCompressedRowLengths[ sliceIdx ];
+ const Index offset = matrix.slicePointers[ sliceIdx ];
+ for( Index j = 0; j < rowLength; j++ )
+ for( Index i = 0; i < SliceSize; i++ ) {
+ tmpValues[ offset + j * SliceSize + i ] = matrix.values[ offset + i * rowLength + j ];
+ tmpColumnIndexes[ offset + j * SliceSize + i ] = matrix.columnIndexes[ offset + i * rowLength + j ];
+ }
+ }
+
+ this->values = tmpValues;
+ this->columnIndexes = tmpColumnIndexes;
+ }
+
+ // cuda -> host
+ if( std::is_same< Device, Devices::Host >::value ) {
+ ValuesVector tmpValues;
+ ColumnIndexesVector tmpColumnIndexes;
+ tmpValues.setLike( matrix.values );
+ tmpColumnIndexes.setLike( matrix.columnIndexes );
+ tmpValues = matrix.values;
+ tmpColumnIndexes = matrix.columnIndexes;
+
+#ifdef HAVE_OPENMP
+#pragma omp parallel for if( Devices::Host::isOMPEnabled() )
+#endif
+ for( Index sliceIdx = 0; sliceIdx < sliceCompressedRowLengths.getSize(); sliceIdx++ ) {
+ const Index rowLength = sliceCompressedRowLengths[ sliceIdx ];
+ const Index offset = slicePointers[ sliceIdx ];
+ for( Index i = 0; i < SliceSize; i++ )
+ for( Index j = 0; j < rowLength; j++ ) {
+ this->values[ offset + i * rowLength + j ] = tmpValues[ offset + j * SliceSize + i ];
+ this->columnIndexes[ offset + i * rowLength + j ] = tmpColumnIndexes[ offset + j * SliceSize + i ];
+ }
+ }
+ }
+
+ if( std::is_same< Device, Devices::MIC >::value ) {
+ throw std::runtime_error("Not Implemented yet for MIC");
+ }
+
+ return *this;
+}
+
+
template< typename Real,
typename Device,
typename Index,
@@ -623,23 +740,30 @@ template< typename Real,
int SliceSize >
void SlicedEllpack< Real, Device, Index, SliceSize >::print( std::ostream& str ) const
{
- for( IndexType row = 0; row < this->getRows(); row++ )
- {
- str <<"Row: " << row << " -> ";
- const IndexType sliceIdx = row / SliceSize;
- const IndexType rowLength = this->sliceCompressedRowLengths.getElement( sliceIdx );
- IndexType elementPtr = this->slicePointers.getElement( sliceIdx ) +
- rowLength * ( row - sliceIdx * SliceSize );
- const IndexType rowEnd( elementPtr + rowLength );
- while( elementPtr < rowEnd &&
- this->columnIndexes.getElement( elementPtr ) < this->columns &&
- this->columnIndexes.getElement( elementPtr ) != this->getPaddingIndex() )
+ if( std::is_same< Device, Devices::Host >::value ) {
+ for( IndexType row = 0; row < this->getRows(); row++ )
{
- const Index column = this->columnIndexes.getElement( elementPtr );
- str << " Col:" << column << "->" << this->values.getElement( elementPtr ) << "\t";
- elementPtr++;
+ str <<"Row: " << row << " -> ";
+ const IndexType sliceIdx = row / SliceSize;
+ const IndexType rowLength = this->sliceCompressedRowLengths.getElement( sliceIdx );
+ IndexType elementPtr = this->slicePointers.getElement( sliceIdx ) +
+ rowLength * ( row - sliceIdx * SliceSize );
+ const IndexType rowEnd( elementPtr + rowLength );
+ while( elementPtr < rowEnd &&
+ this->columnIndexes.getElement( elementPtr ) < this->columns &&
+ this->columnIndexes.getElement( elementPtr ) != this->getPaddingIndex() )
+ {
+ const Index column = this->columnIndexes.getElement( elementPtr );
+ str << " Col:" << column << "->" << this->values.getElement( elementPtr ) << "\t";
+ elementPtr++;
+ }
+ str << std::endl;
}
- str << std::endl;
+ }
+ else {
+ HostType hostMatrix;
+ hostMatrix = *this;
+ hostMatrix.print( str );
}
}
@@ -882,7 +1006,7 @@ class SlicedEllpackDeviceDependentCode< Devices::Cuda >
}
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_rowLengths );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
return true;
}
@@ -923,17 +1047,71 @@ class SlicedEllpackDeviceDependentCode< Devices::Cuda >
inVector.getData(),
outVector.getData(),
gridIdx );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
//Devices::Cuda::freeFromDevice( kernel_this );
//Devices::Cuda::freeFromDevice( kernel_inVector );
//Devices::Cuda::freeFromDevice( kernel_outVector );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
cudaThreadSynchronize();
#endif
}
};
+template<>
+class SlicedEllpackDeviceDependentCode< Devices::MIC >
+{
+ public:
+
+ typedef Devices::MIC Device;
+
+ template< typename Real,
+ typename Index,
+ int SliceSize >
+ static void initRowTraverse( const SlicedEllpack< Real, Device, Index, SliceSize >& matrix,
+ const Index row,
+ Index& rowBegin,
+ Index& rowEnd,
+ Index& step )
+ {
+ throw std::runtime_error("Not Implemented yet SlicedEllpackDeviceDependentCode< Devices::MIC >::initRowTraverse");
+ }
+
+ template< typename Real,
+ typename Index,
+ int SliceSize >
+ __cuda_callable__
+ static void initRowTraverseFast( const SlicedEllpack< Real, Device, Index, SliceSize >& matrix,
+ const Index row,
+ Index& rowBegin,
+ Index& rowEnd,
+ Index& step )
+ {
+ throw std::runtime_error("Not Implemented yet SlicedEllpackDeviceDependentCode< Devices::MIC >::initRowTraverseFast");
+ }
+
+ template< typename Real,
+ typename Index,
+ int SliceSize >
+ static bool computeMaximalRowLengthInSlices( SlicedEllpack< Real, Device, Index, SliceSize >& matrix,
+ const typename SlicedEllpack< Real, Device, Index >::CompressedRowLengthsVector& rowLengths )
+ {
+ throw std::runtime_error("Not Implemented yet SlicedEllpackDeviceDependentCode< Devices::MIC >::computeMaximalRowLengthInSlices");
+ }
+
+ template< typename Real,
+ typename Index,
+ typename InVector,
+ typename OutVector,
+ int SliceSize >
+ static void vectorProduct( const SlicedEllpack< Real, Device, Index, SliceSize >& matrix,
+ const InVector& inVector,
+ OutVector& outVector )
+ {
+ throw std::runtime_error("Not Implemented yet SlicedEllpackDeviceDependentCode< Devices::MIC >::vectorProduct");
+ }
+};
+
} // namespace Matrices
} // namespace TNL
diff --git a/src/TNL/Matrices/Sparse.h b/src/TNL/Matrices/Sparse.h
index c1f6e168ed8dcf7d3be44cea43bbdc3dc4ff707f..2a694826b94e9d757079f72942f3f810ce136885 100644
--- a/src/TNL/Matrices/Sparse.h
+++ b/src/TNL/Matrices/Sparse.h
@@ -34,10 +34,10 @@ class Sparse : public Matrix< Real, Device, Index >
Sparse();
- virtual bool setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths ) = 0;
+ virtual void setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths ) = 0;
template< typename Real2, typename Device2, typename Index2 >
- bool setLike( const Sparse< Real2, Device2, Index2 >& matrix );
+ void setLike( const Sparse< Real2, Device2, Index2 >& matrix );
IndexType getNumberOfMatrixElements() const;
@@ -58,13 +58,20 @@ class Sparse : public Matrix< Real, Device, Index >
protected:
- bool allocateMatrixElements( const IndexType& numberOfMatrixElements );
+ void allocateMatrixElements( const IndexType& numberOfMatrixElements );
Containers::Vector< Index, Device, Index > columnIndexes;
Index maxRowLength;
};
+
+// This cannot be a method of the Sparse class, because the implementation uses
+// methods (marked with __cuda_callable__) which are defined only on the
+// subclasses, but are not virtual methods of Sparse.
+template< typename Matrix1, typename Matrix2 >
+void copySparseMatrix( Matrix1& A, const Matrix2& B );
+
} // namespace Matrices
} // namespace TNL
diff --git a/src/TNL/Matrices/Sparse_impl.h b/src/TNL/Matrices/Sparse_impl.h
index 4f415976a7306b512162272df8142c444b92e5e0..61b3b21d68e4488493431faf890afd5c2537a1fe 100644
--- a/src/TNL/Matrices/Sparse_impl.h
+++ b/src/TNL/Matrices/Sparse_impl.h
@@ -10,8 +10,11 @@
#pragma once
+#include "Sparse.h"
+#include <TNL/DevicePointer.h>
+
namespace TNL {
-namespace Matrices {
+namespace Matrices {
template< typename Real,
typename Device,
@@ -27,12 +30,10 @@ template< typename Real,
template< typename Real2,
typename Device2,
typename Index2 >
-bool Sparse< Real, Device, Index >::setLike( const Sparse< Real2, Device2, Index2 >& matrix )
+void Sparse< Real, Device, Index >::setLike( const Sparse< Real2, Device2, Index2 >& matrix )
{
- if( ! Matrix< Real, Device, Index >::setLike( matrix ) ||
- ! this->allocateMatrixElements( matrix.getNumberOfMatrixElements() ) )
- return false;
- return true;
+ Matrix< Real, Device, Index >::setLike( matrix );
+ this->allocateMatrixElements( matrix.getNumberOfMatrixElements() );
}
template< typename Real,
@@ -112,18 +113,16 @@ bool Sparse< Real, Device, Index >::load( File& file )
template< typename Real,
typename Device,
typename Index >
-bool Sparse< Real, Device, Index >::allocateMatrixElements( const IndexType& numberOfMatrixElements )
+void Sparse< Real, Device, Index >::allocateMatrixElements( const IndexType& numberOfMatrixElements )
{
- if( ! this->values.setSize( numberOfMatrixElements ) ||
- ! this->columnIndexes.setSize( numberOfMatrixElements ) )
- return false;
+ this->values.setSize( numberOfMatrixElements );
+ this->columnIndexes.setSize( numberOfMatrixElements );
/****
* Setting a column index to this->columns means that the
* index is undefined.
*/
this->columnIndexes.setValue( this->columns );
- return true;
}
template< typename Real,
@@ -131,6 +130,147 @@ template< typename Real,
typename Index >
void Sparse< Real, Device, Index >::printStructure( std::ostream& str ) const
{
+ TNL_ASSERT_TRUE( false, "Not implemented yet." );
+}
+
+
+#ifdef HAVE_CUDA
+template< typename Vector, typename Matrix >
+__global__ void
+SparseMatrixSetRowLengthsVectorKernel( Vector* rowLengths,
+ const Matrix* matrix,
+ typename Matrix::IndexType rows,
+ typename Matrix::IndexType cols )
+{
+ using IndexType = typename Matrix::IndexType;
+
+ IndexType rowIdx = blockIdx.x * blockDim.x + threadIdx.x;
+ const IndexType gridSize = blockDim.x * gridDim.x;
+
+ while( rowIdx < rows ) {
+ const auto max_length = matrix->getRowLengthFast( rowIdx );
+ const auto row = matrix->getRow( rowIdx );
+ IndexType length = 0;
+ for( IndexType c_j = 0; c_j < max_length; c_j++ )
+ if( row.getElementColumn( c_j ) < cols )
+ length++;
+ else
+ break;
+ rowLengths[ rowIdx ] = length;
+ rowIdx += gridSize;
+ }
+}
+
+template< typename Matrix1, typename Matrix2 >
+__global__ void
+SparseMatrixCopyKernel( Matrix1* A,
+ const Matrix2* B,
+ const typename Matrix2::IndexType* rowLengths,
+ typename Matrix2::IndexType rows )
+{
+ using IndexType = typename Matrix2::IndexType;
+
+ IndexType rowIdx = blockIdx.x * blockDim.x + threadIdx.x;
+ const IndexType gridSize = blockDim.x * gridDim.x;
+
+ while( rowIdx < rows ) {
+ const auto length = rowLengths[ rowIdx ];
+ const auto rowB = B->getRow( rowIdx );
+ auto rowA = A->getRow( rowIdx );
+ for( IndexType c = 0; c < length; c++ )
+ rowA.setElement( c, rowB.getElementColumn( c ), rowB.getElementValue( c ) );
+ rowIdx += gridSize;
+ }
+}
+#endif
+
+template< typename Matrix1, typename Matrix2 >
+void
+copySparseMatrix( Matrix1& A, const Matrix2& B )
+{
+ static_assert( std::is_same< typename Matrix1::RealType, typename Matrix2::RealType >::value,
+ "The matrices must have the same RealType." );
+ static_assert( std::is_same< typename Matrix1::DeviceType, typename Matrix2::DeviceType >::value,
+ "The matrices must be allocated on the same device." );
+ static_assert( std::is_same< typename Matrix1::IndexType, typename Matrix2::IndexType >::value,
+ "The matrices must have the same IndexType." );
+
+ using RealType = typename Matrix1::RealType;
+ using DeviceType = typename Matrix1::DeviceType;
+ using IndexType = typename Matrix1::IndexType;
+
+ const IndexType rows = B.getRows();
+ const IndexType cols = B.getColumns();
+
+ A.setDimensions( rows, cols );
+
+ if( std::is_same< DeviceType, Devices::Host >::value ) {
+ // set row lengths
+ typename Matrix1::CompressedRowLengthsVector rowLengths;
+ rowLengths.setSize( rows );
+#ifdef HAVE_OPENMP
+#pragma omp parallel for if( Devices::Host::isOMPEnabled() )
+#endif
+ for( IndexType i = 0; i < rows; i++ ) {
+ const auto max_length = B.getRowLength( i );
+ const auto row = B.getRow( i );
+ IndexType length = 0;
+ for( IndexType c_j = 0; c_j < max_length; c_j++ )
+ if( row.getElementColumn( c_j ) < cols )
+ length++;
+ else
+ break;
+ rowLengths[ i ] = length;
+ }
+ A.setCompressedRowLengths( rowLengths );
+
+#ifdef HAVE_OPENMP
+#pragma omp parallel for if( Devices::Host::isOMPEnabled() )
+#endif
+ for( IndexType i = 0; i < rows; i++ ) {
+ const auto length = rowLengths[ i ];
+ const auto rowB = B.getRow( i );
+ auto rowA = A.getRow( i );
+ for( IndexType c = 0; c < length; c++ )
+ rowA.setElement( c, rowB.getElementColumn( c ), rowB.getElementValue( c ) );
+ }
+ }
+
+ if( std::is_same< DeviceType, Devices::Cuda >::value ) {
+#ifdef HAVE_CUDA
+ dim3 blockSize( 256 );
+ dim3 gridSize;
+ const IndexType desGridSize = 32 * Devices::CudaDeviceInfo::getCudaMultiprocessors( Devices::CudaDeviceInfo::getActiveDevice() );
+ gridSize.x = min( desGridSize, Devices::Cuda::getNumberOfBlocks( rows, blockSize.x ) );
+
+ typename Matrix1::CompressedRowLengthsVector rowLengths;
+ rowLengths.setSize( rows );
+
+ DevicePointer< Matrix1 > Apointer( A );
+ const DevicePointer< const Matrix2 > Bpointer( B );
+
+ // set row lengths
+ Devices::Cuda::synchronizeDevice();
+ SparseMatrixSetRowLengthsVectorKernel<<< gridSize, blockSize >>>(
+ rowLengths.getData(),
+ &Bpointer.template getData< TNL::Devices::Cuda >(),
+ rows,
+ cols );
+ TNL_CHECK_CUDA_DEVICE;
+ Apointer->setCompressedRowLengths( rowLengths );
+
+ // copy rows
+ Devices::Cuda::synchronizeDevice();
+ SparseMatrixCopyKernel<<< gridSize, blockSize >>>(
+ &Apointer.template modifyData< TNL::Devices::Cuda >(),
+ &Bpointer.template getData< TNL::Devices::Cuda >(),
+ rowLengths.getData(),
+ rows );
+ TNL_CHECK_CUDA_DEVICE;
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
+ }
}
} // namespace Matrices
diff --git a/src/TNL/Matrices/Tridiagonal.h b/src/TNL/Matrices/Tridiagonal.h
index 88c9cc9164d8bff79a1cef9b304b368e6ac3c606..0769ca83f2f0355ee3ad4c726c601bd16bae8c3f 100644
--- a/src/TNL/Matrices/Tridiagonal.h
+++ b/src/TNL/Matrices/Tridiagonal.h
@@ -25,8 +25,16 @@ template< typename Real = double,
typename Index = int >
class Tridiagonal : public Matrix< Real, Device, Index >
{
- public:
+private:
+ // convenient template alias for controlling the selection of copy-assignment operator
+ template< typename Device2 >
+ using Enabler = std::enable_if< ! std::is_same< Device2, Device >::value >;
+ // friend class will be needed for templated assignment operators
+ template< typename Real2, typename Device2, typename Index2 >
+ friend class Tridiagonal;
+
+public:
typedef Real RealType;
typedef Device DeviceType;
typedef Index IndexType;
@@ -43,17 +51,24 @@ class Tridiagonal : public Matrix< Real, Device, Index >
String getTypeVirtual() const;
- bool setDimensions( const IndexType rows,
+ static String getSerializationType();
+
+ virtual String getSerializationTypeVirtual() const;
+
+ void setDimensions( const IndexType rows,
const IndexType columns );
- bool setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
+ void setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths );
IndexType getRowLength( const IndexType row ) const;
+ __cuda_callable__
+ IndexType getRowLengthFast( const IndexType row ) const;
+
IndexType getMaxRowLength() const;
template< typename Real2, typename Device2, typename Index2 >
- bool setLike( const Tridiagonal< Real2, Device2, Index2 >& m );
+ void setLike( const Tridiagonal< Real2, Device2, Index2 >& m );
IndexType getNumberOfMatrixElements() const;
@@ -148,15 +163,9 @@ class Tridiagonal : public Matrix< Real, Device, Index >
const RealType& matrixMultiplicator = 1.0,
const RealType& thisMatrixMultiplicator = 1.0 );
-#ifdef HAVE_NOT_CXX11
template< typename Real2, typename Index2 >
void getTransposition( const Tridiagonal< Real2, Device, Index2 >& matrix,
const RealType& matrixMultiplicator = 1.0 );
-#else
- template< typename Real2, typename Index2 >
- void getTransposition( const Tridiagonal< Real2, Device, Index2 >& matrix,
- const RealType& matrixMultiplicator = 1.0 );
-#endif
template< typename Vector >
__cuda_callable__
@@ -165,6 +174,14 @@ class Tridiagonal : public Matrix< Real, Device, Index >
Vector& x,
const RealType& omega = 1.0 ) const;
+ // copy assignment
+ Tridiagonal& operator=( const Tridiagonal& matrix );
+
+ // cross-device copy assignment
+ template< typename Real2, typename Device2, typename Index2,
+ typename = typename Enabler< Device2 >::type >
+ Tridiagonal& operator=( const Tridiagonal< Real2, Device2, Index2 >& matrix );
+
bool save( File& file ) const;
bool load( File& file );
@@ -175,7 +192,7 @@ class Tridiagonal : public Matrix< Real, Device, Index >
void print( std::ostream& str ) const;
- protected:
+protected:
__cuda_callable__
IndexType getElementIndex( const IndexType row,
diff --git a/src/TNL/Matrices/Tridiagonal_impl.h b/src/TNL/Matrices/Tridiagonal_impl.h
index 9d15ba841de1a7d64d284aa4e6301d8b9e6f6e48..f3c073cd0d65285643df34d8fc2885e802c77e36 100644
--- a/src/TNL/Matrices/Tridiagonal_impl.h
+++ b/src/TNL/Matrices/Tridiagonal_impl.h
@@ -48,44 +48,65 @@ String Tridiagonal< Real, Device, Index >::getTypeVirtual() const
template< typename Real,
typename Device,
typename Index >
-bool Tridiagonal< Real, Device, Index >::setDimensions( const IndexType rows,
- const IndexType columns )
+String Tridiagonal< Real, Device, Index >::getSerializationType()
{
- if( ! Matrix< Real, Device, Index >::setDimensions( rows, columns ) )
- return false;
- if( ! values.setSize( 3*min( rows, columns ) ) )
- return false;
+ return getType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+String Tridiagonal< Real, Device, Index >::getSerializationTypeVirtual() const
+{
+ return this->getSerializationType();
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+void Tridiagonal< Real, Device, Index >::setDimensions( const IndexType rows,
+ const IndexType columns )
+{
+ Matrix< Real, Device, Index >::setDimensions( rows, columns );
+ values.setSize( 3*min( rows, columns ) );
this->values.setValue( 0.0 );
- return true;
}
template< typename Real,
typename Device,
typename Index >
-bool Tridiagonal< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
+void Tridiagonal< Real, Device, Index >::setCompressedRowLengths( const CompressedRowLengthsVector& rowLengths )
{
if( rowLengths[ 0 ] > 2 )
- return false;
+ throw std::logic_error( "Too many non-zero elements per row in a tri-diagonal matrix." );
const IndexType diagonalLength = min( this->getRows(), this->getColumns() );
for( Index i = 1; i < diagonalLength-1; i++ )
if( rowLengths[ i ] > 3 )
- return false;
+ throw std::logic_error( "Too many non-zero elements per row in a tri-diagonal matrix." );
if( this->getRows() > this->getColumns() )
if( rowLengths[ this->getRows()-1 ] > 1 )
- return false;
+ throw std::logic_error( "Too many non-zero elements per row in a tri-diagonal matrix." );
if( this->getRows() == this->getColumns() )
if( rowLengths[ this->getRows()-1 ] > 2 )
- return false;
+ throw std::logic_error( "Too many non-zero elements per row in a tri-diagonal matrix." );
if( this->getRows() < this->getColumns() )
if( rowLengths[ this->getRows()-1 ] > 3 )
- return false;
- return true;
+ throw std::logic_error( "Too many non-zero elements per row in a tri-diagonal matrix." );
}
template< typename Real,
typename Device,
typename Index >
Index Tridiagonal< Real, Device, Index >::getRowLength( const IndexType row ) const
+{
+ return this->getRowLengthFast( row );
+}
+
+template< typename Real,
+ typename Device,
+ typename Index >
+__cuda_callable__
+Index Tridiagonal< Real, Device, Index >::getRowLengthFast( const IndexType row ) const
{
const IndexType diagonalLength = min( this->getRows(), this->getColumns() );
if( row == 0 )
@@ -111,9 +132,9 @@ template< typename Real,
typename Device,
typename Index >
template< typename Real2, typename Device2, typename Index2 >
-bool Tridiagonal< Real, Device, Index >::setLike( const Tridiagonal< Real2, Device2, Index2 >& m )
+void Tridiagonal< Real, Device, Index >::setLike( const Tridiagonal< Real2, Device2, Index2 >& m )
{
- return this->setDimensions( m.getRows(), m.getColumns() );
+ this->setDimensions( m.getRows(), m.getColumns() );
}
template< typename Real,
@@ -506,7 +527,7 @@ void Tridiagonal< Real, Device, Index >::getTransposition( const Tridiagonal< Re
}
Devices::Cuda::freeFromDevice( kernel_this );
Devices::Cuda::freeFromDevice( kernel_inMatrix );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
}
@@ -529,6 +550,39 @@ void Tridiagonal< Real, Device, Index >::performSORIteration( const Vector& b,
x[ row ] = ( 1.0 - omega ) * x[ row ] + omega / this->getElementFast( row, row ) * ( b[ row ] - sum );
}
+
+// copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+Tridiagonal< Real, Device, Index >&
+Tridiagonal< Real, Device, Index >::operator=( const Tridiagonal& matrix )
+{
+ this->setLike( matrix );
+ this->values = matrix.values;
+ return *this;
+}
+
+// cross-device copy assignment
+template< typename Real,
+ typename Device,
+ typename Index >
+ template< typename Real2, typename Device2, typename Index2, typename >
+Tridiagonal< Real, Device, Index >&
+Tridiagonal< Real, Device, Index >::operator=( const Tridiagonal< Real2, Device2, Index2 >& matrix )
+{
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::Cuda >::value,
+ "unknown device" );
+ static_assert( std::is_same< Device2, Devices::Host >::value || std::is_same< Device2, Devices::Cuda >::value,
+ "unknown device" );
+
+ this->setLike( matrix );
+
+ std::cerr << "Cross-device assignment for the Tridiagonal format is not implemented yet." << std::endl;
+ throw 1;
+}
+
+
template< typename Real,
typename Device,
typename Index >
diff --git a/src/TNL/Meshes/CMakeLists.txt b/src/TNL/Meshes/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Meshes/DummyMesh.h b/src/TNL/Meshes/DummyMesh.h
index efbf1be40615463610df3a4ceec361b634884582..ce4cd0bf42f92d8911fa482ca240e5e919ba00e1 100644
--- a/src/TNL/Meshes/DummyMesh.h
+++ b/src/TNL/Meshes/DummyMesh.h
@@ -25,10 +25,7 @@ class DummyMesh
typedef Index IndexType;
typedef DummyMesh< Real, Device, Index > ThisType;
- static const int meshDimension = 1;
-
- constexpr static int getMeshDimension() { return meshDimension; }
-
+ constexpr static int getMeshDimension() { return 1; }
const Real& getParametricStep(){ return 0.0; }
diff --git a/src/TNL/Meshes/GridDetails/CMakeLists.txt b/src/TNL/Meshes/GridDetails/CMakeLists.txt
index 7e39623121c8dba1ee95332a69fada92a561bbd5..eeecd7bc50bbbac8492066be8f7bf9bf20b19e4d 100644
--- a/src/TNL/Meshes/GridDetails/CMakeLists.txt
+++ b/src/TNL/Meshes/GridDetails/CMakeLists.txt
@@ -15,11 +15,11 @@ SET( headers BoundaryGridEntityChecker.h
GridEntityTopology.h
GridTraverser.h
GridTraverser_impl.h
- NeighbourGridEntitiesStorage.h
- NeighbourGridEntityGetter1D_impl.h
- NeighbourGridEntityGetter2D_impl.h
- NeighbourGridEntityGetter3D_impl.h
- NeighbourGridEntityGetter.h
+ NeighborGridEntitiesStorage.h
+ NeighborGridEntityGetter1D_impl.h
+ NeighborGridEntityGetter2D_impl.h
+ NeighborGridEntityGetter3D_impl.h
+ NeighborGridEntityGetter.h
Traverser_Grid1D.h
Traverser_Grid1D_impl.h
Traverser_Grid2D.h
diff --git a/src/TNL/Meshes/GridDetails/Grid1D.h b/src/TNL/Meshes/GridDetails/Grid1D.h
index 0cec1e2deb13a08b7e4ad6613ab0d8c078b4f7c4..fdf3f932ea071f78dc605ea9108d3b0a35528cf8 100644
--- a/src/TNL/Meshes/GridDetails/Grid1D.h
+++ b/src/TNL/Meshes/GridDetails/Grid1D.h
@@ -14,7 +14,7 @@
#include <TNL/Logger.h>
#include <TNL/Meshes/GridDetails/GridEntityTopology.h>
#include <TNL/Meshes/GridDetails/GridEntityGetter.h>
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter.h>
#include <TNL/Meshes/GridEntity.h>
#include <TNL/Meshes/GridEntityConfig.h>
#include <TNL/Meshes/DistributedGrid.h>
@@ -31,25 +31,26 @@ class Grid< 1, Real, Device, Index > : public Object
typedef Real RealType;
typedef Device DeviceType;
- typedef Index IndexType;
+ typedef Index GlobalIndexType;
typedef Containers::StaticVector< 1, Real > PointType;
typedef Containers::StaticVector< 1, Index > CoordinatesType;
typedef Grid< 1, Real, Devices::Host, Index > HostType;
typedef Grid< 1, Real, Devices::Cuda, Index > CudaType;
typedef Grid< 1, Real, Device, Index > ThisType;
+
+ // TODO: deprecated and to be removed (GlobalIndexType shall be used instead)
+ typedef Index IndexType;
- static const int meshDimension = 1;
+ static constexpr int getMeshDimension() { return 1; };
template< int EntityDimension,
typename Config = GridEntityCrossStencilStorage< 1 > >
- using MeshEntity = GridEntity< ThisType, EntityDimension, Config >;
+ using EntityType = GridEntity< ThisType, EntityDimension, Config >;
- typedef MeshEntity< meshDimension, GridEntityCrossStencilStorage< 1 > > Cell;
- typedef MeshEntity< 0 > Face;
- typedef MeshEntity< 0 > Vertex;
+ typedef EntityType< getMeshDimension(), GridEntityCrossStencilStorage< 1 > > Cell;
+ typedef EntityType< 0 > Face;
+ typedef EntityType< 0 > Vertex;
- static constexpr int getMeshDimension() { return meshDimension; };
-
Grid();
static String getType();
@@ -79,29 +80,22 @@ class Grid< 1, Real, Device, Index > : public Object
__cuda_callable__
inline const PointType& getProportions() const;
- template< typename EntityType >
+
+ template< int EntiytDimension >
__cuda_callable__
IndexType getEntitiesCount() const;
-
- template< int EntiytDimensions >
+
+ template< typename Entity >
__cuda_callable__
IndexType getEntitiesCount() const;
-
- template< typename EntityType >
+ template< typename Entity >
__cuda_callable__
- EntityType getEntity( const IndexType& entityIndex ) const;
-
- template< typename EntityType >
- __cuda_callable__
- Index getEntityIndex( const EntityType& entity ) const;
-
- template< typename EntityType >
- __cuda_callable__
- RealType getEntityMeasure( const EntityType& entity ) const;
+ inline Entity getEntity( const IndexType& entityIndex ) const;
+ template< typename Entity >
__cuda_callable__
- inline const RealType& getCellMeasure() const;
+ inline Index getEntityIndex( const Entity& entity ) const;
__cuda_callable__
@@ -115,7 +109,10 @@ class Grid< 1, Real, Device, Index > : public Object
const RealType& getSpaceStepsProducts() const;
__cuda_callable__
- RealType getSmallestSpaceStep() const;
+ inline const RealType& getCellMeasure() const;
+
+ __cuda_callable__
+ inline RealType getSmallestSpaceStep() const;
template< typename GridFunction >
diff --git a/src/TNL/Meshes/GridDetails/Grid1D_impl.h b/src/TNL/Meshes/GridDetails/Grid1D_impl.h
index f372599a12b8f667a213f1fbe4f009befaf30e24..f4504c17e897f5dade777c54322bf959fff6fce0 100644
--- a/src/TNL/Meshes/GridDetails/Grid1D_impl.h
+++ b/src/TNL/Meshes/GridDetails/Grid1D_impl.h
@@ -14,9 +14,10 @@
#include <iomanip>
#include <TNL/String.h>
#include <TNL/Assert.h>
+#include <TNL/Logger.h>
#include <TNL/Meshes/GridDetails/GnuplotWriter.h>
#include <TNL/Meshes/GridDetails/GridEntityGetter_impl.h>
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter1D_impl.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter1D_impl.h>
#include <TNL/Meshes/GridDetails/Grid1D.h>
#include <TNL/Meshes/GridDetails/GridEntityMeasureGetter.h>
@@ -117,7 +118,7 @@ template< typename Real,
typename Index >
void Grid< 1, Real, Device, Index >::setDimensions( const Index xSize )
{
- TNL_ASSERT( xSize > 0, std::cerr << "xSize = " << xSize );
+ TNL_ASSERT_GT( xSize, 0, "Grid size must be positive." );
this->dimensions.x() = xSize;
this->numberOfCells = xSize;
this->numberOfVertices = xSize + 1;
@@ -173,19 +174,20 @@ const typename Grid< 1, Real, Device, Index >::PointType&
return this->proportions;
}
+
template< typename Real,
typename Device,
typename Index >
- template< typename EntityType >
+ template< int EntityDimension >
__cuda_callable__ inline
Index
Grid< 1, Real, Device, Index >::
getEntitiesCount() const
{
- static_assert( EntityType::entityDimension <= 1 &&
- EntityType::entityDimension >= 0, "Wrong grid entity dimension." );
+ static_assert( EntityDimension <= 1 &&
+ EntityDimension >= 0, "Wrong grid entity dimensions." );
- switch( EntityType::entityDimension )
+ switch( EntityDimension )
{
case 1:
return this->numberOfCells;
@@ -198,77 +200,43 @@ getEntitiesCount() const
template< typename Real,
typename Device,
typename Index >
- template< int EntityDimensions >
+ template< typename Entity >
__cuda_callable__ inline
Index
Grid< 1, Real, Device, Index >::
getEntitiesCount() const
{
- static_assert( EntityDimensions <= 1 &&
- EntityDimensions >= 0, "Wrong grid entity dimensions." );
-
- switch( EntityDimensions )
- {
- case 1:
- return this->numberOfCells;
- case 0:
- return this->numberOfVertices;
- }
- return -1;
+ return getEntitiesCount< Entity::getEntityDimension() >();
}
-
template< typename Real,
typename Device,
typename Index >
- template< typename EntityType >
+ template< typename Entity >
__cuda_callable__ inline
-EntityType
+Entity
Grid< 1, Real, Device, Index >::
getEntity( const IndexType& entityIndex ) const
{
- static_assert( EntityType::entityDimension <= 1 &&
- EntityType::entityDimension >= 0, "Wrong grid entity dimension." );
+ static_assert( Entity::getEntityDimension() <= 1 &&
+ Entity::getEntityDimension() >= 0, "Wrong grid entity dimensions." );
- return GridEntityGetter< ThisType, EntityType >::getEntity( *this, entityIndex );
+ return GridEntityGetter< ThisType, Entity >::getEntity( *this, entityIndex );
}
template< typename Real,
typename Device,
typename Index >
- template< typename EntityType >
+ template< typename Entity >
__cuda_callable__ inline
Index
Grid< 1, Real, Device, Index >::
-getEntityIndex( const EntityType& entity ) const
+getEntityIndex( const Entity& entity ) const
{
- static_assert( EntityType::entityDimension <= 1 &&
- EntityType::entityDimension >= 0, "Wrong grid entity dimension." );
+ static_assert( Entity::getEntityDimension() <= 1 &&
+ Entity::getEntityDimension() >= 0, "Wrong grid entity dimensions." );
- return GridEntityGetter< ThisType, EntityType >::getEntityIndex( *this, entity );
-}
-
-template< typename Real,
- typename Device,
- typename Index >
- template< typename EntityType >
-__cuda_callable__
-Real
-Grid< 1, Real, Device, Index >::
-getEntityMeasure( const EntityType& entity ) const
-{
- return GridEntityMeasureGetter< ThisType, EntityType::getDimensions() >::getMeasure( *this, entity );
-}
-
-template< typename Real,
- typename Device,
- typename Index >
-__cuda_callable__
-const Real&
-Grid< 1, Real, Device, Index >::
-getCellMeasure() const
-{
- return this->template getSpaceStepsProducts< 1 >();
+ return GridEntityGetter< ThisType, Entity >::getEntityIndex( *this, entity );
}
template< typename Real,
@@ -303,11 +271,21 @@ const Real&
Grid< 1, Real, Device, Index >::
getSpaceStepsProducts() const
{
- TNL_ASSERT( xPow >= -2 && xPow <= 2,
- std::cerr << " xPow = " << xPow );
+ static_assert( xPow >= -2 && xPow <= 2, "unsupported value of xPow" );
return this->spaceStepsProducts[ xPow + 2 ];
}
+template< typename Real,
+ typename Device,
+ typename Index >
+__cuda_callable__
+const Real&
+Grid< 1, Real, Device, Index >::
+getCellMeasure() const
+{
+ return this->template getSpaceStepsProducts< 1 >();
+}
+
template< typename Real,
typename Device,
typename Index >
@@ -469,7 +447,7 @@ bool Grid< 1, Real, Device, Index >::write( const MeshFunction& function,
const RealType hx = getSpaceSteps(). x();
if( format == "gnuplot" )
{
- typename ThisType::template MeshEntity< getMeshDimension() > entity( *this );
+ typename ThisType::template EntityType< getMeshDimension() > entity( *this );
for( entity.getCoordinates().x() = 0;
entity.getCoordinates().x() < getDimensions(). x();
entity.getCoordinates().x() ++ )
@@ -496,6 +474,8 @@ writeProlog( Logger& logger )
logger.writeParameter( "Domain proportions:", this->proportions );
logger.writeParameter( "Domain dimensions:", this->dimensions );
logger.writeParameter( "Space steps:", this->getSpaceSteps() );
+ logger.writeParameter( "Number of cells:", getEntitiesCount< Cell >() );
+ logger.writeParameter( "Number of vertices:", getEntitiesCount< Vertex >() );
}
} // namespace Meshes
diff --git a/src/TNL/Meshes/GridDetails/Grid2D.h b/src/TNL/Meshes/GridDetails/Grid2D.h
index 9e2bcdfb52b491f9879f9d1c4a07e64a73635189..41a0a27d6f2cc82ab0ecf75a0a67a0c995b869b9 100644
--- a/src/TNL/Meshes/GridDetails/Grid2D.h
+++ b/src/TNL/Meshes/GridDetails/Grid2D.h
@@ -11,9 +11,10 @@
#pragma once
#include <TNL/Meshes/Grid.h>
+#include <TNL/Meshes/GridEntity.h>
#include <TNL/Meshes/GridDetails/GridEntityTopology.h>
#include <TNL/Meshes/GridDetails/GridEntityGetter.h>
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter.h>
namespace TNL {
namespace Meshes {
@@ -27,33 +28,34 @@ class Grid< 2, Real, Device, Index > : public Object
typedef Real RealType;
typedef Device DeviceType;
- typedef Index IndexType;
+ typedef Index GlobalIndexType;
typedef Containers::StaticVector< 2, Real > PointType;
typedef Containers::StaticVector< 2, Index > CoordinatesType;
typedef Grid< 2, Real, Devices::Host, Index > HostType;
typedef Grid< 2, Real, Devices::Cuda, Index > CudaType;
typedef Grid< 2, Real, Device, Index > ThisType;
- static const int meshDimension = 2;
+ // TODO: deprecated and to be removed (GlobalIndexType shall be used instead)
+ typedef Index IndexType;
+
+ static constexpr int getMeshDimension() { return 2; };
template< int EntityDimension,
typename Config = GridEntityNoStencilStorage >//CrossStencilStorage< 1 > >
- using MeshEntity = GridEntity< ThisType, EntityDimension, Config >;
+ using EntityType = GridEntity< ThisType, EntityDimension, Config >;
- typedef MeshEntity< meshDimension, GridEntityCrossStencilStorage< 1 > > Cell;
- typedef MeshEntity< meshDimension - 1, GridEntityNoStencilStorage > Face;
- typedef MeshEntity< 0 > Vertex;
+ typedef EntityType< getMeshDimension(), GridEntityCrossStencilStorage< 1 > > Cell;
+ typedef EntityType< getMeshDimension() - 1, GridEntityNoStencilStorage > Face;
+ typedef EntityType< 0 > Vertex;
// TODO: remove this
//template< int EntityDimension,
// typename Config = GridEntityNoStencilStorage >//CrossStencilStorage< 1 > >
- //using TestMeshEntity = tnlTestGridEntity< ThisType, EntityDimension, Config >;
- //typedef TestMeshEntity< meshDimension, GridEntityCrossStencilStorage< 1 > > TestCell;
+ //using TestEntityType = tnlTestGridEntity< ThisType, EntityDimension, Config >;
+ //typedef TestEntityType< getMeshDimension(), GridEntityCrossStencilStorage< 1 > > TestCell;
/////
- static constexpr int getMeshDimension() { return meshDimension; };
-
Grid();
static String getType();
@@ -83,29 +85,22 @@ class Grid< 2, Real, Device, Index > : public Object
inline const PointType& getProportions() const;
- template< int EntityDimensions >
+ template< int EntityDimension >
__cuda_callable__
IndexType getEntitiesCount() const;
- template< typename EntityType >
- __cuda_callable__
- IndexType getEntitiesCount() const;
-
- template< typename EntityType >
- __cuda_callable__
- EntityType getEntity( const IndexType& entityIndex ) const;
-
- template< typename EntityType >
+ template< typename Entity >
__cuda_callable__
- Index getEntityIndex( const EntityType& entity ) const;
-
- template< typename EntityType >
- __cuda_callable__
- RealType getEntityMeasure( const EntityType& entity ) const;
+ inline IndexType getEntitiesCount() const;
+ template< typename Entity >
__cuda_callable__
- inline const RealType& getCellMeasure() const;
+ inline Entity getEntity( const IndexType& entityIndex ) const;
+ template< typename Entity >
+ __cuda_callable__
+ inline Index getEntityIndex( const Entity& entity ) const;
+
__cuda_callable__
inline const PointType& getSpaceSteps() const;
@@ -117,7 +112,10 @@ class Grid< 2, Real, Device, Index > : public Object
const RealType& getSpaceStepsProducts() const;
__cuda_callable__
- RealType getSmallestSpaceStep() const;
+ inline const RealType& getCellMeasure() const;
+
+ __cuda_callable__
+ inline RealType getSmallestSpaceStep() const;
template< typename GridFunction >
diff --git a/src/TNL/Meshes/GridDetails/Grid2D_impl.h b/src/TNL/Meshes/GridDetails/Grid2D_impl.h
index bb4a3c2944d496c1438e1bab70c437f75192b4f0..1df3f13630dcee66d2eebee7bca19a6c7ca20957 100644
--- a/src/TNL/Meshes/GridDetails/Grid2D_impl.h
+++ b/src/TNL/Meshes/GridDetails/Grid2D_impl.h
@@ -12,10 +12,13 @@
#include <fstream>
#include <iomanip>
+#include <TNL/String.h>
#include <TNL/Assert.h>
+#include <TNL/Logger.h>
#include <TNL/Meshes/GridDetails/GnuplotWriter.h>
#include <TNL/Meshes/GridDetails/GridEntityGetter_impl.h>
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter2D_impl.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter2D_impl.h>
+#include <TNL/Meshes/GridDetails/Grid2D.h>
#include <TNL/Meshes/GridDetails/GridEntityMeasureGetter.h>
namespace TNL {
@@ -152,8 +155,8 @@ template< typename Real,
typename Index >
void Grid< 2, Real, Device, Index > :: setDimensions( const Index xSize, const Index ySize )
{
- TNL_ASSERT( xSize > 0, std::cerr << "xSize = " << xSize );
- TNL_ASSERT( ySize > 0, std::cerr << "ySize = " << ySize );
+ TNL_ASSERT_GT( xSize, 0, "Grid size must be positive." );
+ TNL_ASSERT_GT( ySize, 0, "Grid size must be positive." );
this->dimensions.x() = xSize;
this->dimensions.y() = ySize;
@@ -222,19 +225,20 @@ const typename Grid< 2, Real, Device, Index > :: PointType&
return this->proportions;
}
+
template< typename Real,
typename Device,
typename Index >
- template< int EntityDimensions >
+ template< int EntityDimension >
__cuda_callable__ inline
Index
Grid< 2, Real, Device, Index >::
getEntitiesCount() const
{
- static_assert( EntityDimensions <= 2 &&
- EntityDimensions >= 0, "Wrong grid entity dimensions." );
+ static_assert( EntityDimension <= 2 &&
+ EntityDimension >= 0, "Wrong grid entity dimensions." );
- switch( EntityDimensions )
+ switch( EntityDimension )
{
case 2:
return this->numberOfCells;
@@ -246,85 +250,48 @@ getEntitiesCount() const
return -1;
}
-
template< typename Real,
typename Device,
typename Index >
- template< typename EntityType >
+ template< typename Entity >
__cuda_callable__ inline
Index
Grid< 2, Real, Device, Index >::
getEntitiesCount() const
{
- static_assert( EntityType::entityDimension <= 2 &&
- EntityType::entityDimension >= 0, "Wrong grid entity dimension." );
-
- switch( EntityType::entityDimension )
- {
- case 2:
- return this->numberOfCells;
- case 1:
- return this->numberOfFaces;
- case 0:
- return this->numberOfVertices;
- }
- return -1;
+ return getEntitiesCount< Entity::getEntityDimension() >();
}
template< typename Real,
typename Device,
typename Index >
- template< typename EntityType >
+ template< typename Entity >
__cuda_callable__ inline
-EntityType
+Entity
Grid< 2, Real, Device, Index >::
getEntity( const IndexType& entityIndex ) const
{
- static_assert( EntityType::entityDimension <= 2 &&
- EntityType::entityDimension >= 0, "Wrong grid entity dimension." );
+ static_assert( Entity::getEntityDimension() <= 2 &&
+ Entity::getEntityDimension() >= 0, "Wrong grid entity dimensions." );
- return GridEntityGetter< ThisType, EntityType >::getEntity( *this, entityIndex );
+ return GridEntityGetter< ThisType, Entity >::getEntity( *this, entityIndex );
}
template< typename Real,
typename Device,
typename Index >
- template< typename EntityType >
+ template< typename Entity >
__cuda_callable__ inline
Index
Grid< 2, Real, Device, Index >::
-getEntityIndex( const EntityType& entity ) const
+getEntityIndex( const Entity& entity ) const
{
- static_assert( EntityType::entityDimension <= 2 &&
- EntityType::entityDimension >= 0, "Wrong grid entity dimension." );
+ static_assert( Entity::getEntityDimension() <= 2 &&
+ Entity::getEntityDimension() >= 0, "Wrong grid entity dimensions." );
- return GridEntityGetter< ThisType, EntityType >::getEntityIndex( *this, entity );
+ return GridEntityGetter< ThisType, Entity >::getEntityIndex( *this, entity );
}
-template< typename Real,
- typename Device,
- typename Index >
- template< typename EntityType >
-__cuda_callable__
-Real
-Grid< 2, Real, Device, Index >::
-getEntityMeasure( const EntityType& entity ) const
-{
- return GridEntityMeasureGetter< ThisType, EntityType::getDimensions() >::getMeasure( *this, entity );
-}
-
-template< typename Real,
- typename Device,
- typename Index >
-__cuda_callable__
-const Real&
-Grid< 2, Real, Device, Index >::
-getCellMeasure() const
-{
- return this->template getSpaceStepsProducts< 1, 1 >();
-}
-
-
template< typename Real,
typename Device,
typename Index >
@@ -357,14 +324,22 @@ const Real&
Grid< 2, Real, Device, Index >::
getSpaceStepsProducts() const
{
- TNL_ASSERT( xPow >= -2 && xPow <= 2,
- std::cerr << " xPow = " << xPow );
- TNL_ASSERT( yPow >= -2 && yPow <= 2,
- std::cerr << " yPow = " << yPow );
-
+ static_assert( xPow >= -2 && xPow <= 2, "unsupported value of xPow" );
+ static_assert( yPow >= -2 && yPow <= 2, "unsupported value of yPow" );
return this->spaceStepsProducts[ xPow + 2 ][ yPow + 2 ];
}
+template< typename Real,
+ typename Device,
+ typename Index >
+__cuda_callable__
+const Real&
+Grid< 2, Real, Device, Index >::
+getCellMeasure() const
+{
+ return this->template getSpaceStepsProducts< 1, 1 >();
+}
+
template< typename Real,
typename Device,
typename Index >
@@ -531,7 +506,7 @@ bool Grid< 2, Real, Device, Index > :: writeMesh( const String& fileName,
<< this->getProportions(). x() << "cm , "
<< this->getProportions(). y() << "cm );"
<< std::endl << std::endl;
- MeshEntity< 0 > vertex( *this );
+ Vertex vertex( *this );
CoordinatesType& vertexCoordinates = vertex.getCoordinates();
PointType v;
for( Index j = 0; j < this->dimensions. y(); j ++ )
@@ -569,7 +544,7 @@ bool Grid< 2, Real, Device, Index > :: writeMesh( const String& fileName,
}
file << std::endl;
- MeshEntity< 2 > cell( *this );
+ Cell cell( *this );
CoordinatesType& cellCoordinates = cell.getCoordinates();
const RealType cellMeasure = this->getSpaceSteps().x() * this->getSpaceSteps().y();
for( Index i = 0; i < this->dimensions. x(); i ++ )
@@ -692,6 +667,9 @@ writeProlog( Logger& logger )
logger.writeParameter( "Domain proportions:", this->proportions );
logger.writeParameter( "Domain dimensions:", this->dimensions );
logger.writeParameter( "Space steps:", this->getSpaceSteps() );
+ logger.writeParameter( "Number of cells:", getEntitiesCount< Cell >() );
+ logger.writeParameter( "Number of faces:", getEntitiesCount< Face >() );
+ logger.writeParameter( "Number of vertices:", getEntitiesCount< Vertex >() );
}
} // namespace Meshes
diff --git a/src/TNL/Meshes/GridDetails/Grid3D.h b/src/TNL/Meshes/GridDetails/Grid3D.h
index 3ba3b2505e4bdf8f1c65bafd95f2be4680ce4768..2321d888fa557abaab05fd5f0deec0bb5aceb99f 100644
--- a/src/TNL/Meshes/GridDetails/Grid3D.h
+++ b/src/TNL/Meshes/GridDetails/Grid3D.h
@@ -11,9 +11,10 @@
#pragma once
#include <TNL/Meshes/Grid.h>
+#include <TNL/Meshes/GridEntity.h>
#include <TNL/Meshes/GridDetails/GridEntityTopology.h>
#include <TNL/Meshes/GridDetails/GridEntityGetter.h>
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter.h>
namespace TNL {
namespace Meshes {
@@ -27,25 +28,26 @@ class Grid< 3, Real, Device, Index > : public Object
typedef Real RealType;
typedef Device DeviceType;
- typedef Index IndexType;
+ typedef Index GlobalIndexType;
typedef Containers::StaticVector< 3, Real > PointType;
typedef Containers::StaticVector< 3, Index > CoordinatesType;
typedef Grid< 3, Real, Devices::Host, Index > HostType;
typedef Grid< 3, Real, Devices::Cuda, Index > CudaType;
typedef Grid< 3, Real, Device, Index > ThisType;
- static const int meshDimension = 3;
+ // TODO: deprecated and to be removed (GlobalIndexType shall be used instead)
+ typedef Index IndexType;
+
+ static constexpr int getMeshDimension() { return 3; };
template< int EntityDimension,
typename Config = GridEntityCrossStencilStorage< 1 > >
- using MeshEntity = GridEntity< ThisType, EntityDimension, Config >;
-
- typedef MeshEntity< meshDimension, GridEntityCrossStencilStorage< 1 > > Cell;
- typedef MeshEntity< meshDimension - 1 > Face;
- typedef MeshEntity< 1 > Edge;
- typedef MeshEntity< 0 > Vertex;
+ using EntityType = GridEntity< ThisType, EntityDimension, Config >;
- static constexpr int getMeshDimension() { return meshDimension; };
+ typedef EntityType< getMeshDimension(), GridEntityCrossStencilStorage< 1 > > Cell;
+ typedef EntityType< getMeshDimension() - 1 > Face;
+ typedef EntityType< 1 > Edge;
+ typedef EntityType< 0 > Vertex;
Grid();
@@ -72,29 +74,23 @@ class Grid< 3, Real, Device, Index > : public Object
__cuda_callable__
inline const PointType& getProportions() const;
- template< typename EntityType >
+
+ template< int EntityDimension >
__cuda_callable__
IndexType getEntitiesCount() const;
-
- template< int Dimensions >
+
+ template< typename Entity >
__cuda_callable__
IndexType getEntitiesCount() const;
-
- template< typename EntityType >
- __cuda_callable__
- EntityType getEntity( const IndexType& entityIndex ) const;
-
- template< typename EntityType >
- __cuda_callable__
- Index getEntityIndex( const EntityType& entity ) const;
-
- template< typename EntityType >
+
+ template< typename Entity >
__cuda_callable__
- RealType getEntityMeasure( const EntityType& entity ) const;
+ inline Entity getEntity( const IndexType& entityIndex ) const;
+ template< typename Entity >
__cuda_callable__
- inline const RealType& getCellMeasure() const;
-
+ inline Index getEntityIndex( const Entity& entity ) const;
+
__cuda_callable__
inline const PointType& getSpaceSteps() const;
@@ -102,6 +98,9 @@ class Grid< 3, Real, Device, Index > : public Object
__cuda_callable__
const RealType& getSpaceStepsProducts() const;
+ __cuda_callable__
+ inline const RealType& getCellMeasure() const;
+
__cuda_callable__
RealType getSmallestSpaceStep() const;
@@ -155,7 +154,7 @@ class Grid< 3, Real, Device, Index > : public Object
PointType origin, proportions;
- IndexType cellZNeighboursStep;
+ IndexType cellZNeighborsStep;
PointType spaceSteps;
@@ -165,7 +164,7 @@ class Grid< 3, Real, Device, Index > : public Object
friend class GridEntityGetter;
template< typename, int, typename >
- friend class NeighbourGridEntityGetter;
+ friend class NeighborGridEntityGetter;
};
} // namespace Meshes
diff --git a/src/TNL/Meshes/GridDetails/Grid3D_impl.h b/src/TNL/Meshes/GridDetails/Grid3D_impl.h
index d8e4d175fcfacf6192542f536f8f873ac0728130..b6cb533c3dc723a99163cc0b2b84dece2b9e3866 100644
--- a/src/TNL/Meshes/GridDetails/Grid3D_impl.h
+++ b/src/TNL/Meshes/GridDetails/Grid3D_impl.h
@@ -10,10 +10,14 @@
#pragma once
+#include <fstream>
#include <iomanip>
+#include <TNL/String.h>
#include <TNL/Assert.h>
+#include <TNL/Logger.h>
+#include <TNL/Meshes/GridDetails/GnuplotWriter.h>
#include <TNL/Meshes/GridDetails/GridEntityGetter_impl.h>
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter3D_impl.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter3D_impl.h>
#include <TNL/Meshes/GridDetails/Grid3D.h>
#include <TNL/Meshes/GridDetails/GridEntityMeasureGetter.h>
@@ -45,7 +49,7 @@ template< typename Real,
String Grid< 3, Real, Device, Index > :: getType()
{
return String( "Meshes::Grid< " ) +
- String( meshDimension ) + ", " +
+ String( getMeshDimension() ) + ", " +
String( TNL::getType< RealType >() ) + ", " +
String( Device :: getDeviceType() ) + ", " +
String( TNL::getType< IndexType >() ) + " >";
@@ -164,9 +168,9 @@ template< typename Real,
typename Index >
void Grid< 3, Real, Device, Index > :: setDimensions( const Index xSize, const Index ySize, const Index zSize )
{
- TNL_ASSERT( xSize > 0, std::cerr << "xSize = " << xSize );
- TNL_ASSERT( ySize > 0, std::cerr << "ySize = " << ySize );
- TNL_ASSERT( zSize > 0, std::cerr << "zSize = " << zSize );
+ TNL_ASSERT_GT( xSize, 0, "Grid size must be positive." );
+ TNL_ASSERT_GT( ySize, 0, "Grid size must be positive." );
+ TNL_ASSERT_GT( zSize, 0, "Grid size must be positive." );
this->dimensions.x() = xSize;
this->dimensions.y() = ySize;
@@ -188,7 +192,7 @@ void Grid< 3, Real, Device, Index > :: setDimensions( const Index xSize, const I
this->numberOfDzEdges;
this->numberOfVertices = ( xSize + 1 ) * ( ySize + 1 ) * ( zSize + 1 );
- this->cellZNeighboursStep = xSize * ySize;
+ this->cellZNeighborsStep = xSize * ySize;
computeSpaceSteps();
}
@@ -242,19 +246,20 @@ const typename Grid< 3, Real, Device, Index > :: PointType&
return this->proportions;
}
+
template< typename Real,
typename Device,
typename Index >
- template< typename EntityType >
+ template< int EntityDimension >
__cuda_callable__ inline
Index
Grid< 3, Real, Device, Index >::
getEntitiesCount() const
{
- static_assert( EntityType::entityDimension <= 3 &&
- EntityType::entityDimension >= 0, "Wrong grid entity dimension." );
+ static_assert( EntityDimension <= 3 &&
+ EntityDimension >= 0, "Wrong grid entity dimensions." );
- switch( EntityType::entityDimension )
+ switch( EntityDimension )
{
case 3:
return this->numberOfCells;
@@ -271,80 +276,43 @@ getEntitiesCount() const
template< typename Real,
typename Device,
typename Index >
- template< int EntityDimensions >
+ template< typename Entity >
__cuda_callable__ inline
Index
Grid< 3, Real, Device, Index >::
getEntitiesCount() const
{
- static_assert( EntityDimensions <= 3 &&
- EntityDimensions >= 0, "Wrong grid entity dimensions." );
-
- switch( EntityDimensions )
- {
- case 3:
- return this->numberOfCells;
- case 2:
- return this->numberOfFaces;
- case 1:
- return this->numberOfEdges;
- case 0:
- return this->numberOfVertices;
- }
- return -1;
+ return getEntitiesCount< Entity::getEntityDimension() >();
}
template< typename Real,
typename Device,
typename Index >
- template< typename EntityType >
+ template< typename Entity >
__cuda_callable__ inline
-EntityType
+Entity
Grid< 3, Real, Device, Index >::
getEntity( const IndexType& entityIndex ) const
{
- static_assert( EntityType::entityDimension <= 3 &&
- EntityType::entityDimension >= 0, "Wrong grid entity dimension." );
+ static_assert( Entity::getEntityDimension() <= 3 &&
+ Entity::getEntityDimension() >= 0, "Wrong grid entity dimensions." );
- return GridEntityGetter< ThisType, EntityType >::getEntity( *this, entityIndex );
+ return GridEntityGetter< ThisType, Entity >::getEntity( *this, entityIndex );
}
template< typename Real,
typename Device,
typename Index >
- template< typename EntityType >
+ template< typename Entity >
__cuda_callable__ inline
Index
Grid< 3, Real, Device, Index >::
-getEntityIndex( const EntityType& entity ) const
+getEntityIndex( const Entity& entity ) const
{
- static_assert( EntityType::entityDimension <= 3 &&
- EntityType::entityDimension >= 0, "Wrong grid entity dimension." );
+ static_assert( Entity::getEntityDimension() <= 3 &&
+ Entity::getEntityDimension() >= 0, "Wrong grid entity dimensions." );
- return GridEntityGetter< ThisType, EntityType >::getEntityIndex( *this, entity );
-}
-
-template< typename Real,
- typename Device,
- typename Index >
- template< typename EntityType >
-__cuda_callable__
-Real
-Grid< 3, Real, Device, Index >::
-getEntityMeasure( const EntityType& entity ) const
-{
- return GridEntityMeasureGetter< ThisType, EntityType::getDimensions() >::getMeasure( *this, entity );
-}
-
-template< typename Real,
- typename Device,
- typename Index >
-__cuda_callable__
-const Real&
-Grid< 3, Real, Device, Index >::
-getCellMeasure() const
-{
- return this->template getSpaceStepsProducts< 1, 1, 1 >();
+ return GridEntityGetter< ThisType, Entity >::getEntityIndex( *this, entity );
}
template< typename Real,
@@ -367,16 +335,23 @@ const Real&
Grid< 3, Real, Device, Index >::
getSpaceStepsProducts() const
{
- TNL_ASSERT( xPow >= -2 && xPow <= 2,
- std::cerr << " xPow = " << xPow );
- TNL_ASSERT( yPow >= -2 && yPow <= 2,
- std::cerr << " yPow = " << yPow );
- TNL_ASSERT( zPow >= -2 && zPow <= 2,
- std::cerr << " zPow = " << zPow );
-
+ static_assert( xPow >= -2 && xPow <= 2, "unsupported value of xPow" );
+ static_assert( yPow >= -2 && yPow <= 2, "unsupported value of yPow" );
+ static_assert( zPow >= -2 && zPow <= 2, "unsupported value of zPow" );
return this->spaceStepsProducts[ xPow + 2 ][ yPow + 2 ][ zPow + 2 ];
}
+template< typename Real,
+ typename Device,
+ typename Index >
+__cuda_callable__
+const Real&
+Grid< 3, Real, Device, Index >::
+getCellMeasure() const
+{
+ return this->template getSpaceStepsProducts< 1, 1, 1 >();
+}
+
template< typename Real,
typename Device,
typename Index >
@@ -405,7 +380,7 @@ typename GridFunction::RealType
const typename GridFunction::RealType& p ) const
{
typename GridFunction::RealType lpNorm( 0.0 );
- MeshEntity< getMeshDimension() > cell;
+ Cell cell;
for( cell.getCoordinates().z() = 0;
cell.getCoordinates().z() < getDimensions().z();
cell.getCoordinates().z()++ )
@@ -433,7 +408,7 @@ template< typename Real,
const GridFunction& f2 ) const
{
typename GridFunction::RealType maxDiff( -1.0 );
- MeshEntity< getMeshDimension() > cell( *this );
+ Cell cell( *this );
for( cell.getCoordinates().z() = 0;
cell.getCoordinates().z() < getDimensions().z();
cell.getCoordinates().z()++ )
@@ -460,8 +435,7 @@ template< typename Real,
const typename GridFunction::RealType& p ) const
{
typename GridFunction::RealType lpNorm( 0.0 );
- MeshEntity< getMeshDimension() > cell( *this );
-
+ Cell cell( *this );
for( cell.getCoordinates().z() = 0;
cell.getCoordinates().z() < getDimensions().z();
cell.getCoordinates().z()++ )
@@ -537,7 +511,9 @@ template< typename Real,
bool Grid< 3, Real, Device, Index >::writeMesh( const String& fileName,
const String& format ) const
{
- TNL_ASSERT( false, std::cerr << "TODO: FIX THIS"); // TODO: FIX THIS
+ /*****
+ * TODO: implement this
+ */
return true;
}
@@ -604,6 +580,9 @@ writeProlog( Logger& logger )
logger.writeParameter( "Domain proportions:", this->proportions );
logger.writeParameter( "Domain dimensions:", this->dimensions );
logger.writeParameter( "Space steps:", this->getSpaceSteps() );
+ logger.writeParameter( "Number of cells:", getEntitiesCount< Cell >() );
+ logger.writeParameter( "Number of faces:", getEntitiesCount< Face >() );
+ logger.writeParameter( "Number of vertices:", getEntitiesCount< Vertex >() );
}
} // namespace Meshes
diff --git a/src/TNL/Meshes/GridDetails/GridEntityGetter.h b/src/TNL/Meshes/GridDetails/GridEntityGetter.h
index 17f764052a8bda7d266dc209bab4bc1d43abe88a..1c092e41be541a66b30cb330a8ef6689cdcaf717 100644
--- a/src/TNL/Meshes/GridDetails/GridEntityGetter.h
+++ b/src/TNL/Meshes/GridDetails/GridEntityGetter.h
@@ -15,7 +15,7 @@ namespace Meshes {
template< typename Grid,
typename GridEntity,
- int EntityDimension = GridEntity::entityDimension >
+ int EntityDimension = GridEntity::getEntityDimension() >
class GridEntityGetter
{
//static_assert( false, "Wrong mesh type or entity topology." );
diff --git a/src/TNL/Meshes/GridDetails/GridEntityGetter_impl.h b/src/TNL/Meshes/GridDetails/GridEntityGetter_impl.h
index 15ad36fac490dd33f351566bde7cbd0af1228044..49bc86956d659edff903b15927a1f4753c80fab2 100644
--- a/src/TNL/Meshes/GridDetails/GridEntityGetter_impl.h
+++ b/src/TNL/Meshes/GridDetails/GridEntityGetter_impl.h
@@ -44,10 +44,8 @@ class GridEntityGetter<
static GridEntity getEntity( const GridType& grid,
const IndexType& index )
{
- TNL_ASSERT( index >= 0 && index < grid.template getEntitiesCount< GridEntity >(),
- std::cerr << " index = " << index
- << " grid.getEntitiesCount<>() = " << grid.template getEntitiesCount< GridEntity >()
- << " entityDimension = " << entityDimension );
+ TNL_ASSERT_GE( index, 0, "Index must be non-negative." );
+ TNL_ASSERT_LT( index, grid.template getEntitiesCount< GridEntity >(), "Index is out of bounds." );
return GridEntity
( grid,
CoordinatesType( index ),
@@ -59,11 +57,8 @@ class GridEntityGetter<
static IndexType getEntityIndex( const GridType& grid,
const GridEntity& entity )
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() < grid.getDimensions() + CoordinatesType( 1 - entityDimension ),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " grid.getDimensions() = " << grid.getDimensions()
- << " EntityDimension = " << entityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), grid.getDimensions() + CoordinatesType( 1 - entityDimension ), "wrong coordinates" );
return entity.getCoordinates().x();
}
};
@@ -90,10 +85,8 @@ class GridEntityGetter< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 2 >
static GridEntity getEntity( const GridType& grid,
const IndexType& index )
{
- TNL_ASSERT( index >= 0 && index < grid.template getEntitiesCount< GridEntity >(),
- std::cerr << " index = " << index
- << " grid.getEntitiesCount<>() = " << grid.template getEntitiesCount< GridEntity >()
- << " entityDimension = " << entityDimension );
+ TNL_ASSERT_GE( index, 0, "Index must be non-negative." );
+ TNL_ASSERT_LT( index, grid.template getEntitiesCount< GridEntity >(), "Index is out of bounds." );
const CoordinatesType dimensions = grid.getDimensions();
@@ -109,10 +102,8 @@ class GridEntityGetter< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 2 >
static IndexType getEntityIndex( const GridType& grid,
const GridEntity& entity )
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() < grid.getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " grid.getDimensions() = " << grid.getDimensions() );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), grid.getDimensions(), "wrong coordinates" );
//const CoordinatesType coordinates = entity.getCoordinates();
//const CoordinatesType dimensions = grid.getDimensions();
@@ -142,10 +133,8 @@ class GridEntityGetter< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 1 >
static GridEntity getEntity( const GridType& grid,
const IndexType& index )
{
- TNL_ASSERT( index >= 0 && index < grid.template getEntitiesCount< GridEntity >(),
- std::cerr << " index = " << index
- << " grid.getEntitiesCount<>() = " << grid.template getEntitiesCount< GridEntity >()
- << " entityDimension = " << entityDimension );
+ TNL_ASSERT_GE( index, 0, "Index must be non-negative." );
+ TNL_ASSERT_LT( index, grid.template getEntitiesCount< GridEntity >(), "Index is out of bounds." );
const CoordinatesType dimensions = grid.getDimensions();
@@ -171,11 +160,8 @@ class GridEntityGetter< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 1 >
static IndexType getEntityIndex( const GridType& grid,
const GridEntity& entity )
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() < grid.getDimensions() + abs( entity.getOrientation() ),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " grid.getDimensions() = " << grid.getDimensions()
- << " abs( entity.getOrientation() ) = " << abs( entity.getOrientation() ) );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), grid.getDimensions() + abs( entity.getOrientation() ), "wrong coordinates" );
const CoordinatesType coordinates = entity.getCoordinates();
const CoordinatesType dimensions = grid.getDimensions();
@@ -205,10 +191,8 @@ class GridEntityGetter< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 0 >
static GridEntity getEntity( const GridType& grid,
const IndexType& index )
{
- TNL_ASSERT( index >= 0 && index < grid.template getEntitiesCount< GridEntity >(),
- std::cerr << " index = " << index
- << " grid.getEntitiesCount<>() = " << grid.template getEntitiesCount< GridEntity >()
- << " entityDimension = " << entityDimension );
+ TNL_ASSERT_GE( index, 0, "Index must be non-negative." );
+ TNL_ASSERT_LT( index, grid.template getEntitiesCount< GridEntity >(), "Index is out of bounds." );
const CoordinatesType dimensions = grid.getDimensions();
@@ -225,9 +209,8 @@ class GridEntityGetter< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 0 >
static IndexType getEntityIndex( const GridType& grid,
const GridEntity& entity )
{
- TNL_ASSERT( entity.getCoordinates() >= 0 && entity.getCoordinates() <= grid.getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " grid.getDimensions() = " << grid.getDimensions() );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), grid.getDimensions(), "wrong coordinates" );
const CoordinatesType coordinates = entity.getCoordinates();
const CoordinatesType dimensions = grid.getDimensions();
@@ -258,10 +241,8 @@ class GridEntityGetter< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 3 >
static GridEntity getEntity( const GridType& grid,
const IndexType& index )
{
- TNL_ASSERT( index >= 0 && index < grid.template getEntitiesCount< GridEntity >(),
- std::cerr << " index = " << index
- << " grid.getEntitiesCount<>() = " << grid.template getEntitiesCount< GridEntity >()
- << " entityDimension = " << entityDimension );
+ TNL_ASSERT_GE( index, 0, "Index must be non-negative." );
+ TNL_ASSERT_LT( index, grid.template getEntitiesCount< GridEntity >(), "Index is out of bounds." );
const CoordinatesType dimensions = grid.getDimensions();
@@ -278,10 +259,8 @@ class GridEntityGetter< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 3 >
static IndexType getEntityIndex( const GridType& grid,
const GridEntity& entity )
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < grid.getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " grid.getDimensions() = " << grid.getDimensions() );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), grid.getDimensions(), "wrong coordinates" );
const CoordinatesType coordinates = entity.getCoordinates();
const CoordinatesType dimensions = grid.getDimensions();
@@ -310,10 +289,8 @@ class GridEntityGetter< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 2 >
static GridEntity getEntity( const GridType& grid,
const IndexType& index )
{
- TNL_ASSERT( index >= 0 && index < grid.template getEntitiesCount< GridEntity >(),
- std::cerr << " index = " << index
- << " grid.getEntitiesCount<>() = " << grid.template getEntitiesCount< GridEntity >()
- << " entityDimension = " << entityDimension );
+ TNL_ASSERT_GE( index, 0, "Index must be non-negative." );
+ TNL_ASSERT_LT( index, grid.template getEntitiesCount< GridEntity >(), "Index is out of bounds." );
const CoordinatesType dimensions = grid.getDimensions();
@@ -354,11 +331,8 @@ class GridEntityGetter< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 2 >
static IndexType getEntityIndex( const GridType& grid,
const GridEntity& entity )
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < grid.getDimensions() + abs( entity.getOrientation() ),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " grid.getDimensions() = " << grid.getDimensions()
- << " abs( entity.getOrientation() ) = " << abs( entity.getOrientation() ) );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), grid.getDimensions() + abs( entity.getOrientation() ), "wrong coordinates" );
const CoordinatesType coordinates = entity.getCoordinates();
const CoordinatesType dimensions = grid.getDimensions();
@@ -400,10 +374,8 @@ class GridEntityGetter< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 1 >
static GridEntity getEntity( const GridType& grid,
const IndexType& index )
{
- TNL_ASSERT( index >= 0 && index < grid.template getEntitiesCount< GridEntity >(),
- std::cerr << " index = " << index
- << " grid.getEntitiesCount<>() = " << grid.template getEntitiesCount< GridEntity >()
- << " entityDimension = " << entityDimension );
+ TNL_ASSERT_GE( index, 0, "Index must be non-negative." );
+ TNL_ASSERT_LT( index, grid.template getEntitiesCount< GridEntity >(), "Index is out of bounds." );
const CoordinatesType dimensions = grid.getDimensions();
@@ -447,12 +419,10 @@ class GridEntityGetter< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 1 >
static IndexType getEntityIndex( const GridType& grid,
const GridEntity& entity )
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < grid.getDimensions() +
- CoordinatesType( 1, 1, 1 ) - entity.getBasis(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " grid.getDimensions() = " << grid.getDimensions()
- << " CoordinatesType( 1, 1, 1 ) - entity.getBasis() = " << CoordinatesType( 1, 1, 1 ) - entity.getBasis() );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(),
+ grid.getDimensions() + CoordinatesType( 1, 1, 1 ) - entity.getBasis(),
+ "wrong coordinates" );
const CoordinatesType coordinates = entity.getCoordinates();
const CoordinatesType dimensions = grid.getDimensions();
@@ -490,10 +460,8 @@ class GridEntityGetter< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 0 >
static GridEntity getEntity( const GridType& grid,
const IndexType& index )
{
- TNL_ASSERT( index >= 0 && index < grid.template getEntitiesCount< GridEntity >(),
- std::cerr << " index = " << index
- << " grid.getEntitiesCount<>() = " << grid.template getEntitiesCount< GridEntity >()
- << " entityDimension = " << entityDimension );
+ TNL_ASSERT_GE( index, 0, "Index must be non-negative." );
+ TNL_ASSERT_LT( index, grid.template getEntitiesCount< GridEntity >(), "Index is out of bounds." );
const CoordinatesType dimensions = grid.getDimensions();
@@ -512,9 +480,8 @@ class GridEntityGetter< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 0 >
static IndexType getEntityIndex( const GridType& grid,
const GridEntity& entity )
{
- TNL_ASSERT( entity.getCoordinates() >= 0 && entity.getCoordinates() <= grid.getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " grid.getDimensions() = " << grid.getDimensions() );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), grid.getDimensions(), "wrong coordinates" );
const CoordinatesType coordinates = entity.getCoordinates();
const CoordinatesType dimensions = grid.getDimensions();
@@ -527,4 +494,3 @@ class GridEntityGetter< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 0 >
} // namespace Meshes
} // namespace TNL
-
diff --git a/src/TNL/Meshes/GridDetails/GridEntityTopology.h b/src/TNL/Meshes/GridDetails/GridEntityTopology.h
index 716b2dcc2d662174dbf2006c450993adcce81bb4..9a9ba9e317c23926ce171063162686c8c57e954d 100644
--- a/src/TNL/Meshes/GridDetails/GridEntityTopology.h
+++ b/src/TNL/Meshes/GridDetails/GridEntityTopology.h
@@ -25,8 +25,8 @@ class GridEntityTopology
typedef Grid GridType;
// TODO: restore when CUDA allows it
- //static const int meshDimension = GridType::Dimension;
- enum { meshDimension = GridType::Dimension };
+ //static const int meshDimension = GridType::getMeshDimension();
+ enum { meshDimension = GridType::getMeshDimension() };
static const int entityDimension = EntityDimension;
diff --git a/src/TNL/Meshes/GridDetails/GridEntity_impl.h b/src/TNL/Meshes/GridDetails/GridEntity_impl.h
index 842ac4ad69f150adb962843db4aeb8f4b58e8441..8703c064fd6e2271ea3705ddb4355261edf14cd9 100644
--- a/src/TNL/Meshes/GridDetails/GridEntity_impl.h
+++ b/src/TNL/Meshes/GridDetails/GridEntity_impl.h
@@ -44,7 +44,7 @@ GridEntity( const Meshes::Grid< Dimension, Real, Device, Index >& grid )
coordinates( 0 ),
orientation( 0 ),
basis( 0 ),
- neighbourEntitiesStorage( *this )
+ neighborEntitiesStorage( *this )
{
}
@@ -65,7 +65,7 @@ GridEntity( const Meshes::Grid< Dimension, Real, Device, Index >& grid,
coordinates( coordinates ),
orientation( orientation ),
basis( basis ),
- neighbourEntitiesStorage( *this )
+ neighborEntitiesStorage( *this )
{
}
@@ -123,7 +123,7 @@ GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, EntityDimension, Con
refresh()
{
this->entityIndex = this->grid.getEntityIndex( *this );
- this->neighbourEntitiesStorage.refresh( this->grid, this->entityIndex );
+ this->neighborEntitiesStorage.refresh( this->grid, this->entityIndex );
}
template< int Dimension,
@@ -138,14 +138,12 @@ GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, EntityDimension, Con
getIndex() const
{
typedef Meshes::Grid< Dimension, Real, Device, Index > GridType;
- typedef typename GridType::template MeshEntity< EntityDimension > EntityType;
- TNL_ASSERT( this->entityIndex >= 0 &&
- this-> entityIndex < grid.template getEntitiesCount< EntityType >(),
- std::cerr << "this->entityIndex = " << this->entityIndex
- << " grid.template getEntitiesCount< EntityDimension >() = " << grid.template getEntitiesCount< EntityType >() );
- TNL_ASSERT( this->entityIndex == grid.getEntityIndex( *this ),
- std::cerr << "this->entityIndex = " << this->entityIndex
- << " grid.getEntityIndex( *this ) = " << grid.getEntityIndex( *this ) );
+ typedef typename GridType::template EntityType< EntityDimension > EntityType;
+ TNL_ASSERT_GE( this->entityIndex, 0, "Entity index is not non-negative." );
+ TNL_ASSERT_LT( this->entityIndex, grid.template getEntitiesCount< EntityDimension >(),
+ "Entity index is out of bounds." );
+ TNL_ASSERT_EQ( this->entityIndex, grid.getEntityIndex( *this ),
+ "Wrong value of stored index." );
return this->entityIndex;
}
@@ -213,13 +211,13 @@ template< int Dimension,
typename Index,
int EntityDimension,
typename Config >
- template< int NeighbourEntityDimension >
+ template< int NeighborEntityDimension >
__cuda_callable__ inline
-const typename GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, EntityDimension, Config >::template NeighbourEntities< NeighbourEntityDimension >&
+const typename GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, EntityDimension, Config >::template NeighborEntities< NeighborEntityDimension >&
GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, EntityDimension, Config >::
-getNeighbourEntities() const
+getNeighborEntities() const
{
- return neighbourEntitiesStorage.template getNeighbourEntities< NeighbourEntityDimension >();
+ return neighborEntitiesStorage.template getNeighborEntities< NeighborEntityDimension >();
}
template< int Dimension,
@@ -301,7 +299,7 @@ GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension, Config >:
GridEntity( const GridType& grid )
: grid( grid ),
entityIndex( -1 ),
- neighbourEntitiesStorage( *this )
+ neighborEntitiesStorage( *this )
{
this->coordinates = CoordinatesType( ( Index ) 0 );
}
@@ -320,7 +318,7 @@ GridEntity( const GridType& grid,
: grid( grid ),
entityIndex( -1 ),
coordinates( coordinates ),
- neighbourEntitiesStorage( *this )
+ neighborEntitiesStorage( *this )
{
}
@@ -374,7 +372,7 @@ GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension, Config >:
refresh()
{
this->entityIndex = this->grid.getEntityIndex( *this );
- this->neighbourEntitiesStorage.refresh( this->grid, this->entityIndex );
+ this->neighborEntitiesStorage.refresh( this->grid, this->entityIndex );
}
template< int Dimension,
@@ -387,13 +385,11 @@ Index
GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension, Config >::
getIndex() const
{
- TNL_ASSERT( this->entityIndex >= 0 &&
- this-> entityIndex < grid.template getEntitiesCount< ThisType >(),
- std::cerr << "this->entityIndex = " << this->entityIndex
- << " grid.template getEntitiesCount< Dimension >() = " << grid.template getEntitiesCount< ThisType >() );
- TNL_ASSERT( this->entityIndex == grid.getEntityIndex( *this ),
- std::cerr << "this->index = " << this->entityIndex
- << " grid.getEntityIndex( *this ) = " << grid.getEntityIndex( *this ) );
+ TNL_ASSERT_GE( this->entityIndex, 0, "Entity index is not non-negative." );
+ TNL_ASSERT_LT( this->entityIndex, grid.template getEntitiesCount< Dimension >(),
+ "Entity index is out of bounds." );
+ TNL_ASSERT_EQ( this->entityIndex, grid.getEntityIndex( *this ),
+ "Wrong value of stored index." );
return this->entityIndex;
}
@@ -428,13 +424,13 @@ template< int Dimension,
typename Device,
typename Index,
typename Config >
- template< int NeighbourEntityDimension >
+ template< int NeighborEntityDimension >
__cuda_callable__ inline
-const typename GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension, Config >::template NeighbourEntities< NeighbourEntityDimension >&
+const typename GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension, Config >::template NeighborEntities< NeighborEntityDimension >&
GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension, Config >::
-getNeighbourEntities() const
+getNeighborEntities() const
{
- return neighbourEntitiesStorage.template getNeighbourEntities< NeighbourEntityDimension >();
+ return neighborEntitiesStorage.template getNeighborEntities< NeighborEntityDimension >();
}
template< int Dimension,
@@ -518,7 +514,7 @@ GridEntity( const GridType& grid )
: grid( grid ),
entityIndex( -1 ),
coordinates( 0 ),
- neighbourEntitiesStorage( *this )
+ neighborEntitiesStorage( *this )
{
}
@@ -536,7 +532,7 @@ GridEntity( const GridType& grid,
: grid( grid ),
entityIndex( -1 ),
coordinates( coordinates ),
- neighbourEntitiesStorage( *this )
+ neighborEntitiesStorage( *this )
{
}
@@ -590,7 +586,7 @@ GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, 0, Config >::
refresh()
{
this->entityIndex = this->grid.getEntityIndex( *this );
- this->neighbourEntitiesStorage.refresh( this->grid, this->entityIndex );
+ this->neighborEntitiesStorage.refresh( this->grid, this->entityIndex );
}
template< int Dimension,
@@ -605,13 +601,11 @@ getIndex() const
{
typedef Meshes::Grid< Dimension, Real, Device, Index > GridType;
typedef typename GridType::Vertex Vertex;
- TNL_ASSERT( this->entityIndex >= 0 &&
- this-> entityIndex < grid.template getEntitiesCount< Vertex >(),
- std::cerr << "this->entityIndex = " << this->entityIndex
- << " grid.template getEntitiesCount< 0 >() = " << grid.template getEntitiesCount< Vertex >() );
- TNL_ASSERT( this->entityIndex == grid.getEntityIndex( *this ),
- std::cerr << "this->entityIndex = " << this->entityIndex
- << " grid.getEntityIndex( *this ) = " << grid.getEntityIndex( *this ) );
+ TNL_ASSERT_GE( this->entityIndex, 0, "Entity index is not non-negative." );
+ TNL_ASSERT_LT( this->entityIndex, grid.template getEntitiesCount< 0 >(),
+ "Entity index is out of bounds." );
+ TNL_ASSERT_EQ( this->entityIndex, grid.getEntityIndex( *this ),
+ "Wrong value of stored index." );
return this->entityIndex;
}
@@ -646,13 +640,13 @@ template< int Dimension,
typename Device,
typename Index,
typename Config >
- template< int NeighbourEntityDimension >
+ template< int NeighborEntityDimension >
__cuda_callable__ inline
-const typename GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, 0, Config >::template NeighbourEntities< NeighbourEntityDimension >&
+const typename GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, 0, Config >::template NeighborEntities< NeighborEntityDimension >&
GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, 0, Config >::
-getNeighbourEntities() const
+getNeighborEntities() const
{
- return neighbourEntitiesStorage.template getNeighbourEntities< NeighbourEntityDimension >();
+ return neighborEntitiesStorage.template getNeighborEntities< NeighborEntityDimension >();
}
template< int Dimension,
diff --git a/src/TNL/Meshes/GridDetails/GridTraverser.h b/src/TNL/Meshes/GridDetails/GridTraverser.h
index 8365f41d7944e04291f13a849afb57832cbc2274..7e821d714bcdb3b6081ac59ae160ab898434979b 100644
--- a/src/TNL/Meshes/GridDetails/GridTraverser.h
+++ b/src/TNL/Meshes/GridDetails/GridTraverser.h
@@ -85,6 +85,38 @@ class GridTraverser< Meshes::Grid< 1, Real, Devices::Cuda, Index > >
const int& stream = 0 );
};
+/****
+ * 1D grid, Devices::MIC
+ */
+template< typename Real,
+ typename Index >
+class GridTraverser< Meshes::Grid< 1, Real, Devices::MIC, Index > >
+{
+ public:
+
+ typedef Meshes::Grid< 1, Real, Devices::MIC, Index > GridType;
+ typedef SharedPointer< GridType > GridPointer;
+ typedef Real RealType;
+ typedef Devices::MIC DeviceType;
+ typedef Index IndexType;
+ typedef typename GridType::CoordinatesType CoordinatesType;
+
+ template<
+ typename GridEntity,
+ typename EntitiesProcessor,
+ typename UserData,
+ bool processOnlyBoundaryEntities >
+ static void
+ processEntities(
+ const GridPointer& gridPointer,
+ const CoordinatesType& begin,
+ const CoordinatesType& end,
+ SharedPointer< UserData, DeviceType >& userData,
+ const int& stream = 0 );
+};
+
+
+
/****
* 2D grid, Devices::Host
*/
@@ -161,6 +193,44 @@ class GridTraverser< Meshes::Grid< 2, Real, Devices::Cuda, Index > >
const GridEntityParameters&... gridEntityParameters );
};
+/****
+ * 2D grid, Devices::MIC
+ */
+template< typename Real,
+ typename Index >
+class GridTraverser< Meshes::Grid< 2, Real, Devices::MIC, Index > >
+{
+ public:
+
+ typedef Meshes::Grid< 2, Real, Devices::MIC, Index > GridType;
+ typedef SharedPointer< GridType > GridPointer;
+ typedef Real RealType;
+ typedef Devices::MIC DeviceType;
+ typedef Index IndexType;
+ typedef typename GridType::CoordinatesType CoordinatesType;
+
+ template<
+ typename GridEntity,
+ typename EntitiesProcessor,
+ typename UserData,
+ bool processOnlyBoundaryEntities,
+ int XOrthogonalBoundary = 1,
+ int YOrthogonalBoundary = 1,
+ typename... GridEntityParameters >
+ static void
+ processEntities(
+ const GridPointer& gridPointer,
+ const CoordinatesType& begin,
+ const CoordinatesType& end,
+ SharedPointer< UserData, DeviceType >& userData,
+ // FIXME: hack around nvcc bug (error: default argument not at end of parameter list)
+// const int& stream = 0,
+ const int& stream,
+ // gridEntityParameters are passed to GridEntity's constructor
+ // (i.e. orientation and basis for faces)
+ const GridEntityParameters&... gridEntityParameters );
+};
+
/****
* 3D grid, Devices::Host
*/
@@ -239,6 +309,45 @@ class GridTraverser< Meshes::Grid< 3, Real, Devices::Cuda, Index > >
const GridEntityParameters&... gridEntityParameters );
};
+/****
+ * 3D grid, Devices::Cuda
+ */
+template< typename Real,
+ typename Index >
+class GridTraverser< Meshes::Grid< 3, Real, Devices::MIC, Index > >
+{
+ public:
+
+ typedef Meshes::Grid< 3, Real, Devices::MIC, Index > GridType;
+ typedef SharedPointer< GridType > GridPointer;
+ typedef Real RealType;
+ typedef Devices::MIC DeviceType;
+ typedef Index IndexType;
+ typedef typename GridType::CoordinatesType CoordinatesType;
+
+ template<
+ typename GridEntity,
+ typename EntitiesProcessor,
+ typename UserData,
+ bool processOnlyBoundaryEntities,
+ int XOrthogonalBoundary = 1,
+ int YOrthogonalBoundary = 1,
+ int ZOrthogonalBoundary = 1,
+ typename... GridEntityParameters >
+ static void
+ processEntities(
+ const GridPointer& gridPointer,
+ const CoordinatesType& begin,
+ const CoordinatesType& end,
+ SharedPointer< UserData, DeviceType >& userData,
+ // FIXME: hack around nvcc bug (error: default argument not at end of parameter list)
+// const int& stream = 0,
+ const int& stream,
+ // gridEntityParameters are passed to GridEntity's constructor
+ // (i.e. orientation and basis for faces and edges)
+ const GridEntityParameters&... gridEntityParameters );
+};
+
} // namespace Meshes
} // namespace TNL
diff --git a/src/TNL/Meshes/GridDetails/GridTraverser_impl.h b/src/TNL/Meshes/GridDetails/GridTraverser_impl.h
index 2adfb1b4cd34988f60760ff77635b51097226194..4443e161c3d3296f7a7924d927f00a3581248cec 100644
--- a/src/TNL/Meshes/GridDetails/GridTraverser_impl.h
+++ b/src/TNL/Meshes/GridDetails/GridTraverser_impl.h
@@ -8,8 +8,14 @@
/* See Copyright Notice in tnl/Copyright */
+#include <TNL/Devices/MIC.h>
+
#pragma once
+#include "GridTraverser.h"
+
+#include <TNL/Exceptions/CudaSupportMissing.h>
+
namespace TNL {
namespace Meshes {
@@ -193,11 +199,75 @@ processEntities(
if( stream == 0 )
{
cudaStreamSynchronize( s );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
+#else
+ throw Exceptions::CudaSupportMissing();
#endif
}
+/****
+ * 1D traverser, MIC
+ */
+
+template< typename Real,
+ typename Index >
+ template<
+ typename GridEntity,
+ typename EntitiesProcessor,
+ typename UserData,
+ bool processOnlyBoundaryEntities >
+void
+GridTraverser< Meshes::Grid< 1, Real, Devices::MIC, Index > >::
+processEntities(
+ const GridPointer& gridPointer,
+ const CoordinatesType& begin,
+ const CoordinatesType& end,
+ SharedPointer< UserData, DeviceType >& userDataPointer,
+ const int& stream )
+{
+ std::cout << "Not Implemented yet Grid Traverser <1, Real, Device::MIC>" << std::endl;
+/*
+ auto& pool = CudaStreamPool::getInstance();
+ const cudaStream_t& s = pool.getStream( stream );
+
+ Devices::Cuda::synchronizeDevice();
+ if( processOnlyBoundaryEntities )
+ {
+ dim3 cudaBlockSize( 2 );
+ dim3 cudaBlocks( 1 );
+ GridBoundaryTraverser1D< Real, Index, GridEntity, UserData, EntitiesProcessor >
+ <<< cudaBlocks, cudaBlockSize, 0, s >>>
+ ( &gridPointer.template getData< Devices::Cuda >(),
+ &userDataPointer.template modifyData< Devices::Cuda >(),
+ begin,
+ end );
+ }
+ else
+ {
+ dim3 cudaBlockSize( 256 );
+ dim3 cudaBlocks;
+ cudaBlocks.x = Devices::Cuda::getNumberOfBlocks( end.x() - begin.x() + 1, cudaBlockSize.x );
+ const IndexType cudaXGrids = Devices::Cuda::getNumberOfGrids( cudaBlocks.x );
+
+ for( IndexType gridXIdx = 0; gridXIdx < cudaXGrids; gridXIdx ++ )
+ GridTraverser1D< Real, Index, GridEntity, UserData, EntitiesProcessor >
+ <<< cudaBlocks, cudaBlockSize, 0, s >>>
+ ( &gridPointer.template getData< Devices::Cuda >(),
+ &userDataPointer.template modifyData< Devices::Cuda >(),
+ begin,
+ end,
+ gridXIdx );
+ }
+
+ // only launches into the stream 0 are synchronized
+ if( stream == 0 )
+ {
+ cudaStreamSynchronize( s );
+ checkCudaDevice;
+ }
+*/
+}
/****
* 2D traverser, host
@@ -499,7 +569,7 @@ processEntities(
cudaStreamSynchronize( s2 );
cudaStreamSynchronize( s3 );
cudaStreamSynchronize( s4 );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
else
{
@@ -533,9 +603,109 @@ processEntities(
if( stream == 0 )
{
cudaStreamSynchronize( s );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
}
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
+}
+
+/****
+ * 2D traverser, MIC
+ */
+template< typename Real,
+ typename Index >
+ template<
+ typename GridEntity,
+ typename EntitiesProcessor,
+ typename UserData,
+ bool processOnlyBoundaryEntities,
+ int XOrthogonalBoundary,
+ int YOrthogonalBoundary,
+ typename... GridEntityParameters >
+void
+GridTraverser< Meshes::Grid< 2, Real, Devices::MIC, Index > >::
+processEntities(
+ const GridPointer& gridPointer,
+ const CoordinatesType& begin,
+ const CoordinatesType& end,
+ SharedPointer< UserData, DeviceType >& userDataPointer,
+ const int& stream,
+ const GridEntityParameters&... gridEntityParameters )
+{
+
+
+#ifdef HAVE_MIC
+ Devices::MIC::synchronizeDevice();
+
+ //TOHLE JE PRUSER -- nemim poslat vypustku --
+ //GridEntity entity( gridPointer.template getData< Devices::MIC >(), begin, gridEntityParameters... );
+
+
+ Devices::MICHider<const GridType> hMicGrid;
+ hMicGrid.pointer=& gridPointer.template getData< Devices::MIC >();
+ Devices::MICHider<UserData> hMicUserData;
+ hMicUserData.pointer=& userDataPointer.template modifyData<Devices::MIC>();
+ TNLMICSTRUCT(begin, const CoordinatesType);
+ TNLMICSTRUCT(end, const CoordinatesType);
+
+ #pragma offload target(mic) in(sbegin,send,hMicUserData,hMicGrid)
+ {
+
+ #pragma omp parallel firstprivate( sbegin, send )
+ {
+ TNLMICSTRUCTUSE(begin, const CoordinatesType);
+ TNLMICSTRUCTUSE(end, const CoordinatesType);
+ GridEntity entity( *(hMicGrid.pointer), *(kernelbegin) );
+
+ if( processOnlyBoundaryEntities )
+ {
+ if( YOrthogonalBoundary )
+ #pragma omp for
+ for( auto k = kernelbegin->x();
+ k <= kernelend->x();
+ k ++ )
+ {
+ entity.getCoordinates().x() = k;
+ entity.getCoordinates().y() = kernelbegin->y();
+ entity.refresh();
+ EntitiesProcessor::processEntity( entity.getMesh(), *(hMicUserData.pointer), entity );
+ entity.getCoordinates().y() = kernelend->y();
+ entity.refresh();
+ EntitiesProcessor::processEntity( entity.getMesh(), *(hMicUserData.pointer), entity );
+ }
+ if( XOrthogonalBoundary )
+ #pragma omp for
+ for( auto k = kernelbegin->y();
+ k <= kernelend->y();
+ k ++ )
+ {
+ entity.getCoordinates().y() = k;
+ entity.getCoordinates().x() = kernelbegin->x();
+ entity.refresh();
+ EntitiesProcessor::processEntity( entity.getMesh(), *(hMicUserData.pointer), entity );
+ entity.getCoordinates().x() = kernelend->x();
+ entity.refresh();
+ EntitiesProcessor::processEntity( entity.getMesh(), *(hMicUserData.pointer), entity );
+ }
+ }
+ else
+ {
+ #pragma omp for
+ for( IndexType y = kernelbegin->y(); y <= kernelend->y(); y ++ )
+ for( IndexType x = kernelbegin->x(); x <= kernelend->x(); x ++ )
+ {
+ // std::cerr << x << " " <<y << std::endl;
+ entity.getCoordinates().x() = x;
+ entity.getCoordinates().y() = y;
+ entity.refresh();
+ EntitiesProcessor::processEntity( entity.getMesh(), *(hMicUserData.pointer), entity );
+ }
+ }
+ }
+ }
+
#endif
}
@@ -940,7 +1110,7 @@ processEntities(
cudaStreamSynchronize( s4 );
cudaStreamSynchronize( s5 );
cudaStreamSynchronize( s6 );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
else
{
@@ -976,11 +1146,76 @@ processEntities(
if( stream == 0 )
{
cudaStreamSynchronize( s );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
}
+#else
+ throw Exceptions::CudaSupportMissing();
#endif
}
+/****
+ * 3D traverser, MIC
+ */
+template< typename Real,
+ typename Index >
+ template<
+ typename GridEntity,
+ typename EntitiesProcessor,
+ typename UserData,
+ bool processOnlyBoundaryEntities,
+ int XOrthogonalBoundary,
+ int YOrthogonalBoundary,
+ int ZOrthogonalBoundary,
+ typename... GridEntityParameters >
+void
+GridTraverser< Meshes::Grid< 3, Real, Devices::MIC, Index > >::
+processEntities(
+ const GridPointer& gridPointer,
+ const CoordinatesType& begin,
+ const CoordinatesType& end,
+ SharedPointer< UserData, DeviceType >& userDataPointer,
+ const int& stream,
+ const GridEntityParameters&... gridEntityParameters )
+{
+ std::cout << "Not Implemented yet Grid Traverser <3, Real, Device::MIC>" << std::endl;
+
+/* HAVE_CUDA
+ dim3 cudaBlockSize( 8, 8, 8 );
+ dim3 cudaBlocks;
+ cudaBlocks.x = Devices::Cuda::getNumberOfBlocks( end.x() - begin.x() + 1, cudaBlockSize.x );
+ cudaBlocks.y = Devices::Cuda::getNumberOfBlocks( end.y() - begin.y() + 1, cudaBlockSize.y );
+ cudaBlocks.z = Devices::Cuda::getNumberOfBlocks( end.z() - begin.z() + 1, cudaBlockSize.z );
+ const IndexType cudaXGrids = Devices::Cuda::getNumberOfGrids( cudaBlocks.x );
+ const IndexType cudaYGrids = Devices::Cuda::getNumberOfGrids( cudaBlocks.y );
+ const IndexType cudaZGrids = Devices::Cuda::getNumberOfGrids( cudaBlocks.z );
+
+ auto& pool = CudaStreamPool::getInstance();
+ const cudaStream_t& s = pool.getStream( stream );
+
+ Devices::Cuda::synchronizeDevice();
+ for( IndexType gridZIdx = 0; gridZIdx < cudaZGrids; gridZIdx ++ )
+ for( IndexType gridYIdx = 0; gridYIdx < cudaYGrids; gridYIdx ++ )
+ for( IndexType gridXIdx = 0; gridXIdx < cudaXGrids; gridXIdx ++ )
+ GridTraverser3D< Real, Index, GridEntity, UserData, EntitiesProcessor, processOnlyBoundaryEntities, GridEntityParameters... >
+ <<< cudaBlocks, cudaBlockSize, 0, s >>>
+ ( &gridPointer.template getData< Devices::Cuda >(),
+ &userDataPointer.template modifyData< Devices::Cuda >(),
+ begin,
+ end,
+ gridXIdx,
+ gridYIdx,
+ gridZIdx,
+ gridEntityParameters... );
+
+ // only launches into the stream 0 are synchronized
+ if( stream == 0 )
+ {
+ cudaStreamSynchronize( s );
+ checkCudaDevice;
+ }
+ */
+}
+
} // namespace Meshes
} // namespace TNL
diff --git a/src/TNL/Meshes/GridDetails/NeighborGridEntitiesStorage.h b/src/TNL/Meshes/GridDetails/NeighborGridEntitiesStorage.h
new file mode 100644
index 0000000000000000000000000000000000000000..c8f3a999b90a08e02b9bcb2cfb29b0270476f046
--- /dev/null
+++ b/src/TNL/Meshes/GridDetails/NeighborGridEntitiesStorage.h
@@ -0,0 +1,173 @@
+/***************************************************************************
+ NeighborGridEntitiesStorage.h - description
+ -------------------
+ begin : Dec 18, 2015
+ copyright : (C) 2015 by Tomas Oberhuber
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#pragma once
+
+#include <TNL/Devices/Cuda.h>
+#include <TNL/Meshes/MeshDimensionTag.h>
+#include <TNL/Meshes/GridEntityConfig.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter.h>
+
+namespace TNL {
+namespace Meshes {
+
+template< typename GridEntity,
+ int NeighborEntityDimension,
+ typename GridEntityConfig,
+ bool storage = GridEntityConfig::template neighborEntityStorage< GridEntity >( NeighborEntityDimension ) >
+class NeighborGridEntityLayer{};
+
+template< typename GridEntity,
+ int NeighborEntityDimension,
+ typename GridEntityConfig >
+class NeighborGridEntityLayer< GridEntity, NeighborEntityDimension, GridEntityConfig, true >
+: public NeighborGridEntityLayer< GridEntity, NeighborEntityDimension - 1, GridEntityConfig >
+{
+ public:
+
+ typedef NeighborGridEntityLayer< GridEntity, NeighborEntityDimension - 1, GridEntityConfig > BaseType;
+ typedef NeighborGridEntityGetter< GridEntity, NeighborEntityDimension > NeighborEntityGetterType;
+
+ using BaseType::getNeighborEntities;
+
+ __cuda_callable__
+ NeighborGridEntityLayer( const GridEntity& entity )
+ : BaseType( entity ),
+ neighborEntities( entity )
+ {}
+
+ __cuda_callable__
+ const NeighborEntityGetterType& getNeighborEntities( const MeshDimensionTag< NeighborEntityDimension>& tag ) const
+ {
+ return this->neighborEntities;
+ }
+
+ __cuda_callable__
+ void refresh( const typename GridEntity::GridType& grid,
+ const typename GridEntity::GridType::IndexType& entityIndex )
+ {
+ BaseType::refresh( grid, entityIndex );
+ neighborEntities.refresh( grid, entityIndex );
+ }
+
+ protected:
+
+ NeighborEntityGetterType neighborEntities;
+};
+
+template< typename GridEntity,
+ typename GridEntityConfig >
+class NeighborGridEntityLayer< GridEntity, 0, GridEntityConfig, true >
+{
+ public:
+
+ typedef NeighborGridEntityGetter< GridEntity, 0 > NeighborEntityGetterType;
+
+ __cuda_callable__
+ NeighborGridEntityLayer( const GridEntity& entity )
+ : neighborEntities( entity )
+ {}
+
+ __cuda_callable__
+ const NeighborEntityGetterType& getNeighborEntities( const MeshDimensionTag< 0 >& tag ) const
+ {
+ return this->neighborEntities;
+ }
+
+ __cuda_callable__
+ void refresh( const typename GridEntity::GridType& grid,
+ const typename GridEntity::GridType::IndexType& entityIndex )
+ {
+ neighborEntities.refresh( grid, entityIndex );
+ }
+
+ protected:
+
+ NeighborEntityGetterType neighborEntities;
+};
+
+template< typename GridEntity,
+ int NeighborEntityDimension,
+ typename GridEntityConfig >
+class NeighborGridEntityLayer< GridEntity, NeighborEntityDimension, GridEntityConfig, false >
+: public NeighborGridEntityLayer< GridEntity, NeighborEntityDimension - 1, GridEntityConfig >
+{
+ public:
+
+ typedef NeighborGridEntityLayer< GridEntity, NeighborEntityDimension - 1, GridEntityConfig > BaseType;
+ typedef NeighborGridEntityGetter< GridEntity, NeighborEntityDimension > NeighborEntityGetterType;
+
+ using BaseType::getNeighborEntities;
+
+ __cuda_callable__
+ NeighborGridEntityLayer( const GridEntity& entity )
+ : BaseType( entity )
+ {}
+
+ __cuda_callable__
+ const NeighborEntityGetterType& getNeighborEntities( const MeshDimensionTag< NeighborEntityDimension >& tag ) const {}
+
+ __cuda_callable__
+ void refresh( const typename GridEntity::GridType& grid,
+ const typename GridEntity::GridType::IndexType& entityIndex ) {}
+};
+
+template< typename GridEntity,
+ typename GridEntityConfig >
+class NeighborGridEntityLayer< GridEntity, 0, GridEntityConfig, false >
+{
+ public:
+
+ typedef NeighborGridEntityGetter< GridEntity, 0 > NeighborEntityGetterType;
+
+ __cuda_callable__
+ NeighborGridEntityLayer( const GridEntity& entity ){}
+
+ __cuda_callable__
+ const NeighborEntityGetterType& getNeighborEntities( const MeshDimensionTag< 0 >& tag ) const {}
+
+ __cuda_callable__
+ void refresh( const typename GridEntity::GridType& grid,
+ const typename GridEntity::GridType::IndexType& entityIndex ) {}
+};
+
+
+
+
+template< typename GridEntity,
+ typename GridEntityConfig >
+class NeighborGridEntitiesStorage
+: public NeighborGridEntityLayer< GridEntity, GridEntity::getMeshDimension(), GridEntityConfig >
+{
+ typedef NeighborGridEntityLayer< GridEntity, GridEntity::getMeshDimension(), GridEntityConfig > BaseType;
+
+ public:
+
+ using BaseType::getNeighborEntities;
+ using BaseType::refresh;
+
+ __cuda_callable__
+ NeighborGridEntitiesStorage( const GridEntity& entity )
+ : BaseType( entity )
+ {}
+
+ template< int EntityDimension >
+ __cuda_callable__
+ const NeighborGridEntityGetter< GridEntity, EntityDimension >&
+ getNeighborEntities() const
+ {
+ return BaseType::getNeighborEntities( MeshDimensionTag< EntityDimension >() );
+ }
+};
+
+
+} // namespace Meshes
+} // namespace TNL
+
diff --git a/src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h b/src/TNL/Meshes/GridDetails/NeighborGridEntityGetter.h
similarity index 77%
rename from src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h
rename to src/TNL/Meshes/GridDetails/NeighborGridEntityGetter.h
index ec6cd3f297530a50c5f6f853eccde3efa65af666..84a9c56d9389f31c013206c10f63fceb81ec2e0c 100644
--- a/src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h
+++ b/src/TNL/Meshes/GridDetails/NeighborGridEntityGetter.h
@@ -1,5 +1,5 @@
/***************************************************************************
- NeighbourGridEntityGetter.h - description
+ NeighborGridEntityGetter.h - description
-------------------
begin : Nov 23, 2015
copyright : (C) 2015 by Tomas Oberhuber
@@ -11,23 +11,24 @@
#pragma once
#include <TNL/Assert.h>
+#include <TNL/Devices/Cuda.h>
#include <TNL/Meshes/GridEntityConfig.h>
namespace TNL {
namespace Meshes {
template< typename GridEntity,
- int NeighbourEntityDimension,
+ int NeighborEntityDimension,
typename EntityStencilTag =
- GridEntityStencilStorageTag< GridEntity::ConfigType::template neighbourEntityStorage< GridEntity >( NeighbourEntityDimension ) > >
-class NeighbourGridEntityGetter
+ GridEntityStencilStorageTag< GridEntity::ConfigType::template neighborEntityStorage< GridEntity >( NeighborEntityDimension ) > >
+class NeighborGridEntityGetter
{
public:
// TODO: not all specializations are implemented yet
__cuda_callable__
- NeighbourGridEntityGetter( const GridEntity& entity )
+ NeighborGridEntityGetter( const GridEntity& entity )
{
//TNL_ASSERT( false, );
}
diff --git a/src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter1D_impl.h b/src/TNL/Meshes/GridDetails/NeighborGridEntityGetter1D_impl.h
similarity index 63%
rename from src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter1D_impl.h
rename to src/TNL/Meshes/GridDetails/NeighborGridEntityGetter1D_impl.h
index a8e91bcefc73ef9e4b5eaaf41bf1bf4f8e6f0595..14143220eb451521afb6098e67ebc9889559a68a 100644
--- a/src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter1D_impl.h
+++ b/src/TNL/Meshes/GridDetails/NeighborGridEntityGetter1D_impl.h
@@ -1,5 +1,5 @@
/***************************************************************************
- NeighbourGridEntityGetter1D_impl.h - description
+ NeighborGridEntityGetter1D_impl.h - description
-------------------
begin : Nov 23, 2015
copyright : (C) 2015 by Tomas Oberhuber
@@ -10,7 +10,7 @@
#pragma once
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter.h>
#include <TNL/Meshes/GridDetails/Grid1D.h>
#include <TNL/Meshes/GridDetails/Grid2D.h>
#include <TNL/Meshes/GridDetails/Grid3D.h>
@@ -21,7 +21,7 @@ namespace Meshes {
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 1 | 1 | ---- |
* +-----------------+---------------------------+-------------------+
@@ -30,7 +30,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 1, Real, Device, Index >, 1, Config >,
1,
GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -38,46 +38,40 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 1;
- static const int NeighbourEntityDimension = 1;
+ static const int NeighborEntityDimension = 1;
typedef Meshes::Grid< 1, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int step >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( this->entity.getCoordinates() >= CoordinatesType( 0 ) &&
- this->entity.getCoordinates() < this->entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << this->entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( step ) >= CoordinatesType( 0 ) &&
entity.getCoordinates() + CoordinatesType( step ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntity( CoordinatesType( entity.getCoordinates().x() + step ) );
+ return NeighborGridEntity( CoordinatesType( entity.getCoordinates().x() + step ) );
}
template< int step >
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( step ) >= CoordinatesType( 0 ) &&
entity.getCoordinates() + CoordinatesType( step ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
@@ -97,7 +91,7 @@ class NeighbourGridEntityGetter<
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 1 | 1 | Cross/Full |
* +-----------------+---------------------------+-------------------+
@@ -107,7 +101,7 @@ template< typename Real,
typename Index,
typename Config,
typename StencilStorage >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 1, Real, Device, Index >, 1, Config >,
1,
StencilStorage >
@@ -115,49 +109,43 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 1;
- static const int NeighbourEntityDimension = 1;
+ static const int NeighborEntityDimension = 1;
typedef Meshes::Grid< 1, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
- typedef NeighbourGridEntityGetter< GridEntityType, 1, StencilStorage > ThisType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
+ typedef NeighborGridEntityGetter< GridEntityType, 1, StencilStorage > ThisType;
static const int stencilSize = Config::getStencilSize();
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int step >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( this->entity.getCoordinates() >= CoordinatesType( 0 ) &&
- this->entity.getCoordinates() < this->entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << this->entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( step ) >= CoordinatesType( 0 ) &&
entity.getCoordinates() + CoordinatesType( step ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->entity.getMesh(), CoordinatesType( entity.getCoordinates().x() + step ) );
+ return NeighborGridEntityType( this->entity.getMesh(), CoordinatesType( entity.getCoordinates().x() + step ) );
}
template< int step >
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( step ) >= CoordinatesType( 0 ) &&
entity.getCoordinates() + CoordinatesType( step ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
@@ -179,9 +167,9 @@ class NeighbourGridEntityGetter<
public:
__cuda_callable__
- static void exec( ThisType& neighbourEntityGetter, const IndexType& entityIndex )
+ static void exec( ThisType& neighborEntityGetter, const IndexType& entityIndex )
{
- neighbourEntityGetter.stencil[ index + stencilSize ] = entityIndex + index;
+ neighborEntityGetter.stencil[ index + stencilSize ] = entityIndex + index;
}
};
@@ -202,7 +190,7 @@ class NeighbourGridEntityGetter<
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 1 | 0 | None |
* +-----------------+---------------------------+-------------------+
@@ -211,7 +199,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 1, Real, Device, Index >, 1, Config >,
0,
GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -219,46 +207,40 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 1;
- static const int NeighbourEntityDimension = 0;
+ static const int NeighborEntityDimension = 0;
typedef Meshes::Grid< 1, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int step >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates().x() + step + ( step < 0 ) >= CoordinatesType( 0 ) &&
entity.getCoordinates().x() + step + ( step < 0 ) <= entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntity( CoordinatesType( entity.getCoordinates().x() + step + ( step < 0 ) ) );
+ return NeighborGridEntity( CoordinatesType( entity.getCoordinates().x() + step + ( step < 0 ) ) );
}
template< int step >
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates().x() + step + ( step < 0 ) >= CoordinatesType( 0 ).x() &&
entity.getCoordinates().x() + step + ( step < 0 ) <= entity.getMesh().getDimensions().x(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
@@ -274,13 +256,13 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 0 | 1 | None |
* +-----------------+---------------------------+-------------------+
@@ -290,7 +272,7 @@ template< typename Real,
typename Index,
typename Config,
typename StencilStorage >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 1, Real, Device, Index >, 0, Config >,
1,
StencilStorage > //GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -298,17 +280,17 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 0;
- static const int NeighbourEntityDimension = 1;
+ static const int NeighborEntityDimension = 1;
typedef Meshes::Grid< 1, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
@@ -316,30 +298,24 @@ class NeighbourGridEntityGetter<
template< int step >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates().x() + step - ( step > 0 ) >= CoordinatesType( 0 ) &&
entity.getCoordinates().x() + step - ( step > 0 ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntity( CoordinatesType( entity.getCoordinates().x() + step - ( step > 0 ) ) );
+ return NeighborGridEntity( CoordinatesType( entity.getCoordinates().x() + step - ( step > 0 ) ) );
}
template< int step >
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates().x() + step - ( step > 0 ) >= 0 &&
entity.getCoordinates().x() + step - ( step > 0 ) < entity.getMesh().getDimensions().x(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
@@ -358,7 +334,7 @@ class NeighbourGridEntityGetter<
/**** TODO: Implement this, now it is only a copy of specialization for none stencil storage
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 0 | 1 | Cross |
* +-----------------+---------------------------+-------------------+
@@ -367,7 +343,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 1, Real, Device, Index >, 0, Config >,
1,
GridEntityStencilStorageTag< GridEntityCrossStencil > >
@@ -375,47 +351,41 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 0;
- static const int NeighbourEntityDimension = 1;
+ static const int NeighborEntityDimension = 1;
typedef Meshes::Grid< 1, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int step >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates().x() + step - ( step > 0 ) >= CoordinatesType( 0 ) &&
entity.getCoordinates().x() + step - ( step > 0 ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntity( CoordinatesType( entity.getCoordinates().x() + step - ( step > 0 ) ) );
+ return NeighborGridEntity( CoordinatesType( entity.getCoordinates().x() + step - ( step > 0 ) ) );
}
template< int step >
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates().x() + step - ( step > 0 ) >= CoordinatesType( 0 ) &&
entity.getCoordinates().x() + step - ( step > 0 ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
@@ -435,7 +405,7 @@ class NeighbourGridEntityGetter<
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 0 | 0 | None |
* +-----------------+---------------------------+-------------------+
@@ -444,7 +414,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 1, Real, Device, Index >, 0, Config >,
0,
GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -452,46 +422,40 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 0;
- static const int NeighbourEntityDimension = 0;
+ static const int NeighborEntityDimension = 0;
typedef Meshes::Grid< 1, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int step >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates().x() + step >= CoordinatesType( 0 ) &&
entity.getCoordinates().x() + step <= entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntity( CoordinatesType( entity.getCoordinates().x() + step ) );
+ return NeighborGridEntity( CoordinatesType( entity.getCoordinates().x() + step ) );
}
template< int step >
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates().x() + step >= CoordinatesType( 0 ) &&
entity.getCoordinates().x() + step <= entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
@@ -511,5 +475,4 @@ class NeighbourGridEntityGetter<
};
} // namespace Meshes
-} // namespace TNL
-
+} // namespace TNL
\ No newline at end of file
diff --git a/src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter2D_impl.h b/src/TNL/Meshes/GridDetails/NeighborGridEntityGetter2D_impl.h
similarity index 70%
rename from src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter2D_impl.h
rename to src/TNL/Meshes/GridDetails/NeighborGridEntityGetter2D_impl.h
index 2ccb4568fa73c30356d306f1a0bf98075481d3aa..b760748cd23a84552891d095f9053058d4b012ca 100644
--- a/src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter2D_impl.h
+++ b/src/TNL/Meshes/GridDetails/NeighborGridEntityGetter2D_impl.h
@@ -1,5 +1,5 @@
/***************************************************************************
- NeighbourGridEntityGetter2D_impl.h - description
+ NeighborGridEntityGetter2D_impl.h - description
-------------------
begin : Nov 23, 2015
copyright : (C) 2015 by Tomas Oberhuber
@@ -10,7 +10,7 @@
#pragma once
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter.h>
#include <TNL/Meshes/GridDetails/Grid1D.h>
#include <TNL/Meshes/GridDetails/Grid2D.h>
#include <TNL/Meshes/GridDetails/Grid3D.h>
@@ -20,7 +20,7 @@ namespace Meshes {
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stencil Storage |
+ * | EntityDimenions | NeighborEntityDimension | Stencil Storage |
* +-----------------+---------------------------+-------------------+
* | 2 | 2 | No specialization |
* +-----------------+---------------------------+-------------------+
@@ -30,7 +30,7 @@ template< typename Real,
typename Index,
typename Config,
typename StencilStorage >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 2, Real, Device, Index >, 2, Config >,
2,
StencilStorage >
@@ -38,35 +38,32 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 2;
- static const int NeighbourEntityDimension = 2;
+ static const int NeighborEntityDimension = 2;
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY ) >= CoordinatesType( 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY ) = " << entity.getCoordinates() + CoordinatesType( stepX, stepY )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->grid,
+ return NeighborGridEntityType( this->grid,
CoordinatesType( entity.getCoordinates().x() + stepX,
entity.getCoordinates().y() + stepY ) );
}
@@ -75,11 +72,8 @@ class NeighbourGridEntityGetter<
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY ) >= CoordinatesType( 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY ) = " << entity.getCoordinates() + CoordinatesType( stepX, stepY )
@@ -95,12 +89,12 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stencil Storage |
+ * | EntityDimenions | NeighborEntityDimension | Stencil Storage |
* +-----------------+---------------------------+-------------------+
* | 2 | 2 | Cross |
* +-----------------+---------------------------+-------------------+
@@ -109,7 +103,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 2, Real, Device, Index >, 2, Config >,
2,
GridEntityStencilStorageTag< GridEntityCrossStencil > >
@@ -117,40 +111,37 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 2;
- static const int NeighbourEntityDimension = 2;
+ static const int NeighborEntityDimension = 2;
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
typedef GridEntityStencilStorageTag< GridEntityCrossStencil > StencilStorage;
- typedef NeighbourGridEntityGetter< GridEntityType, 2, StencilStorage > ThisType;
+ typedef NeighborGridEntityGetter< GridEntityType, 2, StencilStorage > ThisType;
static const int stencilSize = Config::getStencilSize();
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY ) >= CoordinatesType( 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY ) = " << entity.getCoordinates() + CoordinatesType( stepX, stepY )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->entity.getMesh(),
+ return NeighborGridEntityType( this->entity.getMesh(),
CoordinatesType( entity.getCoordinates().x() + stepX,
entity.getCoordinates().y() + stepY ) );
}
@@ -159,11 +150,8 @@ class NeighbourGridEntityGetter<
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY ) >= CoordinatesType( 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY ) = " << entity.getCoordinates() + CoordinatesType( stepX, stepY )
@@ -189,9 +177,9 @@ class NeighbourGridEntityGetter<
public:
__cuda_callable__
- static void exec( ThisType& neighbourEntityGetter, const IndexType& entityIndex )
+ static void exec( ThisType& neighborEntityGetter, const IndexType& entityIndex )
{
- neighbourEntityGetter.stencilX[ index + stencilSize ] = entityIndex + index;
+ neighborEntityGetter.stencilX[ index + stencilSize ] = entityIndex + index;
}
};
@@ -201,10 +189,10 @@ class NeighbourGridEntityGetter<
public:
__cuda_callable__
- static void exec( ThisType& neighbourEntityGetter, const IndexType& entityIndex )
+ static void exec( ThisType& neighborEntityGetter, const IndexType& entityIndex )
{
- neighbourEntityGetter.stencilY[ index + stencilSize ] =
- entityIndex + index * neighbourEntityGetter.entity.getMesh().getDimensions().x();
+ neighborEntityGetter.stencilY[ index + stencilSize ] =
+ entityIndex + index * neighborEntityGetter.entity.getMesh().getDimensions().x();
}
};
@@ -226,12 +214,12 @@ class NeighbourGridEntityGetter<
IndexType stencilX[ 2 * stencilSize + 1 ];
IndexType stencilY[ 2 * stencilSize + 1 ];
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stencil Storage |
+ * | EntityDimenions | NeighborEntityDimension | Stencil Storage |
* +-----------------+---------------------------+-------------------+
* | 2 | 1 | None |
* +-----------------+---------------------------+-------------------+
@@ -241,7 +229,7 @@ template< typename Real,
typename Index,
typename Config,
typename StencilStorage >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 2, Real, Device, Index >, 2, Config >,
1,
StencilStorage >
@@ -249,33 +237,29 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 2;
- static const int NeighbourEntityDimension = 1;
+ static const int NeighborEntityDimension = 1;
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
typedef typename GridEntityType::EntityOrientationType EntityOrientationType;
typedef typename GridEntityType::EntityBasisType EntityBasisType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( ! stepX + ! stepY == 1,
- std::cerr << "Only one of the steps can be non-zero: stepX = " << stepX << " stepY = " << stepY );
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ static_assert( ! stepX + ! stepY == 1, "Only one of the steps can be non-zero." );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() +
CoordinatesType( stepX + ( stepX < 0 ),
stepY + ( stepY < 0 ) ) >= CoordinatesType( 0, 0 ) &&
@@ -287,7 +271,7 @@ class NeighbourGridEntityGetter<
<< entity.getCoordinates() + CoordinatesType( stepX + ( stepX < 0 ), stepY + ( stepY < 0 ) )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->entity.getMesh(),
+ return NeighborGridEntityType( this->entity.getMesh(),
CoordinatesType( entity.getCoordinates().x() + stepX + ( stepX < 0 ),
entity.getCoordinates().y() + stepY + ( stepY < 0 ) ),
EntityOrientationType( stepX ? (stepX > 0 ? 1 : -1) : 0,
@@ -312,7 +296,7 @@ class NeighbourGridEntityGetter<
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stencil Storage |
+ * | EntityDimenions | NeighborEntityDimension | Stencil Storage |
* +-----------------+---------------------------+-------------------+
* | 2 | 0 | None |
* +-----------------+---------------------------+-------------------+
@@ -322,7 +306,7 @@ template< typename Real,
typename Index,
typename Config,
typename StencilStorage >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 2, Real, Device, Index >, 2, Config >,
0,
StencilStorage >
@@ -330,31 +314,28 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 2;
- static const int NeighbourEntityDimension = 0;
+ static const int NeighborEntityDimension = 0;
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
TNL_ASSERT( stepX != 0 && stepY != 0,
std::cerr << " stepX = " << stepX << " stepY = " << stepY );
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() +
CoordinatesType( stepX + ( stepX < 0 ), stepY + ( stepY < 0 ) ) >= CoordinatesType( 0, 0 ) &&
entity.getCoordinates() +
@@ -365,7 +346,7 @@ class NeighbourGridEntityGetter<
<< " entity.getMesh().getDimensions() + CoordinatesType( sign( stepX ), sign( stepY ) ) = "
<< entity.getMesh().getDimensions() + CoordinatesType( sign( stepX ), sign( stepY ) )
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->grid,
+ return NeighborGridEntityType( this->grid,
CoordinatesType( entity.getCoordinates().x() + stepX + ( stepX < 0 ),
entity.getCoordinates().y() + stepY + ( stepY < 0 ) ) );
}
@@ -384,12 +365,12 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stencil Storage |
+ * | EntityDimenions | NeighborEntityDimension | Stencil Storage |
* +-----------------+---------------------------+-------------------+
* | 1 | 2 | None |
* +-----------------+---------------------------+-------------------+
@@ -399,7 +380,7 @@ template< typename Real,
typename Index,
typename Config,
typename StencilStorage >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 2, Real, Device, Index >, 1, Config >,
2,
StencilStorage >
@@ -407,33 +388,30 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 1;
- static const int NeighbourEntityDimension = 2;
+ static const int NeighborEntityDimension = 2;
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
/*TNL_ASSERT( ( ( !! stepX ) == ( !! entity.getOrientation().x() ) ) &&
( ( !! stepY ) == ( !! entity.getOrientation().y() ) ),
std::cerr << "( stepX, stepY ) cannot be perpendicular to entity coordinates: stepX = " << stepX << " stepY = " << stepY
<< " entity.getOrientation() = " << entity.getOrientation() );*/
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions() + entity.getOrientation(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() + entity.getOrientation() = " << entity.getMesh().getDimensions() + entity.getOrientation()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() +
CoordinatesType( stepX - ( stepX > 0 ) * ( entity.getOrientation().x() != 0.0 ),
stepY - ( stepY > 0 ) * ( entity.getOrientation().y() != 0.0 ) ) >= CoordinatesType( 0, 0 ) &&
@@ -444,7 +422,7 @@ class NeighbourGridEntityGetter<
<< entity.getCoordinates() + CoordinatesType( stepX + ( stepX < 0 ), stepY + ( stepY < 0 ) )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->entity.getMesh(),
+ return NeighborGridEntityType( this->entity.getMesh(),
CoordinatesType( entity.getCoordinates().x() + stepX - ( stepX > 0 ) * ( entity.getOrientation().x() != 0.0 ),
entity.getCoordinates().y() + stepY - ( stepY > 0 ) * ( entity.getOrientation().y() != 0.0 ) ) );
}
@@ -466,7 +444,7 @@ class NeighbourGridEntityGetter<
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stencil Storage |
+ * | EntityDimenions | NeighborEntityDimension | Stencil Storage |
* +-----------------+---------------------------+-------------------+
* | 0 | 0 | None |
* +-----------------+---------------------------+-------------------+
@@ -476,7 +454,7 @@ template< typename Real,
typename Index,
typename Config,
typename StencilStorage >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 2, Real, Device, Index >, 0, Config >,
0,
StencilStorage >
@@ -484,35 +462,32 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 0;
- static const int NeighbourEntityDimension = 0;
+ static const int NeighborEntityDimension = 0;
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY ) >= CoordinatesType( 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY ) <= entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY ) = " << entity.getCoordinates() + CoordinatesType( stepX, stepY )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->grid,
+ return NeighborGridEntityType( this->grid,
CoordinatesType( entity.getCoordinates().x() + stepX,
entity.getCoordinates().y() + stepY ) );
}
@@ -521,11 +496,8 @@ class NeighbourGridEntityGetter<
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY ) >= CoordinatesType( 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY ) <= entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY ) = " << entity.getCoordinates() + CoordinatesType( stepX, stepY )
@@ -541,9 +513,8 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
} // namespace Meshes
} // namespace TNL
-
diff --git a/src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter3D_impl.h b/src/TNL/Meshes/GridDetails/NeighborGridEntityGetter3D_impl.h
similarity index 72%
rename from src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter3D_impl.h
rename to src/TNL/Meshes/GridDetails/NeighborGridEntityGetter3D_impl.h
index 682850bb1b3a442f7bf6a227418859496c690375..a7d2ab6d879cb1779b44dbe56f7f98464cd71473 100644
--- a/src/TNL/Meshes/GridDetails/NeighbourGridEntityGetter3D_impl.h
+++ b/src/TNL/Meshes/GridDetails/NeighborGridEntityGetter3D_impl.h
@@ -1,5 +1,5 @@
/***************************************************************************
- NeighbourGridEntityGetter3D_impl.h - description
+ NeighborGridEntityGetter3D_impl.h - description
-------------------
begin : Nov 23, 2015
copyright : (C) 2015 by Tomas Oberhuber
@@ -10,7 +10,7 @@
#pragma once
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntityGetter.h>
#include <TNL/Meshes/GridDetails/Grid1D.h>
#include <TNL/Meshes/GridDetails/Grid2D.h>
#include <TNL/Meshes/GridDetails/Grid3D.h>
@@ -21,7 +21,7 @@ namespace Meshes {
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 3 | 3 | None |
* +-----------------+---------------------------+-------------------+
@@ -30,7 +30,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 3, Real, Device, Index >, 3, Config >,
3,
GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -38,35 +38,32 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 3;
- static const int NeighbourEntityDimension = 3;
+ static const int NeighborEntityDimension = 3;
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY, int stepZ >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY ) >= CoordinatesType( 0, 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY ) = " << entity.getCoordinates() + CoordinatesType( stepX, stepY )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntity( CoordinatesType( entity.getCoordinates().x() + stepX,
+ return NeighborGridEntity( CoordinatesType( entity.getCoordinates().x() + stepX,
entity.getCoordinates().y() + stepY,
entity.getCoordinates().z() + stepZ ) );
}
@@ -75,11 +72,8 @@ class NeighbourGridEntityGetter<
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) >= CoordinatesType( 0, 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) = "
@@ -96,14 +90,14 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 3 | 3 | Cross |
* +-----------------+---------------------------+-------------------+
@@ -112,7 +106,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 3, Real, Device, Index >, 3, Config >,
3,
GridEntityStencilStorageTag< GridEntityCrossStencil > >
@@ -120,39 +114,36 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 3;
- static const int NeighbourEntityDimension = 3;
+ static const int NeighborEntityDimension = 3;
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
typedef GridEntityStencilStorageTag< GridEntityCrossStencil > StencilStorage;
- typedef NeighbourGridEntityGetter< GridEntityType, 3, StencilStorage > ThisType;
+ typedef NeighborGridEntityGetter< GridEntityType, 3, StencilStorage > ThisType;
static const int stencilSize = Config::getStencilSize();
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY, int stepZ >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) >= CoordinatesType( 0, 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY ) = " << entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->entity.getMesh(), CoordinatesType( entity.getCoordinates().x() + stepX,
+ return NeighborGridEntityType( this->entity.getMesh(), CoordinatesType( entity.getCoordinates().x() + stepX,
entity.getCoordinates().y() + stepY,
entity.getCoordinates().z() + stepZ ) );
}
@@ -161,11 +152,8 @@ class NeighbourGridEntityGetter<
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) >= CoordinatesType( 0, 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) < entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) = "
@@ -195,9 +183,9 @@ class NeighbourGridEntityGetter<
public:
__cuda_callable__
- static void exec( ThisType& neighbourEntityGetter, const IndexType& entityIndex )
+ static void exec( ThisType& neighborEntityGetter, const IndexType& entityIndex )
{
- neighbourEntityGetter.stencilX[ index + stencilSize ] = entityIndex + index;
+ neighborEntityGetter.stencilX[ index + stencilSize ] = entityIndex + index;
}
};
@@ -207,10 +195,10 @@ class NeighbourGridEntityGetter<
public:
__cuda_callable__
- static void exec( ThisType& neighbourEntityGetter, const IndexType& entityIndex )
+ static void exec( ThisType& neighborEntityGetter, const IndexType& entityIndex )
{
- neighbourEntityGetter.stencilY[ index + stencilSize ] =
- entityIndex + index * neighbourEntityGetter.entity.getMesh().getDimensions().x();
+ neighborEntityGetter.stencilY[ index + stencilSize ] =
+ entityIndex + index * neighborEntityGetter.entity.getMesh().getDimensions().x();
}
};
@@ -220,10 +208,10 @@ class NeighbourGridEntityGetter<
public:
__cuda_callable__
- static void exec( ThisType& neighbourEntityGetter, const IndexType& entityIndex )
+ static void exec( ThisType& neighborEntityGetter, const IndexType& entityIndex )
{
- neighbourEntityGetter.stencilZ[ index + stencilSize ] =
- entityIndex + index * neighbourEntityGetter.entity.getMesh().cellZNeighboursStep;
+ neighborEntityGetter.stencilZ[ index + stencilSize ] =
+ entityIndex + index * neighborEntityGetter.entity.getMesh().cellZNeighborsStep;
}
};
@@ -248,12 +236,12 @@ class NeighbourGridEntityGetter<
IndexType stencilY[ 2 * stencilSize + 1 ];
IndexType stencilZ[ 2 * stencilSize + 1 ];
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 3 | 2 | None |
* +-----------------+---------------------------+-------------------+
@@ -262,7 +250,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 3, Real, Device, Index >, 3, Config >,
2,
GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -270,35 +258,29 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 3;
- static const int NeighbourEntityDimension = 2;
+ static const int NeighborEntityDimension = 2;
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
typedef typename GridEntityType::EntityOrientationType EntityOrientationType;
typedef typename GridEntityType::EntityBasisType EntityBasisType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY, int stepZ >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( ! stepX + ! stepY + ! stepZ == 2,
- std::cerr << "Only one of the steps can be non-zero: stepX = " << stepX
- << " stepY = " << stepY
- << " stepZ = " << stepZ );
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ static_assert( ! stepX + ! stepY + ! stepZ == 2, "Only one of the steps can be non-zero." );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() +
CoordinatesType( stepX + ( stepX < 0 ),
stepY + ( stepY < 0 ),
@@ -313,7 +295,7 @@ class NeighbourGridEntityGetter<
<< entity.getCoordinates() + CoordinatesType( stepX + ( stepX < 0 ), stepY + ( stepY < 0 ), stepZ + ( stepZ < 0 ) )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->entity.getMesh(),
+ return NeighborGridEntityType( this->entity.getMesh(),
CoordinatesType( entity.getCoordinates().x() + stepX + ( stepX < 0 ),
entity.getCoordinates().y() + stepY + ( stepY < 0 ),
entity.getCoordinates().z() + stepZ + ( stepZ < 0 ) ),
@@ -337,12 +319,12 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/**** TODO: Finish it, knonw it is only a copy of specialization for none stored stencil
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 3 | 2 | Cross |
* +-----------------+---------------------------+-------------------+
@@ -351,7 +333,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 3, Real, Device, Index >, 3, Config >,
2,
GridEntityStencilStorageTag< GridEntityCrossStencil > >
@@ -359,35 +341,29 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 3;
- static const int NeighbourEntityDimension = 2;
+ static const int NeighborEntityDimension = 2;
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
typedef typename GridEntityType::EntityOrientationType EntityOrientationType;
typedef typename GridEntityType::EntityBasisType EntityBasisType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY, int stepZ >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( ! stepX + ! stepY + ! stepZ == 2,
- std::cerr << "Only one of the steps can be non-zero: stepX = " << stepX
- << " stepY = " << stepY
- << " stepZ = " << stepZ );
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ static_assert( ! stepX + ! stepY + ! stepZ == 2, "Only one of the steps can be non-zero." );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() +
CoordinatesType( stepX + ( stepX < 0 ),
stepY + ( stepY < 0 ),
@@ -402,7 +378,7 @@ class NeighbourGridEntityGetter<
<< entity.getCoordinates() + CoordinatesType( stepX + ( stepX < 0 ), stepY + ( stepY < 0 ), stepZ + ( stepZ < 0 ) )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( this->entity.getMesh(),
+ return NeighborGridEntityType( this->entity.getMesh(),
CoordinatesType( entity.getCoordinates().x() + stepX + ( stepX < 0 ),
entity.getCoordinates().y() + stepY + ( stepY < 0 ),
entity.getCoordinates().z() + stepZ + ( stepZ < 0 ) ),
@@ -426,13 +402,13 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 3 | 1 | None |
* +-----------------+---------------------------+-------------------+
@@ -441,7 +417,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 3, Real, Device, Index >, 3, Config >,
1,
GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -449,35 +425,29 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 3;
- static const int NeighbourEntityDimension = 1;
+ static const int NeighborEntityDimension = 1;
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
typedef typename GridEntityType::EntityOrientationType EntityOrientationType;
typedef typename GridEntityType::EntityBasisType EntityBasisType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY, int stepZ >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( ! stepX + ! stepY + ! stepZ == 1,
- std::cerr << "Exactly two of the steps must be non-zero: stepX = " << stepX
- << " stepY = " << stepY
- << " stepZ = " << stepZ );
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ static_assert( ! stepX + ! stepY + ! stepZ == 1, "Exactly two of the steps must be non-zero." );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() +
CoordinatesType( stepX + ( stepX < 0 ),
stepY + ( stepY < 0 ),
@@ -492,7 +462,7 @@ class NeighbourGridEntityGetter<
<< entity.getCoordinates() + CoordinatesType( stepX + ( stepX < 0 ), stepY + ( stepY < 0 ), stepZ + ( stepZ < 0 ) )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntity( CoordinatesType( entity.getCoordinates().x() + stepX + ( stepX < 0 ),
+ return NeighborGridEntity( CoordinatesType( entity.getCoordinates().x() + stepX + ( stepX < 0 ),
entity.getCoordinates().y() + stepY + ( stepY < 0 ),
entity.getCoordinates().z() + stepZ + ( stepZ < 0 ) ),
EntityOrientationType( !!stepX, !!stepY, !!stepZ ),
@@ -513,13 +483,13 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 3 | 0 | None |
* +-----------------+---------------------------+-------------------+
@@ -528,7 +498,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 3, Real, Device, Index >, 3, Config >,
0,
GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -536,33 +506,30 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 3;
- static const int NeighbourEntityDimension = 0;
+ static const int NeighborEntityDimension = 0;
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY,int stepZ >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
TNL_ASSERT( stepX != 0 && stepY != 0 && stepZ != 0,
std::cerr << " stepX = " << stepX
<< " stepY = " << stepY
<< " stepZ = " << stepZ );
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() +
CoordinatesType( stepX + ( stepX < 0 ),
stepY + ( stepY < 0 ),
@@ -578,7 +545,7 @@ class NeighbourGridEntityGetter<
<< " entity.getMesh().getDimensions() + CoordinatesType( sign( stepX ), sign( stepY ), sign( stepZ ) ) = "
<< entity.getMesh().getDimensions() + CoordinatesType( sign( stepX ), sign( stepY ), sign( stepZ ) )
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntity( CoordinatesType( entity.getCoordinates().x() + stepX + ( stepX < 0 ),
+ return NeighborGridEntity( CoordinatesType( entity.getCoordinates().x() + stepX + ( stepX < 0 ),
entity.getCoordinates().y() + stepY + ( stepY < 0 ),
entity.getCoordinates().z() + stepZ + ( stepZ < 0 ) ) );
}
@@ -597,12 +564,12 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 2 | 3 | None |
* +-----------------+---------------------------+-------------------+
@@ -611,7 +578,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 3, Real, Device, Index >, 2, Config >,
3,
GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -619,23 +586,23 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 2;
- static const int NeighbourEntityDimension = 3;
+ static const int NeighborEntityDimension = 3;
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY, int stepZ >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
/*TNL_ASSERT( ( ( !! stepX ) == ( !! entity.getOrientation().x() ) ) &&
( ( !! stepY ) == ( !! entity.getOrientation().y() ) ) &&
@@ -643,11 +610,8 @@ class NeighbourGridEntityGetter<
std::cerr << "( stepX, stepY, stepZ ) cannot be perpendicular to entity coordinates: stepX = " << stepX
<< " stepY = " << stepY << " stepZ = " << stepZ
<< " entity.getOrientation() = " << entity.getOrientation() );*/
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() < entity.getMesh().getDimensions() + entity.getOrientation(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() + entity.getOrientation() = " << entity.getMesh().getDimensions() + entity.getOrientation()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LT( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() +
CoordinatesType( stepX - ( stepX > 0 ) * ( entity.getOrientation().x() != 0.0 ),
stepY - ( stepY > 0 ) * ( entity.getOrientation().y() != 0.0 ),
@@ -663,7 +627,7 @@ class NeighbourGridEntityGetter<
stepZ + ( stepZ < 0 ) * ( entity.getOrientation().z() != 0.0 ) )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntityType( entity.getMesh(),
+ return NeighborGridEntityType( entity.getMesh(),
CoordinatesType( entity.getCoordinates().x() + stepX - ( stepX > 0 ) * ( entity.getOrientation().x() != 0.0 ),
entity.getCoordinates().y() + stepY - ( stepY > 0 ) * ( entity.getOrientation().y() != 0.0 ),
entity.getCoordinates().z() + stepZ - ( stepZ > 0 ) * ( entity.getOrientation().z() != 0.0 ) ) );
@@ -683,12 +647,12 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stored Stencil |
+ * | EntityDimenions | NeighborEntityDimension | Stored Stencil |
* +-----------------+---------------------------+-------------------+
* | 0 | 0 | None |
* +-----------------+---------------------------+-------------------+
@@ -697,7 +661,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class NeighbourGridEntityGetter<
+class NeighborGridEntityGetter<
GridEntity< Meshes::Grid< 3, Real, Device, Index >, 0, Config >,
0,
GridEntityStencilStorageTag< GridEntityNoStencil > >
@@ -705,36 +669,33 @@ class NeighbourGridEntityGetter<
public:
static const int EntityDimension = 0;
- static const int NeighbourEntityDimension = 0;
+ static const int NeighborEntityDimension = 0;
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetterType;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetterType;
__cuda_callable__ inline
- NeighbourGridEntityGetter( const GridEntityType& entity )
+ NeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
template< int stepX, int stepY, int stepZ >
__cuda_callable__ inline
- NeighbourGridEntityType getEntity() const
+ NeighborGridEntityType getEntity() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) >= CoordinatesType( 0, 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) <= entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) = "
<< entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ )
<< " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
<< " EntityDimension = " << EntityDimension );
- return NeighbourGridEntity( CoordinatesType( entity.getCoordinates().x() + stepX,
+ return NeighborGridEntity( CoordinatesType( entity.getCoordinates().x() + stepX,
entity.getCoordinates().y() + stepY,
entity.getCoordinates().z() + stepZ ) );
}
@@ -743,11 +704,8 @@ class NeighbourGridEntityGetter<
__cuda_callable__ inline
IndexType getEntityIndex() const
{
- TNL_ASSERT( entity.getCoordinates() >= CoordinatesType( 0, 0, 0 ) &&
- entity.getCoordinates() <= entity.getMesh().getDimensions(),
- std::cerr << "entity.getCoordinates() = " << entity.getCoordinates()
- << " entity.getMesh().getDimensions() = " << entity.getMesh().getDimensions()
- << " EntityDimension = " << EntityDimension );
+ TNL_ASSERT_GE( entity.getCoordinates(), CoordinatesType( 0, 0, 0 ), "wrong coordinates" );
+ TNL_ASSERT_LE( entity.getCoordinates(), entity.getMesh().getDimensions(), "wrong coordinates" );
TNL_ASSERT( entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) >= CoordinatesType( 0, 0, 0 ) &&
entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) <= entity.getMesh().getDimensions(),
std::cerr << "entity.getCoordinates() + CoordinatesType( stepX, stepY, stepZ ) = "
@@ -764,10 +722,9 @@ class NeighbourGridEntityGetter<
const GridEntityType& entity;
- //NeighbourGridEntityGetter(){};
+ //NeighborGridEntityGetter(){};
};
} // namespace Meshes
-} // namespace TNL
-
+} // namespace TNL
\ No newline at end of file
diff --git a/src/TNL/Meshes/GridDetails/NeighbourGridEntitiesStorage.h b/src/TNL/Meshes/GridDetails/NeighbourGridEntitiesStorage.h
deleted file mode 100644
index ec28a6fffd7ee33933b7d3b8897ab41d110d46b1..0000000000000000000000000000000000000000
--- a/src/TNL/Meshes/GridDetails/NeighbourGridEntitiesStorage.h
+++ /dev/null
@@ -1,173 +0,0 @@
-/***************************************************************************
- NeighbourGridEntitiesStorage.h - description
- -------------------
- begin : Dec 18, 2015
- copyright : (C) 2015 by Tomas Oberhuber
- email : tomas.oberhuber@fjfi.cvut.cz
- ***************************************************************************/
-
-/* See Copyright Notice in tnl/Copyright */
-
-#pragma once
-
-#include <TNL/Devices/Cuda.h>
-#include <TNL/Meshes/MeshDimensionTag.h>
-#include <TNL/Meshes/GridEntityConfig.h>
-#include <TNL/Meshes/GridDetails/NeighbourGridEntityGetter.h>
-
-namespace TNL {
-namespace Meshes {
-
-template< typename GridEntity,
- int NeighbourEntityDimension,
- typename GridEntityConfig,
- bool storage = GridEntityConfig::template neighbourEntityStorage< GridEntity >( NeighbourEntityDimension ) >
-class NeighbourGridEntityLayer{};
-
-template< typename GridEntity,
- int NeighbourEntityDimension,
- typename GridEntityConfig >
-class NeighbourGridEntityLayer< GridEntity, NeighbourEntityDimension, GridEntityConfig, true >
-: public NeighbourGridEntityLayer< GridEntity, NeighbourEntityDimension - 1, GridEntityConfig >
-{
- public:
-
- typedef NeighbourGridEntityLayer< GridEntity, NeighbourEntityDimension - 1, GridEntityConfig > BaseType;
- typedef NeighbourGridEntityGetter< GridEntity, NeighbourEntityDimension > NeighbourEntityGetterType;
-
- using BaseType::getNeighbourEntities;
-
- __cuda_callable__
- NeighbourGridEntityLayer( const GridEntity& entity )
- : BaseType( entity ),
- neighbourEntities( entity )
- {}
-
- __cuda_callable__
- const NeighbourEntityGetterType& getNeighbourEntities( const MeshDimensionTag< NeighbourEntityDimension>& tag ) const
- {
- return this->neighbourEntities;
- }
-
- __cuda_callable__
- void refresh( const typename GridEntity::GridType& grid,
- const typename GridEntity::GridType::IndexType& entityIndex )
- {
- BaseType::refresh( grid, entityIndex );
- neighbourEntities.refresh( grid, entityIndex );
- }
-
- protected:
-
- NeighbourEntityGetterType neighbourEntities;
-};
-
-template< typename GridEntity,
- typename GridEntityConfig >
-class NeighbourGridEntityLayer< GridEntity, 0, GridEntityConfig, true >
-{
- public:
-
- typedef NeighbourGridEntityGetter< GridEntity, 0 > NeighbourEntityGetterType;
-
- __cuda_callable__
- NeighbourGridEntityLayer( const GridEntity& entity )
- : neighbourEntities( entity )
- {}
-
- __cuda_callable__
- const NeighbourEntityGetterType& getNeighbourEntities( const MeshDimensionTag< 0 >& tag ) const
- {
- return this->neighbourEntities;
- }
-
- __cuda_callable__
- void refresh( const typename GridEntity::GridType& grid,
- const typename GridEntity::GridType::IndexType& entityIndex )
- {
- neighbourEntities.refresh( grid, entityIndex );
- }
-
- protected:
-
- NeighbourEntityGetterType neighbourEntities;
-};
-
-template< typename GridEntity,
- int NeighbourEntityDimension,
- typename GridEntityConfig >
-class NeighbourGridEntityLayer< GridEntity, NeighbourEntityDimension, GridEntityConfig, false >
-: public NeighbourGridEntityLayer< GridEntity, NeighbourEntityDimension - 1, GridEntityConfig >
-{
- public:
-
- typedef NeighbourGridEntityLayer< GridEntity, NeighbourEntityDimension - 1, GridEntityConfig > BaseType;
- typedef NeighbourGridEntityGetter< GridEntity, NeighbourEntityDimension > NeighbourEntityGetterType;
-
- using BaseType::getNeighbourEntities;
-
- __cuda_callable__
- NeighbourGridEntityLayer( const GridEntity& entity )
- : BaseType( entity )
- {}
-
- __cuda_callable__
- const NeighbourEntityGetterType& getNeighbourEntities( const MeshDimensionTag< NeighbourEntityDimension >& tag ) const {}
-
- __cuda_callable__
- void refresh( const typename GridEntity::GridType& grid,
- const typename GridEntity::GridType::IndexType& entityIndex ) {}
-};
-
-template< typename GridEntity,
- typename GridEntityConfig >
-class NeighbourGridEntityLayer< GridEntity, 0, GridEntityConfig, false >
-{
- public:
-
- typedef NeighbourGridEntityGetter< GridEntity, 0 > NeighbourEntityGetterType;
-
- __cuda_callable__
- NeighbourGridEntityLayer( const GridEntity& entity ){}
-
- __cuda_callable__
- const NeighbourEntityGetterType& getNeighbourEntities( const MeshDimensionTag< 0 >& tag ) const {}
-
- __cuda_callable__
- void refresh( const typename GridEntity::GridType& grid,
- const typename GridEntity::GridType::IndexType& entityIndex ) {}
-};
-
-
-
-
-template< typename GridEntity,
- typename GridEntityConfig >
-class NeighbourGridEntitiesStorage
-: public NeighbourGridEntityLayer< GridEntity, GridEntity::meshDimension, GridEntityConfig >
-{
- typedef NeighbourGridEntityLayer< GridEntity, GridEntity::meshDimension, GridEntityConfig > BaseType;
-
- public:
-
- using BaseType::getNeighbourEntities;
- using BaseType::refresh;
-
- __cuda_callable__
- NeighbourGridEntitiesStorage( const GridEntity& entity )
- : BaseType( entity )
- {}
-
- template< int EntityDimension >
- __cuda_callable__
- const NeighbourGridEntityGetter< GridEntity, EntityDimension >&
- getNeighbourEntities() const
- {
- return BaseType::getNeighbourEntities( MeshDimensionTag< EntityDimension >() );
- }
-};
-
-
-} // namespace Meshes
-} // namespace TNL
-
diff --git a/src/TNL/Meshes/GridDetails/Traverser_Grid1D.h b/src/TNL/Meshes/GridDetails/Traverser_Grid1D.h
index 400f71531a5f6f5d92606d9da52b991839098b94..a9310fd1e54faa1e417086f4c5f85765be4b6974 100644
--- a/src/TNL/Meshes/GridDetails/Traverser_Grid1D.h
+++ b/src/TNL/Meshes/GridDetails/Traverser_Grid1D.h
@@ -25,25 +25,22 @@ class Traverser< Meshes::Grid< 1, Real, Device, Index >, GridEntity, 1 >
public:
typedef Meshes::Grid< 1, Real, Device, Index > GridType;
typedef SharedPointer< GridType > GridPointer;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
template< typename UserData,
typename EntitiesProcessor >
void processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
};
@@ -57,25 +54,22 @@ class Traverser< Meshes::Grid< 1, Real, Device, Index >, GridEntity, 0 >
public:
typedef Meshes::Grid< 1, Real, Device, Index > GridType;
typedef SharedPointer< GridType > GridPointer;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
template< typename UserData,
typename EntitiesProcessor >
void processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
};
} // namespace Meshes
diff --git a/src/TNL/Meshes/GridDetails/Traverser_Grid1D_impl.h b/src/TNL/Meshes/GridDetails/Traverser_Grid1D_impl.h
index 6fcc348b509bff617b4739f7dfd02cac2f750651..c6102d04c574a32aa4023f5a0ccd0630528db5f7 100644
--- a/src/TNL/Meshes/GridDetails/Traverser_Grid1D_impl.h
+++ b/src/TNL/Meshes/GridDetails/Traverser_Grid1D_impl.h
@@ -29,12 +29,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 1, Real, Device, Index >, GridEntity, 1 >::
processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Boundary cells
*/
- static_assert( GridEntity::entityDimension == 1, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 1, "The entity has wrong dimensions." );
auto distributedgrid=gridPointer->GetDistGrid();
if(distributedgrid==nullptr||!distributedgrid->isMPIUsed())
@@ -80,12 +80,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 1, Real, Device, Index >, GridEntity, 1 >::
processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Interior cells
*/
- static_assert( GridEntity::entityDimension == 1, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 1, "The entity has wrong dimension." );
auto distributedgrid=gridPointer->GetDistGrid();
if(distributedgrid==nullptr||!distributedgrid->isMPIUsed())
@@ -132,13 +132,13 @@ void
Traverser< Meshes::Grid< 1, Real, Device, Index >, GridEntity, 1 >::
processAllEntities(
const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* All cells
*/
- static_assert( GridEntity::entityDimension == 1, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 1, "The entity has wrong dimensions." );
auto distributedgrid=gridPointer->GetDistGrid();
if(distributedgrid==nullptr||!distributedgrid->isMPIUsed())
@@ -187,12 +187,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 1, Real, Device, Index >, GridEntity, 0 >::
processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Boundary vertices
*/
- static_assert( GridEntity::entityDimension == 0, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 0, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, true >(
gridPointer,
@@ -210,12 +210,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 1, Real, Device, Index >, GridEntity, 0 >::
processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Interior vertices
*/
- static_assert( GridEntity::entityDimension == 0, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 0, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false >(
gridPointer,
@@ -234,12 +234,12 @@ void
Traverser< Meshes::Grid< 1, Real, Device, Index >, GridEntity, 0 >::
processAllEntities(
const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* All vertices
*/
- static_assert( GridEntity::entityDimension == 0, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 0, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false >(
gridPointer,
diff --git a/src/TNL/Meshes/GridDetails/Traverser_Grid2D.h b/src/TNL/Meshes/GridDetails/Traverser_Grid2D.h
index 6123ce9d1a712a916775c45fe3d856e285f8196c..d80205feede8fb2e08cbcd4e92fe91d3c852904f 100644
--- a/src/TNL/Meshes/GridDetails/Traverser_Grid2D.h
+++ b/src/TNL/Meshes/GridDetails/Traverser_Grid2D.h
@@ -25,24 +25,21 @@ class Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 2 >
public:
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef SharedPointer< GridType > GridPointer;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
template< typename UserData,
typename EntitiesProcessor >
void processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
};
@@ -55,25 +52,22 @@ class Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 1 >
public:
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef SharedPointer< GridType > GridPointer;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
template< typename UserData,
typename EntitiesProcessor >
void processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
};
@@ -86,25 +80,22 @@ class Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 0 >
public:
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef SharedPointer< GridType > GridPointer;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
template< typename UserData,
typename EntitiesProcessor >
void processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
};
} // namespace Meshes
diff --git a/src/TNL/Meshes/GridDetails/Traverser_Grid2D_impl.h b/src/TNL/Meshes/GridDetails/Traverser_Grid2D_impl.h
index 544076f8328c301347813c81e169f879d36e79d0..8cdc9eafe673588965ba2cdf2f7bb4fa3a144188 100644
--- a/src/TNL/Meshes/GridDetails/Traverser_Grid2D_impl.h
+++ b/src/TNL/Meshes/GridDetails/Traverser_Grid2D_impl.h
@@ -27,12 +27,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 2 >::
processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Boundary cells
*/
- static_assert( GridEntity::entityDimension == 2, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 2, "The entity has wrong dimension." );
auto distributedgrid=gridPointer->GetDistGrid();
if(distributedgrid==nullptr||!distributedgrid->isMPIUsed())
@@ -101,13 +101,13 @@ template< typename Real,
void
Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 2 >::
processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Interior cells
*/
- static_assert( GridEntity::entityDimension == 2, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 2, "The entity has wrong dimensions." );
auto distributedgrid=gridPointer->GetDistGrid();
if(distributedgrid==nullptr||!distributedgrid->isMPIUsed())
@@ -166,12 +166,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 2 >::
processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* All cells
*/
- static_assert( GridEntity::entityDimension == 2, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 2, "The entity has wrong dimension." );
auto distributedgrid=gridPointer->GetDistGrid();
if(distributedgrid==nullptr||!distributedgrid->isMPIUsed())
@@ -233,12 +233,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 1 >::
processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Boundary faces
*/
- static_assert( GridEntity::entityDimension == 1, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 1, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, true, 1, 0, CoordinatesType, CoordinatesType >(
gridPointer,
@@ -268,12 +268,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 1 >::
processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Interior faces
*/
- static_assert( GridEntity::entityDimension == 1, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 1, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false, 1, 1, CoordinatesType, CoordinatesType >(
gridPointer,
@@ -303,12 +303,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 1 >::
processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* All faces
*/
- static_assert( GridEntity::entityDimension == 1, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 1, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false, 1, 1, CoordinatesType, CoordinatesType >(
gridPointer,
@@ -338,12 +338,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 0 >::
processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Boundary vertices
*/
- static_assert( GridEntity::entityDimension == 0, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 0, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, true, 1, 1 >(
gridPointer,
@@ -362,12 +362,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 0 >::
processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Interior vertices
*/
- static_assert( GridEntity::entityDimension == 0, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 0, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false >(
gridPointer,
@@ -386,12 +386,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 2, Real, Device, Index >, GridEntity, 0 >::
processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* All vertices
*/
- static_assert( GridEntity::entityDimension == 0, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 0, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false >(
gridPointer,
diff --git a/src/TNL/Meshes/GridDetails/Traverser_Grid3D.h b/src/TNL/Meshes/GridDetails/Traverser_Grid3D.h
index a03503e302bb346cf096d893859fe61fcb4de24d..705e4bbc033ca36708306894ec15186074d9460a 100644
--- a/src/TNL/Meshes/GridDetails/Traverser_Grid3D.h
+++ b/src/TNL/Meshes/GridDetails/Traverser_Grid3D.h
@@ -25,24 +25,21 @@ class Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 3 >
public:
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef SharedPointer< GridType > GridPointer;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
template< typename UserData,
typename EntitiesProcessor >
void processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
};
@@ -55,24 +52,21 @@ class Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 2 >
public:
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef SharedPointer< GridType > GridPointer;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
template< typename UserData,
typename EntitiesProcessor >
void processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
};
@@ -85,25 +79,22 @@ class Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 1 >
public:
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef SharedPointer< GridType > GridPointer;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
template< typename UserData,
typename EntitiesProcessor >
void processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
};
@@ -116,25 +107,22 @@ class Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 0 >
public:
typedef Meshes::Grid< 3, Real, Device, Index > GridType;
typedef SharedPointer< GridType > GridPointer;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
template< typename UserData,
typename EntitiesProcessor >
void processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
template< typename UserData,
typename EntitiesProcessor >
void processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const;
+ SharedPointer< UserData, Device >& userDataPointer ) const;
};
} // namespace Meshes
diff --git a/src/TNL/Meshes/GridDetails/Traverser_Grid3D_impl.h b/src/TNL/Meshes/GridDetails/Traverser_Grid3D_impl.h
index 5f98a59d55855099162f70718ef4d75f3e8f9107..84e44ef17c29aca27da6c940e6ab29841fb9aa86 100644
--- a/src/TNL/Meshes/GridDetails/Traverser_Grid3D_impl.h
+++ b/src/TNL/Meshes/GridDetails/Traverser_Grid3D_impl.h
@@ -27,12 +27,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 3 >::
processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Boundary cells
*/
- static_assert( GridEntity::entityDimension == 3, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 3, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, true, 1, 1, 1 >(
gridPointer,
@@ -51,12 +51,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 3 >::
processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Interior cells
*/
- static_assert( GridEntity::entityDimension == 3, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 3, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false >(
gridPointer,
@@ -75,12 +75,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 3 >::
processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* All cells
*/
- static_assert( GridEntity::entityDimension == 3, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 3, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false >(
gridPointer,
@@ -102,12 +102,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 2 >::
processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Boundary faces
*/
- static_assert( GridEntity::entityDimension == 2, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 2, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, true, 1, 0, 0, CoordinatesType, CoordinatesType >(
gridPointer,
@@ -146,12 +146,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 2 >::
processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Interior faces
*/
- static_assert( GridEntity::entityDimension == 2, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 2, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false, 1, 1, 1, CoordinatesType, CoordinatesType >(
gridPointer,
@@ -190,12 +190,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 2 >::
processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* All faces
*/
- static_assert( GridEntity::entityDimension == 2, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 2, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false, 1, 1, 1, CoordinatesType, CoordinatesType >(
gridPointer,
CoordinatesType( 0, 0, 0 ),
@@ -236,12 +236,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 1 >::
processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Boundary edges
*/
- static_assert( GridEntity::entityDimension == 1, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 1, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, true, 0, 1, 1, CoordinatesType, CoordinatesType >(
gridPointer,
@@ -280,12 +280,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 1 >::
processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Interior edges
*/
- static_assert( GridEntity::entityDimension == 1, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 1, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false, 1, 1, 1, CoordinatesType, CoordinatesType >(
gridPointer,
@@ -324,12 +324,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 1 >::
processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* All edges
*/
- static_assert( GridEntity::entityDimension == 1, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 1, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false, 1, 1, 1, CoordinatesType, CoordinatesType >(
gridPointer,
CoordinatesType( 0, 0, 0 ),
@@ -370,12 +370,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 0 >::
processBoundaryEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Boundary vertices
*/
- static_assert( GridEntity::entityDimension == 0, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 0, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, true, 1, 1, 1 >(
gridPointer,
@@ -394,12 +394,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 0 >::
processInteriorEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* Interior vertices
*/
- static_assert( GridEntity::entityDimension == 0, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 0, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false >(
gridPointer,
@@ -418,12 +418,12 @@ template< typename Real,
void
Traverser< Meshes::Grid< 3, Real, Device, Index >, GridEntity, 0 >::
processAllEntities( const GridPointer& gridPointer,
- SharedPointer< UserData, DeviceType >& userDataPointer ) const
+ SharedPointer< UserData, Device >& userDataPointer ) const
{
/****
* All vertices
*/
- static_assert( GridEntity::entityDimension == 0, "The entity has wrong dimensions." );
+ static_assert( GridEntity::getEntityDimension() == 0, "The entity has wrong dimension." );
GridTraverser< GridType >::template processEntities< GridEntity, EntitiesProcessor, UserData, false >(
gridPointer,
diff --git a/src/TNL/Meshes/GridEntity.h b/src/TNL/Meshes/GridEntity.h
index dee16543008d5ac79551403d2c7cd210527756e4..4db50173285c267e61736a36827073dcb323a002 100644
--- a/src/TNL/Meshes/GridEntity.h
+++ b/src/TNL/Meshes/GridEntity.h
@@ -10,15 +10,15 @@
#pragma once
-#include <TNL/Meshes/GridDetails/NeighbourGridEntitiesStorage.h>
+#include <TNL/Meshes/GridDetails/NeighborGridEntitiesStorage.h>
namespace TNL {
namespace Meshes {
template< typename GridEntity,
- int NeighbourEntityDimension,
+ int NeighborEntityDimension,
typename StencilStorage >
-class NeighbourGridEntityGetter;
+class NeighborGridEntityGetter;
template< typename GridEntityType >
class BoundaryGridEntityChecker;
@@ -51,28 +51,24 @@ class GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, EntityDimensio
typedef typename GridType::CoordinatesType CoordinatesType;
typedef Config ConfigType;
- static const int meshDimension = GridType::meshDimension;
+ constexpr static int getMeshDimension() { return GridType::getMeshDimension(); };
- static const int entityDimension = EntityDimension;
+ constexpr static int getEntityDimension() { return EntityDimension; };
- constexpr static int getDimension() { return EntityDimension; };
-
- constexpr static int getMeshDimension() { return meshDimension; };
-
- typedef Containers::StaticVector< meshDimension, IndexType > EntityOrientationType;
- typedef Containers::StaticVector< meshDimension, IndexType > EntityBasisType;
- typedef GridEntity< GridType, entityDimension, Config > ThisType;
+ typedef Containers::StaticVector< getMeshDimension(), IndexType > EntityOrientationType;
+ typedef Containers::StaticVector< getMeshDimension(), IndexType > EntityBasisType;
+ typedef GridEntity< GridType, EntityDimension, Config > ThisType;
typedef typename GridType::PointType PointType;
- typedef NeighbourGridEntitiesStorage< ThisType, Config > NeighbourGridEntitiesStorageType;
+ typedef NeighborGridEntitiesStorage< ThisType, Config > NeighborGridEntitiesStorageType;
- template< int NeighbourEntityDimension = entityDimension >
- using NeighbourEntities =
- NeighbourGridEntityGetter<
+ template< int NeighborEntityDimension = getEntityDimension() >
+ using NeighborEntities =
+ NeighborGridEntityGetter<
GridEntity< Meshes::Grid< Dimension, Real, Device, Index >,
EntityDimension,
Config >,
- NeighbourEntityDimension >;
+ NeighborEntityDimension >;
__cuda_callable__ inline
GridEntity( const GridType& grid );
@@ -116,10 +112,10 @@ class GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, EntityDimensio
__cuda_callable__ inline
void setBasis( const EntityBasisType& basis );
- template< int NeighbourEntityDimension = entityDimension >
+ template< int NeighborEntityDimension = getEntityDimension() >
__cuda_callable__ inline
- const NeighbourEntities< NeighbourEntityDimension >&
- getNeighbourEntities() const;
+ const NeighborEntities< NeighborEntityDimension >&
+ getNeighborEntities() const;
__cuda_callable__ inline
bool isBoundaryEntity() const;
@@ -145,7 +141,7 @@ class GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, EntityDimensio
EntityBasisType basis;
- NeighbourGridEntitiesStorageType neighbourEntitiesStorage;
+ NeighborGridEntitiesStorageType neighborEntitiesStorage;
//__cuda_callable__ inline
//GridEntity();
@@ -175,27 +171,23 @@ class GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension, Con
typedef typename GridType::PointType PointType;
typedef Config ConfigType;
- static const int meshDimension = GridType::meshDimension;
-
- static const int entityDimension = meshDimension;
-
- constexpr static int getDimension() { return entityDimension; };
+ constexpr static int getMeshDimension() { return GridType::getMeshDimension(); };
- constexpr static int getMeshDimension() { return meshDimension; };
+ constexpr static int getEntityDimension() { return getMeshDimension(); };
- typedef Containers::StaticVector< meshDimension, IndexType > EntityOrientationType;
- typedef Containers::StaticVector< meshDimension, IndexType > EntityBasisType;
- typedef GridEntity< GridType, entityDimension, Config > ThisType;
- typedef NeighbourGridEntitiesStorage< ThisType, Config > NeighbourGridEntitiesStorageType;
+ typedef Containers::StaticVector< getMeshDimension(), IndexType > EntityOrientationType;
+ typedef Containers::StaticVector< getMeshDimension(), IndexType > EntityBasisType;
+ typedef GridEntity< GridType, Dimension, Config > ThisType;
+ typedef NeighborGridEntitiesStorage< ThisType, Config > NeighborGridEntitiesStorageType;
- template< int NeighbourEntityDimension = entityDimension >
- using NeighbourEntities =
- NeighbourGridEntityGetter<
+ template< int NeighborEntityDimension = getEntityDimension() >
+ using NeighborEntities =
+ NeighborGridEntityGetter<
GridEntity< Meshes::Grid< Dimension, Real, Device, Index >,
- entityDimension,
+ Dimension,
Config >,
- NeighbourEntityDimension >;
+ NeighborEntityDimension >;
__cuda_callable__ inline
@@ -240,10 +232,10 @@ class GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension, Con
__cuda_callable__ inline
void setBasis( const EntityBasisType& basis ){};
- template< int NeighbourEntityDimension = Dimension >
+ template< int NeighborEntityDimension = Dimension >
__cuda_callable__ inline
- const NeighbourEntities< NeighbourEntityDimension >&
- getNeighbourEntities() const;
+ const NeighborEntities< NeighborEntityDimension >&
+ getNeighborEntities() const;
__cuda_callable__ inline
bool isBoundaryEntity() const;
@@ -268,7 +260,7 @@ class GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension, Con
CoordinatesType coordinates;
- NeighbourGridEntitiesStorageType neighbourEntitiesStorage;
+ NeighborGridEntitiesStorageType neighborEntitiesStorage;
//__cuda_callable__ inline
//GridEntity();
@@ -298,26 +290,22 @@ class GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, 0, Config >
typedef typename GridType::PointType PointType;
typedef Config ConfigType;
- static const int meshDimension = GridType::meshDimension;
-
- static const int entityDimension = 0;
-
- constexpr static int getDimension() { return entityDimension; };
+ constexpr static int getMeshDimension() { return GridType::getMeshDimension(); };
- constexpr static int getMeshDimension() { return meshDimension; };
+ constexpr static int getEntityDimension() { return 0; };
- typedef Containers::StaticVector< meshDimension, IndexType > EntityOrientationType;
- typedef Containers::StaticVector< meshDimension, IndexType > EntityBasisType;
- typedef GridEntity< GridType, entityDimension, Config > ThisType;
- typedef NeighbourGridEntitiesStorage< ThisType, Config > NeighbourGridEntitiesStorageType;
+ typedef Containers::StaticVector< getMeshDimension(), IndexType > EntityOrientationType;
+ typedef Containers::StaticVector< getMeshDimension(), IndexType > EntityBasisType;
+ typedef GridEntity< GridType, 0, Config > ThisType;
+ typedef NeighborGridEntitiesStorage< ThisType, Config > NeighborGridEntitiesStorageType;
- template< int NeighbourEntityDimension = entityDimension >
- using NeighbourEntities =
- NeighbourGridEntityGetter<
+ template< int NeighborEntityDimension = getEntityDimension() >
+ using NeighborEntities =
+ NeighborGridEntityGetter<
GridEntity< Meshes::Grid< Dimension, Real, Device, Index >,
- entityDimension,
+ 0,
Config >,
- NeighbourEntityDimension >;
+ NeighborEntityDimension >;
__cuda_callable__ inline
@@ -363,10 +351,10 @@ class GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, 0, Config >
void setBasis( const EntityBasisType& basis ){};
- template< int NeighbourEntityDimension = entityDimension >
+ template< int NeighborEntityDimension = getEntityDimension() >
__cuda_callable__ inline
- const NeighbourEntities< NeighbourEntityDimension >&
- getNeighbourEntities() const;
+ const NeighborEntities< NeighborEntityDimension >&
+ getNeighborEntities() const;
__cuda_callable__ inline
bool isBoundaryEntity() const;
@@ -391,7 +379,7 @@ class GridEntity< Meshes::Grid< Dimension, Real, Device, Index >, 0, Config >
CoordinatesType coordinates;
- NeighbourGridEntitiesStorageType neighbourEntitiesStorage;
+ NeighborGridEntitiesStorageType neighborEntitiesStorage;
friend class BoundaryGridEntityChecker< ThisType >;
diff --git a/src/TNL/Meshes/GridEntityConfig.h b/src/TNL/Meshes/GridEntityConfig.h
index e382bdbba82827dc9e3d39760bd6d92f619b7129..a056910a37021c04c383a6673021cb8a072052b6 100644
--- a/src/TNL/Meshes/GridEntityConfig.h
+++ b/src/TNL/Meshes/GridEntityConfig.h
@@ -29,9 +29,9 @@ class GridEntityStencilStorageTag
};
/****
- * This class says what neighbour grid entity indexes shall be pre-computed and stored in the
- * grid entity structure. If neighbourEntityStorage() returns false, nothing is stored.
- * Otherwise, if neighbour entity storage is enabled, we may store either only neighbour entities in a cross like this
+ * This class says what neighbor grid entity indexes shall be pre-computed and stored in the
+ * grid entity structure. If neighborEntityStorage() returns false, nothing is stored.
+ * Otherwise, if neighbor entity storage is enabled, we may store either only neighbor entities in a cross like this
*
* X
* X X
@@ -39,7 +39,7 @@ class GridEntityStencilStorageTag
* X X
* X
*
- * or all neighbour entities like this
+ * or all neighbor entities like this
*
* XXXXX
* XXX XXXXX
@@ -53,7 +53,7 @@ class GridEntityNoStencilStorage
public:
template< typename GridEntity >
- constexpr static bool neighbourEntityStorage( int neighbourEntityStorage )
+ constexpr static bool neighborEntityStorage( int neighborEntityStorage )
{
return false;
}
@@ -70,12 +70,10 @@ class GridEntityCrossStencilStorage
public:
template< typename GridEntity >
- constexpr static bool neighbourEntityStorage( const int neighbourEntityDimension )
+ constexpr static bool neighborEntityStorage( const int neighborEntityDimension )
{
- return ( GridEntity::entityDimension == GridEntity::GridType::meshDimension &&
- neighbourEntityDimension == GridEntity::GridType::meshDimension )
- // FIXME: how is GridEntityCrossStencil cast to int?
- * GridEntityCrossStencil;
+ return ( GridEntity::getEntityDimension() == GridEntity::GridType::getMeshDimension() &&
+ neighborEntityDimension == GridEntity::GridType::getMeshDimension() );
}
constexpr static int getStencilSize()
@@ -86,4 +84,3 @@ class GridEntityCrossStencilStorage
} // namespace Meshes
} // namespace TNL
-
diff --git a/src/TNL/Meshes/MeshDetails/CMakeLists.txt b/src/TNL/Meshes/MeshDetails/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Meshes/MeshDetails/config/CMakeLists.txt b/src/TNL/Meshes/MeshDetails/config/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Meshes/MeshDetails/layers/CMakeLists.txt b/src/TNL/Meshes/MeshDetails/layers/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Meshes/MeshDetails/traits/CMakeLists.txt b/src/TNL/Meshes/MeshDetails/traits/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Meshes/Topologies/CMakeLists.txt b/src/TNL/Meshes/Topologies/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Meshes/Traverser.h b/src/TNL/Meshes/Traverser.h
index 206227e9b360a2beaf5fccb8ce4ec26aadec375e..c9c647cd7314d800c4a56739945284530945b575 100644
--- a/src/TNL/Meshes/Traverser.h
+++ b/src/TNL/Meshes/Traverser.h
@@ -15,7 +15,7 @@ namespace Meshes {
template< typename Mesh,
typename MeshEntity,
- int EntitiesDimension = MeshEntity::entityDimension >
+ int EntitiesDimension = MeshEntity::getEntityDimension() >
class Traverser{};
} // namespace Meshes
@@ -23,4 +23,4 @@ class Traverser{};
#include <TNL/Meshes/GridDetails/Traverser_Grid1D.h>
#include <TNL/Meshes/GridDetails/Traverser_Grid2D.h>
-#include <TNL/Meshes/GridDetails/Traverser_Grid3D.h>
\ No newline at end of file
+#include <TNL/Meshes/GridDetails/Traverser_Grid3D.h>
diff --git a/src/TNL/Object.cpp b/src/TNL/Object.cpp
index 14418d26da1412d8f3785fce4f0c76fb0391e948..b4f364139de556ce410526f1bf94e1e12b4d8e41 100644
--- a/src/TNL/Object.cpp
+++ b/src/TNL/Object.cpp
@@ -46,11 +46,7 @@ String Object :: getSerializationTypeVirtual() const
bool Object :: save( File& file ) const
{
-#ifdef HAVE_NOT_CXX11
- if( ! file. write< const char, Devices::Host, int >( magic_number, strlen( magic_number ) ) )
-#else
if( ! file. write( magic_number, strlen( magic_number ) ) )
-#endif
return false;
if( ! this->getSerializationTypeVirtual().save( file ) ) return false;
return true;
@@ -80,55 +76,34 @@ bool Object :: boundLoad( File& file )
bool Object :: save( const String& fileName ) const
{
File file;
- if( ! file. open( fileName, tnlWriteMode ) )
+ if( ! file. open( fileName, IOMode::write ) )
{
- std::cerr << "I am not bale to open the file " << fileName << " for writing." << std::endl;
+ std::cerr << "I am not able to open the file " << fileName << " for writing." << std::endl;
return false;
}
- if( ! this->save( file ) )
- return false;
- if( ! file. close() )
- {
- std::cerr << "An error occurred when I was closing the file " << fileName << "." << std::endl;
- return false;
- }
- return true;
+ return this->save( file );
}
bool Object :: load( const String& fileName )
{
File file;
- if( ! file. open( fileName, tnlReadMode ) )
+ if( ! file. open( fileName, IOMode::read ) )
{
- std::cerr << "I am not bale to open the file " << fileName << " for reading." << std::endl;
+ std::cerr << "I am not able to open the file " << fileName << " for reading." << std::endl;
return false;
}
- if( ! this->load( file ) )
- return false;
- if( ! file. close() )
- {
- std::cerr << "An error occurred when I was closing the file " << fileName << "." << std::endl;
- return false;
- }
- return true;
+ return this->load( file );
}
bool Object :: boundLoad( const String& fileName )
{
File file;
- if( ! file. open( fileName, tnlReadMode ) )
+ if( ! file. open( fileName, IOMode::read ) )
{
- std::cerr << "I am not bale to open the file " << fileName << " for reading." << std::endl;
+ std::cerr << "I am not able to open the file " << fileName << " for reading." << std::endl;
return false;
}
- if( ! this->boundLoad( file ) )
- return false;
- if( ! file. close() )
- {
- std::cerr << "An error occurred when I was closing the file " << fileName << "." << std::endl;
- return false;
- }
- return true;
+ return this->boundLoad( file );
}
void Object::setDeprecatedReadMode()
@@ -140,32 +115,34 @@ void Object::setDeprecatedReadMode()
bool getObjectType( File& file, String& type )
{
char mn[ 10 ];
-#ifdef HAVE_NOT_CXX11
- if( ! file. read< char, Devices::Host, int >( mn, strlen( magic_number ) ) )
-#else
if( ! file. read( mn, strlen( magic_number ) ) )
-#endif
{
std::cerr << "Unable to read file " << file. getFileName() << " ... " << std::endl;
return false;
}
if( strncmp( mn, magic_number, 5 ) != 0 &&
- strncmp( mn, "SIM33", 5 ) != 0 ) return false;
- if( ! type. load( file ) ) return false;
+ strncmp( mn, "SIM33", 5 ) != 0 )
+ {
+ std::cout << "Not a TNL file (wrong magic number)." << std::endl;
+ return false;
+ }
+ if( ! type. load( file ) )
+ {
+ std::cerr << "Cannot load the object type." << std::endl;
+ return false;
+ }
return true;
}
bool getObjectType( const String& fileName, String& type )
{
File binaryFile;
- if( ! binaryFile. open( fileName, tnlReadMode ) )
+ if( ! binaryFile. open( fileName, IOMode::read ) )
{
std::cerr << "I am not able to open the file " << fileName << " for detecting the object inside!" << std::endl;
return false;
}
- bool ret_val = getObjectType( binaryFile, type );
- binaryFile. close();
- return ret_val;
+ return getObjectType( binaryFile, type );
}
bool parseObjectType( const String& objectType,
diff --git a/src/TNL/Object.h b/src/TNL/Object.h
index ed0ac90de971773bf52b01bb8715b8f43e425908..932361354be5778e661e140ed2dfa72145520447 100644
--- a/src/TNL/Object.h
+++ b/src/TNL/Object.h
@@ -10,7 +10,7 @@
#pragma once
-#include <TNL/Devices/Cuda.h>
+#include <TNL/Devices/CudaCallable.h>
#include <TNL/String.h>
#include <TNL/File.h>
#include <TNL/Containers/List.h>
@@ -33,8 +33,10 @@ class Object
public:
//! Basic constructor
+#ifndef HAVE_MIC
__cuda_callable__
Object() : deprecatedReadMode( false ) {};
+#endif
/****
* Type getter. This returns the type in C++ style - for example the returned value
@@ -75,8 +77,9 @@ class Object
// FIXME: __cuda_callable__ would have to be added to every overriding destructor,
// even if the object's constructor is not __cuda_callable__
// __cuda_callable__
+#ifndef HAVE_MIC
virtual ~Object(){};
-
+#endif
protected:
diff --git a/src/TNL/Operators/Advection/CMakeLists.txt b/src/TNL/Operators/Advection/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Operators/Advection/LaxFridrichs.h b/src/TNL/Operators/Advection/LaxFridrichs.h
index f72e7fbd3802805adb556dfc338df92811a3c5e5..9dff03402a5bfef768e8ca608bd8dbbd10a38998 100644
--- a/src/TNL/Operators/Advection/LaxFridrichs.h
+++ b/src/TNL/Operators/Advection/LaxFridrichs.h
@@ -89,14 +89,14 @@ class LaxFridrichs< Meshes::Grid< 1, MeshReal, Device, MeshIndex >, Real, Index,
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 1, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 1, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 1, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
typedef Functions::FunctionAdapter< MeshType, VelocityFunctionType > FunctionAdapter;
return ( 0.5 / this->tau ) * this->artificialViscosity * ( u[ west ]- 2.0 * u[ center ] + u[ east ] ) -
FunctionAdapter::getValue( this->velocityField.template getData< DeviceType >()[ 0 ], entity, time ) * ( u[ east ] - u[ west ] ) * hxInverse * 0.5;
@@ -181,18 +181,18 @@ class LaxFridrichs< Meshes::Grid< 2, MeshReal, Device, MeshIndex >, Real, Index,
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 2, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 2, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 2, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
typedef Functions::FunctionAdapter< MeshType, VelocityFunctionType > FunctionAdapter;
return ( 0.25 / this->tau ) * this->artificialViscosity * ( u[ west ] + u[ east ] + u[ north ] + u[ south ] - 4.0 * u[ center ] ) -
@@ -279,20 +279,20 @@ class LaxFridrichs< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, Real, Index,
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- static_assert( MeshEntity::entityDimension == 3, "Wrong mesh entity dimensions." );
+ static_assert( MeshEntity::getEntityDimension() == 3, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == 3, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxInverse = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
const RealType& hyInverse = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const RealType& hzInverse = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
typedef Functions::FunctionAdapter< MeshType, VelocityFunctionType > FunctionAdapter;
return ( 0.25 / this->tau ) * this->artificialViscosity * ( u[ west ] + u[ east ] + u[ north ] + u[ south ] + u[ up ] + u[ down ] - 6.0 * u[ center ] ) -
diff --git a/src/TNL/Operators/Analytic/CMakeLists.txt b/src/TNL/Operators/Analytic/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Operators/Analytic/Identity.h b/src/TNL/Operators/Analytic/Identity.h
index b8eb3f9a92901e8384370de6f49b66c226954570..b0c21a57ea0fd49b35067c6614488bb8e7177931 100644
--- a/src/TNL/Operators/Analytic/Identity.h
+++ b/src/TNL/Operators/Analytic/Identity.h
@@ -52,11 +52,11 @@ class Identity : public Functions::Domain< Dimensions, Functions::SpaceDomain >
const PointType& vertex,
const RealType& time = 0 ) const
{
- return function.getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
+ return function.template getPartialDerivative< XDiffOrder, YDiffOrder, ZDiffOrder >( vertex, time );
}
};
} // namespace Analytic
} // namespace Operators
-} // namespace TNL
\ No newline at end of file
+} // namespace TNL
diff --git a/src/TNL/Operators/CMakeLists.txt b/src/TNL/Operators/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Operators/DirichletBoundaryConditions.h b/src/TNL/Operators/DirichletBoundaryConditions.h
index bd2d54739250d80bf493aace2d22e267cf11adc6..759db071cf8af223abedc6899e1787f45ab76ee3 100644
--- a/src/TNL/Operators/DirichletBoundaryConditions.h
+++ b/src/TNL/Operators/DirichletBoundaryConditions.h
@@ -22,7 +22,7 @@ template< typename Mesh,
typename Function = Functions::Analytic::Constant< Mesh::getMeshDimension(), typename Mesh::RealType >,
int MeshEntitiesDimension = Mesh::getMeshDimension(),
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class DirichletBoundaryConditions
: public Operator< Mesh,
Functions::MeshBoundaryDomain,
@@ -43,7 +43,7 @@ class DirichletBoundaryConditions
typedef Containers::Vector< RealType, DeviceType, IndexType> DofVectorType;
typedef typename MeshType::PointType PointType;
- static constexpr int getDimension() { return MeshType::meshDimension; }
+ static constexpr int getMeshDimension() { return MeshType::getMeshDimension(); }
static void configSetup( Config::ConfigDescription& config,
const String& prefix = "" )
diff --git a/src/TNL/Operators/FiniteDifferences.h b/src/TNL/Operators/FiniteDifferences.h
index 36b8a3948bc9c3b1534258d741eb03d193dcf940..24ca76f82ee9009d6dc7e98f62b6a96340130f59 100644
--- a/src/TNL/Operators/FiniteDifferences.h
+++ b/src/TNL/Operators/FiniteDifferences.h
@@ -33,18 +33,6 @@ class FiniteDifferences< Meshes::Grid< 1, Real, Device, Index > >
typedef typename GridType::Cell CellType;
-#ifdef HAVE_NOT_CXX11
- template< typename GridFunction,
- int XDifferenceOrder,
- int YDifferenceOrder,
- int ZDifferenceOrder,
- int XDifferenceDirection,
- int YDifferenceDirection,
- int ZDifferenceDirection >
- static RealType getDifference( const GridType& grid,
- const GridFunction& inFunction,
- GridFunction& outFunction );
-#else
template< typename GridFunction,
int XDifferenceOrder,
int YDifferenceOrder,
@@ -55,20 +43,7 @@ class FiniteDifferences< Meshes::Grid< 1, Real, Device, Index > >
static RealType getDifference( const GridType& grid,
const GridFunction& inFunction,
GridFunction& outFunction );
-#endif
-#ifdef HAVE_NOT_CXX11
- template< typename GridFunction,
- int XDifferenceOrder,
- int YDifferenceOrder,
- int ZDifferenceOrder,
- int XDifferenceDirection,
- int YDifferenceDirection,
- int ZDifferenceDirection >
- static RealType getDifference( const GridType& grid,
- const CellType& cell,
- const GridFunction& function );
-#else
template< typename GridFunction,
int XDifferenceOrder,
int YDifferenceOrder,
@@ -79,8 +54,6 @@ class FiniteDifferences< Meshes::Grid< 1, Real, Device, Index > >
static RealType getDifference( const GridType& grid,
const CellType& cell,
const GridFunction& function );
-#endif
-
};
template< typename Real, typename Device, typename Index >
@@ -96,18 +69,6 @@ class FiniteDifferences< Meshes::Grid< 2, Real, Device, Index > >
typedef typename GridType::Cell CellType;
-#ifdef HAVE_NOT_CXX11
- template< typename GridFunction,
- int XDifferenceOrder,
- int YDifferenceOrder,
- int ZDifferenceOrder,
- int XDifferenceDirection,
- int YDifferenceDirection,
- int ZDifferenceDirection >
- static RealType getDifference( const GridType& grid,
- const GridFunction& inFunction,
- GridFunction& outFunction );
-#else
template< typename GridFunction,
int XDifferenceOrder,
int YDifferenceOrder,
@@ -118,20 +79,7 @@ class FiniteDifferences< Meshes::Grid< 2, Real, Device, Index > >
static RealType getDifference( const GridType& grid,
const GridFunction& inFunction,
GridFunction& outFunction );
-#endif
-#ifdef HAVE_NOT_CXX11
- template< typename GridFunction,
- int XDifferenceOrder,
- int YDifferenceOrder,
- int ZDifferenceOrder,
- int XDifferenceDirection,
- int YDifferenceDirection,
- int ZDifferenceDirection >
- static RealType getDifference( const GridType& grid,
- const CellType& cell,
- const GridFunction& function );
-#else
template< typename GridFunction,
int XDifferenceOrder,
int YDifferenceOrder,
@@ -142,8 +90,6 @@ class FiniteDifferences< Meshes::Grid< 2, Real, Device, Index > >
static RealType getDifference( const GridType& grid,
const CellType& cell,
const GridFunction& function );
-#endif
-
};
template< typename Real, typename Device, typename Index >
@@ -158,18 +104,6 @@ class FiniteDifferences< Meshes::Grid< 3, Real, Device, Index > >
//typedef typename GridType::CoordinatesType CoordinatesType;
typedef typename GridType::Cell CellType;
-#ifdef HAVE_NOT_CXX11
- template< typename GridFunction,
- int XDifferenceOrder,
- int YDifferenceOrder,
- int ZDifferenceOrder,
- int XDifferenceDirection,
- int YDifferenceDirection,
- int ZDifferenceDirection >
- static RealType getDifference( const GridType& grid,
- const GridFunction& inFunction,
- GridFunction& outFunction );
-#else
template< typename GridFunction,
int XDifferenceOrder,
int YDifferenceOrder,
@@ -180,20 +114,7 @@ class FiniteDifferences< Meshes::Grid< 3, Real, Device, Index > >
static RealType getDifference( const GridType& grid,
const GridFunction& inFunction,
GridFunction& outFunction );
-#endif
-#ifdef HAVE_NOT_CXX11
- template< typename GridFunction,
- int XDifferenceOrder,
- int YDifferenceOrder,
- int ZDifferenceOrder,
- int XDifferenceDirection,
- int YDifferenceDirection,
- int ZDifferenceDirection >
- static RealType getDifference( const GridType& grid,
- const CellType& cell,
- const GridFunction& function );
-#else
template< typename GridFunction,
int XDifferenceOrder,
int YDifferenceOrder,
@@ -204,11 +125,9 @@ class FiniteDifferences< Meshes::Grid< 3, Real, Device, Index > >
static RealType getDifference( const GridType& grid,
const CellType& cell,
const GridFunction& function );
-#endif
-
};
} // namespace Operators
} // namespace TNL
-#include <TNL/Operators/FiniteDifferences_impl.h>
\ No newline at end of file
+#include <TNL/Operators/FiniteDifferences_impl.h>
diff --git a/src/TNL/Operators/FiniteDifferences_impl.h b/src/TNL/Operators/FiniteDifferences_impl.h
index 51cc742999d7fc40bdb2e5fd75363ecc3ec6a335..3d5b5bace6fa1cd9d9566581fbe113ebba06437d 100644
--- a/src/TNL/Operators/FiniteDifferences_impl.h
+++ b/src/TNL/Operators/FiniteDifferences_impl.h
@@ -69,22 +69,22 @@ Real FiniteDifferences< Meshes::Grid< 1, Real, Device, Index > >::getDifference(
if( YDifferenceOrder > 0 || ZDifferenceOrder > 0 )
return 0.0;
const RealType hx = grid.getSpaceSteps().x();
- auto neighbourEntities = cell.getNeighbourEntities();
+ auto neighborEntities = cell.getNeighborEntities();
IndexType cellIndex = grid.getEntityIndex( cell );
if( XDifferenceOrder == 1 )
{
if( XDifferenceDirection == 0 )
- return ( function[ neighbourEntities.template getEntityIndex< 1 >() ] -
- function[ neighbourEntities.template getEntityIndex< -1 >() ] ) / ( 2.0 * hx );
+ return ( function[ neighborEntities.template getEntityIndex< 1 >() ] -
+ function[ neighborEntities.template getEntityIndex< -1 >() ] ) / ( 2.0 * hx );
else
- return ( function[ neighbourEntities.template getEntityIndex< XDifferenceDirection >() ] -
+ return ( function[ neighborEntities.template getEntityIndex< XDifferenceDirection >() ] -
function[ cellIndex ] ) / ( XDifferenceDirection * hx );
}
if( XDifferenceOrder == 2 )
{
- return ( function[ neighbourEntities.template getEntityIndex< 1 >() ] -
+ return ( function[ neighborEntities.template getEntityIndex< 1 >() ] -
2.0 * function[ cellIndex ] +
- function[ neighbourEntities.template getEntityIndex< -1 >() ] ) / ( hx * hx );
+ function[ neighborEntities.template getEntityIndex< -1 >() ] ) / ( hx * hx );
}
}
@@ -121,33 +121,33 @@ Real FiniteDifferences< Meshes::Grid< 2, Real, Device, Index > >::getDifference(
{
if( ZDifferenceOrder > 0 )
return 0.0;
- auto neighbourEntities = cell.getNeighbourEntities();
+ auto neighborEntities = cell.getNeighborEntities();
IndexType cellIndex = grid.getEntityIndex( cell );
if( XDifferenceOrder == 1 )
{
const RealType hx = grid.getSpaceSteps().x();
- return ( function[ neighbourEntities.template getEntityIndex< XDifferenceDirection, 0 >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< XDifferenceDirection, 0 >( cellIndex ) ] -
function[ cellIndex ] ) / ( XDifferenceDirection * hx );
}
if( XDifferenceOrder == 2 )
{
const RealType hx = grid.getSpaceSteps().x();
- return ( function[ neighbourEntities.template getEntityIndex< 1, 0 >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< 1, 0 >( cellIndex ) ] -
2.0 * function[ cellIndex ] +
- function[ neighbourEntities.template getEntityIndex< -1, 0 >( cellIndex ) ] ) / ( hx * hx );
+ function[ neighborEntities.template getEntityIndex< -1, 0 >( cellIndex ) ] ) / ( hx * hx );
}
if( YDifferenceOrder == 1 )
{
const RealType hy = grid.getSpaceSteps().y();
- return ( function[ neighbourEntities.template getEntityIndex< 0, YDifferenceDirection >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< 0, YDifferenceDirection >( cellIndex ) ] -
function[ cellIndex ] ) / ( YDifferenceDirection * hy );
}
if( YDifferenceOrder == 2 )
{
const RealType hy = grid.getSpaceSteps().y();
- return ( function[ neighbourEntities.template getEntityIndex< 0, 1 >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< 0, 1 >( cellIndex ) ] -
2.0 * function[ cellIndex ] +
- function[ neighbourEntities.template getEntityIndex< 0, -1 >( cellIndex ) ] ) / ( hy * hy );
+ function[ neighborEntities.template getEntityIndex< 0, -1 >( cellIndex ) ] ) / ( hy * hy );
}
@@ -184,47 +184,47 @@ Real FiniteDifferences< Meshes::Grid< 3, Real, Device, Index > >::getDifference(
const CellType& cell,
const GridFunction& function )
{
- auto neighbourEntities = cell.getNeighbourEntities();
+ auto neighborEntities = cell.getNeighborEntities();
IndexType cellIndex = grid.getEntityIndex( cell );
if( XDifferenceOrder == 1 )
{
const RealType hx = grid.getSpaceSteps().x();
- return ( function[ neighbourEntities.template getEntityIndex< XDifferenceDirection, 0, 0 >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< XDifferenceDirection, 0, 0 >( cellIndex ) ] -
function[ cellIndex ] ) / ( XDifferenceDirection * hx );
}
if( XDifferenceOrder == 2 )
{
const RealType hx = grid.getSpaceSteps().x();
- return ( function[ neighbourEntities.template getEntityIndex< 1, 0, 0 >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< 1, 0, 0 >( cellIndex ) ] -
2.0 * function[ cellIndex ] +
- function[ neighbourEntities.template getEntityIndex< -1, 0, 0 >( cellIndex ) ] ) / ( hx * hx );
+ function[ neighborEntities.template getEntityIndex< -1, 0, 0 >( cellIndex ) ] ) / ( hx * hx );
}
if( YDifferenceOrder == 1 )
{
const RealType hy = grid.getSpaceSteps().y();
- return ( function[ neighbourEntities.template getEntityIndex< 0, YDifferenceDirection, 0 >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< 0, YDifferenceDirection, 0 >( cellIndex ) ] -
function[ cellIndex ] ) / ( YDifferenceDirection * hy );
}
if( YDifferenceOrder == 2 )
{
const RealType hy = grid.getSpaceSteps().y();
- return ( function[ neighbourEntities.template getEntityIndex< 0, 1, 0 >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< 0, 1, 0 >( cellIndex ) ] -
2.0 * function[ cellIndex ] +
- function[ neighbourEntities.template getEntityIndex< 0, -1, 0 >( cellIndex ) ] ) / ( hy * hy );
+ function[ neighborEntities.template getEntityIndex< 0, -1, 0 >( cellIndex ) ] ) / ( hy * hy );
}
if( ZDifferenceOrder == 1 )
{
const RealType hz = grid.getSpaceSteps().z();
- return ( function[ neighbourEntities.template getEntityIndex< 0, 0, ZDifferenceDirection >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< 0, 0, ZDifferenceDirection >( cellIndex ) ] -
function[ cellIndex ] ) / ( ZDifferenceDirection * hz );
}
if( ZDifferenceOrder == 2 )
{
const RealType hz = grid.getSpaceSteps().z();
- return ( function[ neighbourEntities.template getEntityIndex< 0, 0, 1 >( cellIndex ) ] -
+ return ( function[ neighborEntities.template getEntityIndex< 0, 0, 1 >( cellIndex ) ] -
2.0 * function[ cellIndex ] +
- function[ neighbourEntities.template getEntityIndex< 0, 0, -1 >( cellIndex ) ] ) / ( hz * hz );
+ function[ neighborEntities.template getEntityIndex< 0, 0, -1 >( cellIndex ) ] ) / ( hz * hz );
}
diff --git a/src/TNL/Operators/NeumannBoundaryConditions.h b/src/TNL/Operators/NeumannBoundaryConditions.h
index aa39235068ea28f07258eea3e136aaba379f1e71..15cfbe85c68e8e986e8d50dc8f1c83c24c36fb7b 100644
--- a/src/TNL/Operators/NeumannBoundaryConditions.h
+++ b/src/TNL/Operators/NeumannBoundaryConditions.h
@@ -19,7 +19,7 @@ namespace Operators {
template< typename Mesh,
typename Function,
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class NeumannBoundaryConditions
{
@@ -121,13 +121,13 @@ class NeumannBoundaryConditions< Meshes::Grid< 1, MeshReal, Device, MeshIndex >,
const RealType& time = 0 ) const
{
const MeshType& mesh = entity.getMesh();
- const auto& neighbourEntities = entity.getNeighbourEntities();
+ const auto& neighborEntities = entity.getNeighborEntities();
const IndexType& index = entity.getIndex();
if( entity.getCoordinates().x() == 0 )
- return u[ neighbourEntities.template getEntityIndex< 1 >() ] + entity.getMesh().getSpaceSteps().x() *
+ return u[ neighborEntities.template getEntityIndex< 1 >() ] + entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
else
- return u[ neighbourEntities.template getEntityIndex< -1 >() ] + entity.getMesh().getSpaceSteps().x() *
+ return u[ neighborEntities.template getEntityIndex< -1 >() ] + entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
@@ -154,19 +154,19 @@ class NeumannBoundaryConditions< Meshes::Grid< 1, MeshReal, Device, MeshIndex >,
Matrix& matrix,
Vector& b ) const
{
- const auto& neighbourEntities = entity.getNeighbourEntities();
+ const auto& neighborEntities = entity.getNeighborEntities();
const IndexType& index = entity.getIndex();
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
if( entity.getCoordinates().x() == 0 )
{
matrixRow.setElement( 0, index, 1.0 );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< 1 >(), -1.0 );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< 1 >(), -1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
else
{
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< -1 >(), -1.0 );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< -1 >(), -1.0 );
matrixRow.setElement( 1, index, 1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
@@ -215,27 +215,27 @@ class NeumannBoundaryConditions< Meshes::Grid< 2, MeshReal, Device, MeshIndex >,
const RealType& time = 0 ) const
{
const MeshType& mesh = entity.getMesh();
- const auto& neighbourEntities = entity.getNeighbourEntities();
+ const auto& neighborEntities = entity.getNeighborEntities();
const IndexType& index = entity.getIndex();
if( entity.getCoordinates().x() == 0 )
{
- return u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] + entity.getMesh().getSpaceSteps().x() *
+ return u[ neighborEntities.template getEntityIndex< 1, 0 >() ] + entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().x() == entity.getMesh().getDimensions().x() - 1 )
{
- return u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] + entity.getMesh().getSpaceSteps().x() *
+ return u[ neighborEntities.template getEntityIndex< -1, 0 >() ] + entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().y() == 0 )
{
- return u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] + entity.getMesh().getSpaceSteps().y() *
+ return u[ neighborEntities.template getEntityIndex< 0, 1 >() ] + entity.getMesh().getSpaceSteps().y() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
// The following line is commented to avoid compiler warning
//if( entity.getCoordinates().y() == entity.getMesh().getDimensions().y() - 1 )
{
- return u[ neighbourEntities.template getEntityIndex< 0, -1 >() ] + entity.getMesh().getSpaceSteps().y() *
+ return u[ neighborEntities.template getEntityIndex< 0, -1 >() ] + entity.getMesh().getSpaceSteps().y() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
}
@@ -261,19 +261,19 @@ class NeumannBoundaryConditions< Meshes::Grid< 2, MeshReal, Device, MeshIndex >,
Matrix& matrix,
Vector& b ) const
{
- const auto& neighbourEntities = entity.getNeighbourEntities();
+ const auto& neighborEntities = entity.getNeighborEntities();
const IndexType& index = entity.getIndex();
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
if( entity.getCoordinates().x() == 0 )
{
matrixRow.setElement( 0, index, 1.0 );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< 1, 0 >(), -1.0 );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< 1, 0 >(), -1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().x() == entity.getMesh().getDimensions().x() - 1 )
{
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< -1, 0 >(), -1.0 );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< -1, 0 >(), -1.0 );
matrixRow.setElement( 1, index, 1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
@@ -281,13 +281,13 @@ class NeumannBoundaryConditions< Meshes::Grid< 2, MeshReal, Device, MeshIndex >,
if( entity.getCoordinates().y() == 0 )
{
matrixRow.setElement( 0, index, 1.0 );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< 0, 1 >(), -1.0 );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< 0, 1 >(), -1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().y() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().y() == entity.getMesh().getDimensions().y() - 1 )
{
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< 0, -1 >(), -1.0 );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< 0, -1 >(), -1.0 );
matrixRow.setElement( 1, index, 1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().y() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
@@ -334,37 +334,37 @@ class NeumannBoundaryConditions< Meshes::Grid< 3, MeshReal, Device, MeshIndex >,
const RealType& time = 0 ) const
{
const MeshType& mesh = entity.getMesh();
- const auto& neighbourEntities = entity.getNeighbourEntities();
+ const auto& neighborEntities = entity.getNeighborEntities();
const IndexType& index = entity.getIndex();
if( entity.getCoordinates().x() == 0 )
{
- return u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] + entity.getMesh().getSpaceSteps().x() *
+ return u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] + entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().x() == entity.getMesh().getDimensions().x() - 1 )
{
- return u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] + entity.getMesh().getSpaceSteps().x() *
+ return u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] + entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().y() == 0 )
{
- return u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] + entity.getMesh().getSpaceSteps().y() *
+ return u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] + entity.getMesh().getSpaceSteps().y() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().y() == entity.getMesh().getDimensions().y() - 1 )
{
- return u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] + entity.getMesh().getSpaceSteps().y() *
+ return u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] + entity.getMesh().getSpaceSteps().y() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().z() == 0 )
{
- return u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] + entity.getMesh().getSpaceSteps().z() *
+ return u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] + entity.getMesh().getSpaceSteps().z() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
// The following line is commented to avoid compiler warning
//if( entity.getCoordinates().z() == entity.getMesh().getDimensions().z() - 1 )
{
- return u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] + entity.getMesh().getSpaceSteps().z() *
+ return u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] + entity.getMesh().getSpaceSteps().z() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
}
@@ -391,19 +391,19 @@ class NeumannBoundaryConditions< Meshes::Grid< 3, MeshReal, Device, MeshIndex >,
Matrix& matrix,
Vector& b ) const
{
- const auto& neighbourEntities = entity.getNeighbourEntities();
+ const auto& neighborEntities = entity.getNeighborEntities();
const IndexType& index = entity.getIndex();
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
if( entity.getCoordinates().x() == 0 )
{
matrixRow.setElement( 0, index, 1.0 );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< 1, 0, 0 >(), -1.0 );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< 1, 0, 0 >(), -1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().x() == entity.getMesh().getDimensions().x() - 1 )
{
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< -1, 0, 0 >(), -1.0 );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< -1, 0, 0 >(), -1.0 );
matrixRow.setElement( 1, index, 1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().x() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
@@ -411,13 +411,13 @@ class NeumannBoundaryConditions< Meshes::Grid< 3, MeshReal, Device, MeshIndex >,
if( entity.getCoordinates().y() == 0 )
{
matrixRow.setElement( 0, index, 1.0 );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< 0, 1, 0 >(), -1.0 );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< 0, 1, 0 >(), -1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().y() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().y() == entity.getMesh().getDimensions().y() - 1 )
{
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< 0, -1, 0 >(), -1.0 );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< 0, -1, 0 >(), -1.0 );
matrixRow.setElement( 1, index, 1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().y() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
@@ -425,13 +425,13 @@ class NeumannBoundaryConditions< Meshes::Grid< 3, MeshReal, Device, MeshIndex >,
if( entity.getCoordinates().z() == 0 )
{
matrixRow.setElement( 0, index, 1.0 );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< 0, 0, 1 >(), -1.0 );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< 0, 0, 1 >(), -1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().z() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
}
if( entity.getCoordinates().z() == entity.getMesh().getDimensions().z() - 1 )
{
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< 0, 0, -1 >(), -1.0 );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< 0, 0, -1 >(), -1.0 );
matrixRow.setElement( 1, index, 1.0 );
b[ index ] = entity.getMesh().getSpaceSteps().z() *
Functions::FunctionAdapter< MeshType, FunctionType >::getValue( this->function, entity, time );
diff --git a/src/TNL/Operators/Operator.h b/src/TNL/Operators/Operator.h
index b18e31fc4fbc0b453840733871afdbcd360afc8f..1a78fd5970a73650781f01fd593f40563797a432 100644
--- a/src/TNL/Operators/Operator.h
+++ b/src/TNL/Operators/Operator.h
@@ -20,7 +20,7 @@ template< typename Mesh,
int PreimageEntitiesDimension = Mesh::getMeshDimension(),
int ImageEntitiesDimension = Mesh::getMeshDimension(),
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class Operator : public Functions::Domain< Mesh::getMeshDimension(), DomainType >
{
public:
@@ -28,12 +28,12 @@ class Operator : public Functions::Domain< Mesh::getMeshDimension(), DomainType
typedef Mesh MeshType;
typedef typename MeshType::RealType MeshRealType;
typedef typename MeshType::DeviceType DeviceType;
- typedef typename MeshType::IndexType MeshIndexType;
+ typedef typename MeshType::GlobalIndexType MeshIndexType;
typedef Real RealType;
typedef Index IndexType;
typedef void ExactOperatorType;
- constexpr static int getDimension() { return MeshType::getMeshDimension(); }
+ constexpr static int getMeshDimension() { return MeshType::getMeshDimension(); }
constexpr static int getPreimageEntitiesDimension() { return PreimageEntitiesDimension; }
constexpr static int getImageEntitiesDimension() { return ImageEntitiesDimension; }
diff --git a/src/TNL/Operators/diffusion/CMakeLists.txt b/src/TNL/Operators/diffusion/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Operators/diffusion/FiniteVolumeNonlinearOperator.h b/src/TNL/Operators/diffusion/FiniteVolumeNonlinearOperator.h
index 93e5d48e95acc6b9a66378d1f449791925900ec0..efb17555547b53da1b850b789b6956674c1abbdf 100644
--- a/src/TNL/Operators/diffusion/FiniteVolumeNonlinearOperator.h
+++ b/src/TNL/Operators/diffusion/FiniteVolumeNonlinearOperator.h
@@ -25,7 +25,7 @@ template< typename Mesh,
typename NonlinearDiffusionOperator,
typename OperatorQ,
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class FiniteVolumeNonlinearOperator
{
diff --git a/src/TNL/Operators/diffusion/FiniteVolumeNonlinearOperator_impl.h b/src/TNL/Operators/diffusion/FiniteVolumeNonlinearOperator_impl.h
index 7242115f36088d30eb65d192ac18ec98e24474a5..083160467875cc0e4f40b15c63b7cf59c222a68b 100644
--- a/src/TNL/Operators/diffusion/FiniteVolumeNonlinearOperator_impl.h
+++ b/src/TNL/Operators/diffusion/FiniteVolumeNonlinearOperator_impl.h
@@ -130,14 +130,14 @@ operator()( const MeshEntity& entity,
const Vector& u,
const Real& time ) const
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const IndexType& cellIndex = entity.getIndex();
return operatorQ( entity, u, time ) *
- ( ( u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< -2, 0 >() / operatorQ.operator()( entity, u, time, 1 )
- + ( u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< 0, -2 >() / operatorQ.operator()( entity, u, time, 0, 1 )
- - ( -u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] + u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< -2, 0 >() / operatorQ.operator()( entity, u, time, -1)
- - ( -u[ neighbourEntities.template getEntityIndex< 0,-1 >() ] + u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< 0, -2 >() / operatorQ.operator()( entity, u, time, 0, -1) );
+ ( ( u[ neighborEntities.template getEntityIndex< 1, 0 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< -2, 0 >() / operatorQ.operator()( entity, u, time, 1 )
+ + ( u[ neighborEntities.template getEntityIndex< 0, 1 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< 0, -2 >() / operatorQ.operator()( entity, u, time, 0, 1 )
+ - ( -u[ neighborEntities.template getEntityIndex< -1, 0 >() ] + u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< -2, 0 >() / operatorQ.operator()( entity, u, time, -1)
+ - ( -u[ neighborEntities.template getEntityIndex< 0,-1 >() ] + u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< 0, -2 >() / operatorQ.operator()( entity, u, time, 0, -1) );
}
template< typename MeshReal,
@@ -180,7 +180,7 @@ setMatrixElements( const RealType& time,
Matrix& matrix ) const
{
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType aCoef = - tau * operatorQ.operator()( entity, u, time ) * mesh.template getSpaceStepsProducts< 0, -2 >() /
operatorQ.operator()( entity, u, time, 0, -1 );
const RealType bCoef = - tau * operatorQ.operator()( entity, u, time ) * mesh.template getSpaceStepsProducts< -2, 0 >() /
@@ -194,11 +194,11 @@ setMatrixElements( const RealType& time,
operatorQ.operator()( entity, u, time, 1 );
const RealType eCoef = - tau * operatorQ.operator()( entity, u, time ) * mesh.template getSpaceStepsProducts< 0, -2 >() /
operatorQ.operator()( entity, u, time, 0, 1 );
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< 0, -1 >(), aCoef );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< -1, 0 >(), bCoef );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< 0, -1 >(), aCoef );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< -1, 0 >(), bCoef );
matrixRow.setElement( 2, entity.getIndex(), cCoef );
- matrixRow.setElement( 3, neighbourEntities.template getEntityIndex< 1, 0 >(), dCoef );
- matrixRow.setElement( 4, neighbourEntities.template getEntityIndex< 0, 1 >(), eCoef );
+ matrixRow.setElement( 3, neighborEntities.template getEntityIndex< 1, 0 >(), dCoef );
+ matrixRow.setElement( 4, neighborEntities.template getEntityIndex< 0, 1 >(), eCoef );
}
template< typename MeshReal,
@@ -233,21 +233,21 @@ operator()( const MeshEntity& entity,
const Vector& u,
const Real& time ) const
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const IndexType& cellIndex = entity.getIndex();
return operatorQ( entity, u, time ) *
- ( (u[ neighbourEntities.template getEntityIndex< 1,0,0 >() ] - u[ cellIndex ])
+ ( (u[ neighborEntities.template getEntityIndex< 1,0,0 >() ] - u[ cellIndex ])
* mesh.template getSpaceStepsProducts< -2, 0, 0 >() / operatorQ( entity, u, time, 1 )
- + ( u[ neighbourEntities.template getEntityIndex< 0,1,0 >() ] - u[ cellIndex ]) * mesh.template getSpaceStepsProducts< 0, -2, 0 >()/
+ + ( u[ neighborEntities.template getEntityIndex< 0,1,0 >() ] - u[ cellIndex ]) * mesh.template getSpaceStepsProducts< 0, -2, 0 >()/
operatorQ( entity, u, time, 0, 1 )
- + ( u[ neighbourEntities.template getEntityIndex< 0,0,1 >() ] - u[ cellIndex ]) * mesh.template getSpaceStepsProducts< 0, 0, -2 >()/
+ + ( u[ neighborEntities.template getEntityIndex< 0,0,1 >() ] - u[ cellIndex ]) * mesh.template getSpaceStepsProducts< 0, 0, -2 >()/
operatorQ( entity, u, time, 0, 0, 1 )
- - ( - u[ neighbourEntities.template getEntityIndex< -1,0,0 >() ] + u[ cellIndex ])
+ - ( - u[ neighborEntities.template getEntityIndex< -1,0,0 >() ] + u[ cellIndex ])
* mesh.template getSpaceStepsProducts< -2, 0, 0 >() / operatorQ( entity, u, time, -1)
- -( - u[ neighbourEntities.template getEntityIndex< 0,-1,0 >() ] + u[ cellIndex ]) * mesh.template getSpaceStepsProducts< 0, -2, 0 >()
+ -( - u[ neighborEntities.template getEntityIndex< 0,-1,0 >() ] + u[ cellIndex ]) * mesh.template getSpaceStepsProducts< 0, -2, 0 >()
/operatorQ( entity, u, time, 0, -1)
- -( - u[ neighbourEntities.template getEntityIndex< 0,0,-1 >() ] + u[ cellIndex ]) * mesh.template getSpaceStepsProducts< 0, 0, -2 >()
+ -( - u[ neighborEntities.template getEntityIndex< 0,0,-1 >() ] + u[ cellIndex ]) * mesh.template getSpaceStepsProducts< 0, 0, -2 >()
/operatorQ( entity, u, time, 0, 0, -1) );
}
@@ -293,7 +293,7 @@ setMatrixElements( const RealType& time,
Matrix& matrix ) const
{
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType aCoef = - tau * operatorQ( entity, u, time ) *
mesh.template getSpaceStepsProducts< 0, 0, -2 >() / operatorQ.operator()( entity, u, time, 0, 0, -1 );
const RealType bCoef = - tau * operatorQ( entity, u, time ) *
@@ -313,13 +313,13 @@ setMatrixElements( const RealType& time,
mesh.template getSpaceStepsProducts< 0, -2, 0 >() / operatorQ.operator()( entity, u, time, 0, 1, 0 );
const RealType gCoef = - tau * operatorQ.operator()( entity, u, time ) *
mesh.template getSpaceStepsProducts< 0, 0, -2 >() / operatorQ.operator()( entity, u, time, 0, 0, 1 );
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< 0,0,-1 >(), aCoef );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< 0,-1,0 >(), bCoef );
- matrixRow.setElement( 2, neighbourEntities.template getEntityIndex< -1,0,0 >(), cCoef );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< 0,0,-1 >(), aCoef );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< 0,-1,0 >(), bCoef );
+ matrixRow.setElement( 2, neighborEntities.template getEntityIndex< -1,0,0 >(), cCoef );
matrixRow.setElement( 3, entity.getIndex(), dCoef );
- matrixRow.setElement( 4, neighbourEntities.template getEntityIndex< 1,0,0 >(), eCoef );
- matrixRow.setElement( 5, neighbourEntities.template getEntityIndex< 0,1,0 >(), fCoef );
- matrixRow.setElement( 6, neighbourEntities.template getEntityIndex< 0,0,1 >(), gCoef );
+ matrixRow.setElement( 4, neighborEntities.template getEntityIndex< 1,0,0 >(), eCoef );
+ matrixRow.setElement( 5, neighborEntities.template getEntityIndex< 0,1,0 >(), fCoef );
+ matrixRow.setElement( 6, neighborEntities.template getEntityIndex< 0,0,1 >(), gCoef );
}
} // namespace Operators
diff --git a/src/TNL/Operators/diffusion/LinearDiffusion.h b/src/TNL/Operators/diffusion/LinearDiffusion.h
index 8844e547ad9f8b9f53f12b47eddb08caf8d30f7c..ff9c05f03238eb5e10869f4a1dc2d8a6cd2d821a 100644
--- a/src/TNL/Operators/diffusion/LinearDiffusion.h
+++ b/src/TNL/Operators/diffusion/LinearDiffusion.h
@@ -27,7 +27,7 @@ namespace Operators {
template< typename Mesh,
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class LinearDiffusion
{
@@ -52,7 +52,7 @@ class LinearDiffusion< Meshes::Grid< 1,MeshReal, Device, MeshIndex >, Real, Inde
typedef Index IndexType;
typedef ExactLinearDiffusion< 1 > ExactOperatorType;
- static const int Dimension = MeshType::meshDimension;
+ static const int Dimension = MeshType::getMeshDimension();
static constexpr int getDimension() { return Dimension; }
@@ -103,7 +103,7 @@ class LinearDiffusion< Meshes::Grid< 2, MeshReal, Device, MeshIndex >, Real, Ind
typedef Index IndexType;
typedef ExactLinearDiffusion< 2 > ExactOperatorType;
- static const int Dimension = MeshType::meshDimension;
+ static const int Dimension = MeshType::getMeshDimension();
static constexpr int getDimension() { return Dimension; }
@@ -153,7 +153,7 @@ class LinearDiffusion< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, Real, Ind
typedef Index IndexType;
typedef ExactLinearDiffusion< 3 > ExactOperatorType;
- static const int Dimension = MeshType::meshDimension;
+ static const int Dimension = MeshType::getMeshDimension();
static constexpr int getDimension() { return Dimension; }
diff --git a/src/TNL/Operators/diffusion/LinearDiffusion_impl.h b/src/TNL/Operators/diffusion/LinearDiffusion_impl.h
index c8bc8a3d9b9f30254d7c4247ac356715543b2c02..83a20829ccc4f46a56eb80b1e474990db23856da 100644
--- a/src/TNL/Operators/diffusion/LinearDiffusion_impl.h
+++ b/src/TNL/Operators/diffusion/LinearDiffusion_impl.h
@@ -52,13 +52,13 @@ operator()( const PreimageFunction& u,
const MeshEntity& entity,
const Real& time ) const
{
- static_assert( MeshEntity::entityDimension == 1, "Wrong mesh entity dimension." );
+ static_assert( MeshEntity::getEntityDimension() == 1, "Wrong mesh entity dimensions." );
static_assert( PreimageFunction::getEntitiesDimension() == 1, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxSquareInverse = entity.getMesh().template getSpaceStepsProducts< - 2 >();
- return ( u[ neighbourEntities.template getEntityIndex< -1 >() ]
+ return ( u[ neighborEntities.template getEntityIndex< -1 >() ]
- 2.0 * u[ entity.getIndex() ]
- + u[ neighbourEntities.template getEntityIndex< 1 >() ] ) * hxSquareInverse;
+ + u[ neighborEntities.template getEntityIndex< 1 >() ] ) * hxSquareInverse;
}
template< typename MeshReal,
@@ -98,15 +98,15 @@ setMatrixElements( const PreimageFunction& u,
Matrix& matrix,
Vector& b ) const
{
- static_assert( MeshEntity::entityDimension == 1, "Wrong mesh entity dimension." );
+ static_assert( MeshEntity::getEntityDimension() == 1, "Wrong mesh entity dimensions." );
static_assert( PreimageFunction::getEntitiesDimension() == 1, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const IndexType& index = entity.getIndex();
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
const RealType lambdaX = tau * entity.getMesh().template getSpaceStepsProducts< -2 >();
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< -1 >(), - lambdaX );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< -1 >(), - lambdaX );
matrixRow.setElement( 1, index, 2.0 * lambdaX );
- matrixRow.setElement( 2, neighbourEntities.template getEntityIndex< 1 >(), - lambdaX );
+ matrixRow.setElement( 2, neighborEntities.template getEntityIndex< 1 >(), - lambdaX );
}
template< typename MeshReal,
@@ -157,15 +157,15 @@ operator()( const PreimageFunction& u,
const EntityType& entity,
const Real& time ) const
{
- static_assert( EntityType::entityDimension == 2, "Wrong mesh entity dimension." );
+ static_assert( EntityType::getEntityDimension() == 2, "Wrong mesh entity dimensions." );
static_assert( PreimageFunction::getEntitiesDimension() == 2, "Wrong preimage function" );
- const typename EntityType::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename EntityType::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxSquareInverse = entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const RealType& hySquareInverse = entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
- return ( u[ neighbourEntities.template getEntityIndex< -1, 0 >() ]
- + u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] ) * hxSquareInverse +
- ( u[ neighbourEntities.template getEntityIndex< 0, -1 >() ]
- + u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] ) * hySquareInverse
+ return ( u[ neighborEntities.template getEntityIndex< -1, 0 >() ]
+ + u[ neighborEntities.template getEntityIndex< 1, 0 >() ] ) * hxSquareInverse +
+ ( u[ neighborEntities.template getEntityIndex< 0, -1 >() ]
+ + u[ neighborEntities.template getEntityIndex< 0, 1 >() ] ) * hySquareInverse
- 2.0 * u[ entity.getIndex() ] * ( hxSquareInverse + hySquareInverse );
}
@@ -189,18 +189,18 @@ setMatrixElements( const PreimageFunction& u,
Matrix& matrix,
Vector& b ) const
{
- static_assert( MeshEntity::entityDimension == 2, "Wrong mesh entity dimension." );
+ static_assert( MeshEntity::getEntityDimension() == 2, "Wrong mesh entity dimensions." );
static_assert( PreimageFunction::getEntitiesDimension() == 2, "Wrong preimage function" );
const IndexType& index = entity.getIndex();
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
const RealType lambdaX = tau * entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const RealType lambdaY = tau * entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< 0, -1 >(), -lambdaY );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< -1, 0 >(), -lambdaX );
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< 0, -1 >(), -lambdaY );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< -1, 0 >(), -lambdaX );
matrixRow.setElement( 2, index, 2.0 * ( lambdaX + lambdaY ) );
- matrixRow.setElement( 3, neighbourEntities.template getEntityIndex< 1, 0 >(), -lambdaX );
- matrixRow.setElement( 4, neighbourEntities.template getEntityIndex< 0, 1 >(), -lambdaY );
+ matrixRow.setElement( 3, neighborEntities.template getEntityIndex< 1, 0 >(), -lambdaX );
+ matrixRow.setElement( 4, neighborEntities.template getEntityIndex< 0, 1 >(), -lambdaY );
}
@@ -234,18 +234,18 @@ operator()( const PreimageFunction& u,
const EntityType& entity,
const Real& time ) const
{
- static_assert( EntityType::entityDimension == 3, "Wrong mesh entity dimension." );
+ static_assert( EntityType::getEntityDimension() == 3, "Wrong mesh entity dimensions." );
static_assert( PreimageFunction::getEntitiesDimension() == 3, "Wrong preimage function" );
- const typename EntityType::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename EntityType::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxSquareInverse = entity.getMesh().template getSpaceStepsProducts< -2, 0, 0 >();
const RealType& hySquareInverse = entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const RealType& hzSquareInverse = entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
- return ( u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ]
- + u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] ) * hxSquareInverse +
- ( u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ]
- + u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] ) * hySquareInverse +
- ( u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ]
- + u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] ) * hzSquareInverse
+ return ( u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ]
+ + u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] ) * hxSquareInverse +
+ ( u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ]
+ + u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] ) * hySquareInverse +
+ ( u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ]
+ + u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] ) * hzSquareInverse
- 2.0 * u[ entity.getIndex() ] * ( hxSquareInverse + hySquareInverse + hzSquareInverse );
}
@@ -286,21 +286,21 @@ setMatrixElements( const PreimageFunction& u,
Matrix& matrix,
Vector& b ) const
{
- static_assert( MeshEntity::entityDimension == 3, "Wrong mesh entity dimension." );
+ static_assert( MeshEntity::getEntityDimension() == 3, "Wrong mesh entity dimensions." );
static_assert( PreimageFunction::getEntitiesDimension() == 3, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const IndexType& index = entity.getIndex();
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
const RealType lambdaX = tau * entity.getMesh().template getSpaceStepsProducts< -2, 0, 0 >();
const RealType lambdaY = tau * entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const RealType lambdaZ = tau * entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
- matrixRow.setElement( 0, neighbourEntities.template getEntityIndex< 0, 0, -1 >(), -lambdaZ );
- matrixRow.setElement( 1, neighbourEntities.template getEntityIndex< 0, -1, 0 >(), -lambdaY );
- matrixRow.setElement( 2, neighbourEntities.template getEntityIndex< -1, 0, 0 >(), -lambdaX );
+ matrixRow.setElement( 0, neighborEntities.template getEntityIndex< 0, 0, -1 >(), -lambdaZ );
+ matrixRow.setElement( 1, neighborEntities.template getEntityIndex< 0, -1, 0 >(), -lambdaY );
+ matrixRow.setElement( 2, neighborEntities.template getEntityIndex< -1, 0, 0 >(), -lambdaX );
matrixRow.setElement( 3, index, 2.0 * ( lambdaX + lambdaY + lambdaZ ) );
- matrixRow.setElement( 4, neighbourEntities.template getEntityIndex< 1, 0, 0 >(), -lambdaX );
- matrixRow.setElement( 5, neighbourEntities.template getEntityIndex< 0, 1, 0 >(), -lambdaY );
- matrixRow.setElement( 6, neighbourEntities.template getEntityIndex< 0, 0, 1 >(), -lambdaZ );
+ matrixRow.setElement( 4, neighborEntities.template getEntityIndex< 1, 0, 0 >(), -lambdaX );
+ matrixRow.setElement( 5, neighborEntities.template getEntityIndex< 0, 1, 0 >(), -lambdaY );
+ matrixRow.setElement( 6, neighborEntities.template getEntityIndex< 0, 0, 1 >(), -lambdaZ );
}
} // namespace Operators
diff --git a/src/TNL/Operators/diffusion/OneSidedMeanCurvature.h b/src/TNL/Operators/diffusion/OneSidedMeanCurvature.h
index 5e3e4d59c2aa98d09d8cbee72a105bf2dcfb37c6..91f9848cbc9bc63faca5538913f1ec4c9a1701b0 100644
--- a/src/TNL/Operators/diffusion/OneSidedMeanCurvature.h
+++ b/src/TNL/Operators/diffusion/OneSidedMeanCurvature.h
@@ -30,7 +30,7 @@ namespace Operators {
template< typename Mesh,
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType,
+ typename Index = typename Mesh::GlobalIndexType,
bool EvaluateNonlinearityOnFly = false >
class OneSidedMeanCurvature
: public Operator< Mesh, Functions::MeshInteriorDomain, Mesh::getMeshDimension(), Mesh::getMeshDimension(), Real, Index >
diff --git a/src/TNL/Operators/diffusion/OneSidedNonlinearDiffusion.h b/src/TNL/Operators/diffusion/OneSidedNonlinearDiffusion.h
index 606ab6a05ddca3474b9bfa7ed535e9d89380dbb7..b74abe0b06de5898e86679dcd537d9656cbfcc9b 100644
--- a/src/TNL/Operators/diffusion/OneSidedNonlinearDiffusion.h
+++ b/src/TNL/Operators/diffusion/OneSidedNonlinearDiffusion.h
@@ -26,7 +26,7 @@ namespace Operators {
template< typename Mesh,
typename Nonlinearity,
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class OneSidedNonlinearDiffusion
{
};
@@ -69,12 +69,12 @@ class OneSidedNonlinearDiffusion< Meshes::Grid< 1,MeshReal, Device, MeshIndex >,
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const RealType& hx_div = entity.getMesh().template getSpaceStepsProducts< -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
const RealType& u_c = u[ center ];
const RealType u_x_f = ( u[ east ] - u_c );
const RealType u_x_b = ( u_c - u[ west ] );
@@ -105,10 +105,10 @@ class OneSidedNonlinearDiffusion< Meshes::Grid< 1,MeshReal, Device, MeshIndex >,
Vector& b ) const
{
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
const RealType lambda_x = tau * entity.getMesh().template getSpaceStepsProducts< -2 >();
const RealType& nonlinearity_center = this->nonlinearity[ center ];
const RealType& nonlinearity_west = this->nonlinearity[ west ];
@@ -163,15 +163,15 @@ class OneSidedNonlinearDiffusion< Meshes::Grid< 2, MeshReal, Device, MeshIndex >
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const RealType& hx_div = entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const RealType& hy_div = entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
const RealType& u_c = u[ center ];
const RealType u_x_f = ( u[ east ] - u_c );
const RealType u_x_b = ( u_c - u[ west ] );
@@ -205,12 +205,12 @@ class OneSidedNonlinearDiffusion< Meshes::Grid< 2, MeshReal, Device, MeshIndex >
Vector& b ) const
{
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
const RealType lambda_x = tau * entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const RealType lambda_y = tau * entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
const RealType& nonlinearity_center = this->nonlinearity[ center ];
@@ -272,18 +272,18 @@ class OneSidedNonlinearDiffusion< Meshes::Grid< 3, MeshReal, Device, MeshIndex >
const MeshEntity& entity,
const RealType& time = 0.0 ) const
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const RealType& hx_div = entity.getMesh().template getSpaceStepsProducts< -2, 0, 0 >();
const RealType& hy_div = entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const RealType& hz_div = entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
const RealType& u_c = u[ center ];
const RealType u_x_f = ( u[ east ] - u_c );
@@ -322,14 +322,14 @@ class OneSidedNonlinearDiffusion< Meshes::Grid< 3, MeshReal, Device, MeshIndex >
Vector& b ) const
{
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
const RealType lambda_x = tau * entity.getMesh().template getSpaceStepsProducts< -2, 0, 0 >();
diff --git a/src/TNL/Operators/euler/CMakeLists.txt b/src/TNL/Operators/euler/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Operators/euler/fvm/CMakeLists.txt b/src/TNL/Operators/euler/fvm/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Operators/euler/fvm/LaxFridrichs_impl.h b/src/TNL/Operators/euler/fvm/LaxFridrichs_impl.h
index d0c650b0a1bf3d4f1d9a90952a7548f2b9700506..b2cc2ff06d267933fbacb78be213a24a8496ef5f 100644
--- a/src/TNL/Operators/euler/fvm/LaxFridrichs_impl.h
+++ b/src/TNL/Operators/euler/fvm/LaxFridrichs_impl.h
@@ -147,14 +147,14 @@ void LaxFridrichs< Meshes::Grid< 2, Real, Device, Index, GridGeometry >, Pressur
RealType& rho_u2_t,
const RealType& tau ) const
{
- TNL_ASSERT( mesh, std::cerr << "No mesh has been binded with the Lax-Fridrichs scheme." );
- TNL_ASSERT( pressureGradient, std::cerr << "No pressure gradient was set in the the Lax-Fridrichs scheme." )
+ TNL_ASSERT_TRUE( mesh, "No mesh has been binded with the Lax-Fridrichs scheme." );
+ TNL_ASSERT_TRUE( pressureGradient, "No pressure gradient was set in the the Lax-Fridrichs scheme." )
const IndexType& c = centralVolume;
- const IndexType e = this->mesh -> getElementNeighbour( centralVolume, 1, 0 );
- const IndexType w = this->mesh -> getElementNeighbour( centralVolume, -1, 0 );
- const IndexType n = this->mesh -> getElementNeighbour( centralVolume, 0, 1 );
- const IndexType s = this->mesh -> getElementNeighbour( centralVolume, 0, -1 );
+ const IndexType e = this->mesh -> getElementNeighbor( centralVolume, 1, 0 );
+ const IndexType w = this->mesh -> getElementNeighbor( centralVolume, -1, 0 );
+ const IndexType n = this->mesh -> getElementNeighbor( centralVolume, 0, 1 );
+ const IndexType s = this->mesh -> getElementNeighbor( centralVolume, 0, -1 );
const RealType u1_e = rho_u1[ e ] / regularize( rho[ e ] );
const RealType u1_w = rho_u1[ w ] / regularize( rho[ w ] );
@@ -168,7 +168,7 @@ void LaxFridrichs< Meshes::Grid< 2, Real, Device, Index, GridGeometry >, Pressur
const RealType u2_w = rho_u2[ w ] / regularize( rho[ w ] );
/****
- * Get the central volume and its neighbours (east, north, west, south) coordinates
+ * Get the central volume and its neighbors (east, north, west, south) coordinates
*/
CoordinatesType c_coordinates, e_coordinates, n_coordinates, w_coordinates, s_coordinates;
this->mesh -> getElementCoordinates( c, c_coordinates );
@@ -398,8 +398,8 @@ void LaxFridrichs< Meshes::Grid< 2, Real, Device, Index, tnlIdenticalGridGeometr
RealType& rho_u2_t,
const RealType& tau ) const
{
- TNL_ASSERT( mesh, std::cerr << "No mesh has been binded with the Lax-Fridrichs scheme." );
- TNL_ASSERT( pressureGradient, std::cerr << "No pressure gradient was set in the the Lax-Fridrichs scheme." )
+ TNL_ASSERT_TRUE( mesh, "No mesh has been binded with the Lax-Fridrichs scheme." );
+ TNL_ASSERT_TRUE( pressureGradient, "No pressure gradient was set in the the Lax-Fridrichs scheme." )
const IndexType& xSize = this->mesh -> getDimensions(). x();
const IndexType& ySize = this->mesh -> getDimensions(). y();
@@ -407,10 +407,10 @@ void LaxFridrichs< Meshes::Grid< 2, Real, Device, Index, tnlIdenticalGridGeometr
const RealType hy = this->mesh -> getParametricStep(). y();
const IndexType& c = centralVolume;
- const IndexType e = this->mesh -> getElementNeighbour( centralVolume, 1, 0 );
- const IndexType w = this->mesh -> getElementNeighbour( centralVolume, -1, 0 );
- const IndexType n = this->mesh -> getElementNeighbour( centralVolume, 0, 1 );
- const IndexType s = this->mesh -> getElementNeighbour( centralVolume, 0, -1 );
+ const IndexType e = this->mesh -> getElementNeighbor( centralVolume, 1, 0 );
+ const IndexType w = this->mesh -> getElementNeighbor( centralVolume, -1, 0 );
+ const IndexType n = this->mesh -> getElementNeighbor( centralVolume, 0, 1 );
+ const IndexType s = this->mesh -> getElementNeighbor( centralVolume, 0, -1 );
/****
* rho_t + ( rho u_1 )_x + ( rho u_2 )_y = 0
@@ -456,8 +456,8 @@ void LaxFridrichs< Meshes::Grid< 2, Real, Device, Index, tnlIdenticalGridGeometr
RealType& e_t,
const RealType& tau ) const
{
- TNL_ASSERT( mesh, std::cerr << "No mesh has been binded with the Lax-Fridrichs scheme." );
- TNL_ASSERT( pressureGradient, std::cerr << "No pressure gradient was set in the the Lax-Fridrichs scheme." )
+ TNL_ASSERT_TRUE( mesh, "No mesh has been binded with the Lax-Fridrichs scheme." );
+ TNL_ASSERT_TRUE( pressureGradient, "No pressure gradient was set in the the Lax-Fridrichs scheme." )
const IndexType& xSize = this->mesh -> getDimensions(). x();
const IndexType& ySize = this->mesh -> getDimensions(). y();
@@ -465,10 +465,10 @@ void LaxFridrichs< Meshes::Grid< 2, Real, Device, Index, tnlIdenticalGridGeometr
const RealType hy = this->mesh -> getParametricStep(). y();
const IndexType& c = centralVolume;
- const IndexType e = this->mesh -> getElementNeighbour( centralVolume, 1, 0 );
- const IndexType w = this->mesh -> getElementNeighbour( centralVolume, -1, 0 );
- const IndexType n = this->mesh -> getElementNeighbour( centralVolume, 0, 1 );
- const IndexType s = this->mesh -> getElementNeighbour( centralVolume, 0, -1 );
+ const IndexType e = this->mesh -> getElementNeighbor( centralVolume, 1, 0 );
+ const IndexType w = this->mesh -> getElementNeighbor( centralVolume, -1, 0 );
+ const IndexType n = this->mesh -> getElementNeighbor( centralVolume, 0, 1 );
+ const IndexType s = this->mesh -> getElementNeighbor( centralVolume, 0, -1 );
/****
* rho_t + ( rho u_1 )_x + ( rho u_2 )_y = 0
diff --git a/src/TNL/Operators/fdm/BackwardFiniteDifference.h b/src/TNL/Operators/fdm/BackwardFiniteDifference.h
index ecc94c6a4e34bdae4eb61b856a2513edf25da033..d816499f1e733c4a6cbca7f1a3dc1c0fefe3d320 100644
--- a/src/TNL/Operators/fdm/BackwardFiniteDifference.h
+++ b/src/TNL/Operators/fdm/BackwardFiniteDifference.h
@@ -22,7 +22,7 @@ template< typename Mesh,
int YDifference = 0,
int ZDifference = 0,
typename RealType = typename Mesh::RealType,
- typename IndexType = typename Mesh::IndexType >
+ typename IndexType = typename Mesh::GlobalIndexType >
class BackwardFiniteDifference
{
};
diff --git a/src/TNL/Operators/fdm/CentralFiniteDifference.h b/src/TNL/Operators/fdm/CentralFiniteDifference.h
index a9514d9f3e9354505d72aa675f6cdc0a76a16382..90c9a5887897dd2664e2936514bb9255089c9e2c 100644
--- a/src/TNL/Operators/fdm/CentralFiniteDifference.h
+++ b/src/TNL/Operators/fdm/CentralFiniteDifference.h
@@ -22,7 +22,7 @@ template< typename Mesh,
int YDifference = 0,
int ZDifference = 0,
typename RealType = typename Mesh::RealType,
- typename IndexType = typename Mesh::IndexType >
+ typename IndexType = typename Mesh::GlobalIndexType >
class CentralFiniteDifference
{
};
diff --git a/src/TNL/Operators/fdm/FiniteDifferences_1D.h b/src/TNL/Operators/fdm/FiniteDifferences_1D.h
index 127d454d159ff058ce5e3d7ec5afd7e6b820f228..622e4c4585169179df679ca352610d06dd2ec4e1 100644
--- a/src/TNL/Operators/fdm/FiniteDifferences_1D.h
+++ b/src/TNL/Operators/fdm/FiniteDifferences_1D.h
@@ -66,10 +66,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const Real& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 1 >()] - u_c ) * hxDiv;
+ return ( u[ neighborEntities.template getEntityIndex< 1 >()] - u_c ) * hxDiv;
}
};
@@ -93,10 +93,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const Real& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u_c - u[ neighbourEntities.template getEntityIndex< -1 >()] ) * hxDiv;
+ return ( u_c - u[ neighborEntities.template getEntityIndex< -1 >()] ) * hxDiv;
}
};
@@ -120,10 +120,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const Real& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1 >();
- return ( u[ neighbourEntities.template getEntityIndex< 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1 >() ] ) * ( 0.5 * hxDiv );
+ return ( u[ neighborEntities.template getEntityIndex< 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1 >() ] ) * ( 0.5 * hxDiv );
}
};
@@ -147,12 +147,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< -2 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 2 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 2 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 1 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 1 >() ] ) * hxSquareDiv;
}
};
@@ -173,12 +173,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< -2 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< -2 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< -2 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< -1 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< -1 >() ] ) * hxSquareDiv;
}
};
@@ -199,12 +199,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< -2 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 1 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 1 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< -1 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< -1 >() ] ) * hxSquareDiv;
}
};
diff --git a/src/TNL/Operators/fdm/FiniteDifferences_2D.h b/src/TNL/Operators/fdm/FiniteDifferences_2D.h
index d287b50d210d36abdbe9d373bc4b1567aa70968c..4ea67b93159038745e137e0c6decc17c234e720e 100644
--- a/src/TNL/Operators/fdm/FiniteDifferences_2D.h
+++ b/src/TNL/Operators/fdm/FiniteDifferences_2D.h
@@ -66,10 +66,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 1, 0 >()] - u_c ) * hxDiv;
+ return ( u[ neighborEntities.template getEntityIndex< 1, 0 >()] - u_c ) * hxDiv;
}
};
@@ -90,10 +90,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hyDiv = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 1 >()] - u_c ) * hyDiv;
+ return ( u[ neighborEntities.template getEntityIndex< 0, 1 >()] - u_c ) * hyDiv;
}
};
@@ -117,10 +117,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u_c - u[ neighbourEntities.template getEntityIndex< -1, 0 >()] ) * hxDiv;
+ return ( u_c - u[ neighborEntities.template getEntityIndex< -1, 0 >()] ) * hxDiv;
}
};
@@ -141,10 +141,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hyDiv = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u_c - u[ neighbourEntities.template getEntityIndex< 0, -1 >()] ) * hyDiv;
+ return ( u_c - u[ neighborEntities.template getEntityIndex< 0, -1 >()] ) * hyDiv;
}
};
@@ -168,10 +168,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
- return ( u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] ) * ( 0.5 * hxDiv );
+ return ( u[ neighborEntities.template getEntityIndex< 1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0 >() ] ) * ( 0.5 * hxDiv );
}
};
@@ -192,10 +192,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hyDiv = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
- return ( u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1 >() ] ) * ( 0.5 * hyDiv );
+ return ( u[ neighborEntities.template getEntityIndex< 0, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1 >() ] ) * ( 0.5 * hyDiv );
}
};
@@ -220,12 +220,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< -2,0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 2, 0 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 2, 0 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 1, 0 >() ] ) * hxSquareDiv;
}
};
@@ -246,12 +246,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< -2, 0 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< -2, 0 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< -1, 0 >() ] ) * hxSquareDiv;
}
};
@@ -272,12 +272,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 1, 0 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< -1, 0 >() ] ) * hxSquareDiv;
}
};
@@ -298,12 +298,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 2 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 0, 2 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 0, 1 >() ] ) * hxSquareDiv;
}
};
@@ -324,12 +324,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, -2 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 0, -2 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 0, -1 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 0, -1 >() ] ) * hxSquareDiv;
}
};
@@ -351,12 +351,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const Real& hySquareDiv = entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 0, 1 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 0, -1 >() ] ) * hySquareDiv;
+ u[ neighborEntities.template getEntityIndex< 0, -1 >() ] ) * hySquareDiv;
}
};
diff --git a/src/TNL/Operators/fdm/FiniteDifferences_3D.h b/src/TNL/Operators/fdm/FiniteDifferences_3D.h
index 569a1c7645e4bf5904875ac208dab390c601050d..895aeaddac1be89b57b308becf80b27cd9576c4d 100644
--- a/src/TNL/Operators/fdm/FiniteDifferences_3D.h
+++ b/src/TNL/Operators/fdm/FiniteDifferences_3D.h
@@ -33,10 +33,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >()] - u_c ) * hxDiv;
+ return ( u[ neighborEntities.template getEntityIndex< 1, 0, 0 >()] - u_c ) * hxDiv;
}
};
@@ -57,10 +57,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hyDiv = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >()] - u_c ) * hyDiv;
+ return ( u[ neighborEntities.template getEntityIndex< 0, 1, 0 >()] - u_c ) * hyDiv;
}
};
@@ -81,10 +81,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hzDiv = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >()] - u_c ) * hzDiv;
+ return ( u[ neighborEntities.template getEntityIndex< 0, 0, 1 >()] - u_c ) * hzDiv;
}
};
@@ -108,10 +108,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u_c - u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >()] ) * hxDiv;
+ return ( u_c - u[ neighborEntities.template getEntityIndex< -1, 0, 0 >()] ) * hxDiv;
}
};
@@ -132,10 +132,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hyDiv = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u_c - u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >()] ) * hyDiv;
+ return ( u_c - u[ neighborEntities.template getEntityIndex< 0, -1, 0 >()] ) * hyDiv;
}
};
@@ -156,10 +156,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hzDiv = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u_c - u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >()] ) * hzDiv;
+ return ( u_c - u[ neighborEntities.template getEntityIndex< 0, 0, -1 >()] ) * hzDiv;
}
};
@@ -183,10 +183,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
- return ( u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] ) * ( 0.5 * hxDiv );
+ return ( u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] ) * ( 0.5 * hxDiv );
}
};
@@ -207,10 +207,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hyDiv = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
- return ( u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] ) * ( 0.5 * hyDiv );
+ return ( u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] ) * ( 0.5 * hyDiv );
}
};
@@ -231,10 +231,10 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hzDiv = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
- return ( u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] ) * ( 0.5 * hzDiv );
+ return ( u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] ) * ( 0.5 * hzDiv );
}
};
@@ -258,12 +258,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< -2, 0, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 2, 0, 0 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 2, 0, 0 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] ) * hxSquareDiv;
}
};
@@ -284,12 +284,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< -2, 0, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< -2, 0, 0 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< -2, 0, 0 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] ) * hxSquareDiv;
}
};
@@ -310,12 +310,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< -2, 0, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] ) * hxSquareDiv;
}
};
@@ -336,12 +336,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 2, 0 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 0, 2, 0 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] ) * hxSquareDiv;
}
};
@@ -362,12 +362,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, -2, 0 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 0, -2, 0 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] ) * hxSquareDiv;
}
};
@@ -388,12 +388,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hySquareDiv = entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] ) * hySquareDiv;
+ u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] ) * hySquareDiv;
}
};
@@ -414,12 +414,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 0, 2 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 0, 0, 2 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] ) * hxSquareDiv;
}
};
@@ -440,12 +440,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hxSquareDiv = entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 0, -2 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 0, 0, -2 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] ) * hxSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] ) * hxSquareDiv;
}
};
@@ -466,12 +466,12 @@ class FiniteDifferences<
static Real getValue( const MeshFunction& u,
const MeshEntity& entity )
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const Real& hzSquareDiv = entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
const Real& u_c = u[ entity.getIndex() ];
- return ( u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] -
+ return ( u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] -
2.0 * u_c +
- u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] ) * hzSquareDiv;
+ u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] ) * hzSquareDiv;
}
};
diff --git a/src/TNL/Operators/fdm/ForwardFiniteDifference.h b/src/TNL/Operators/fdm/ForwardFiniteDifference.h
index 1de01680cd16c115aaaed34d95437537d3d67e88..9c75e05dadd03213494d1229fe814771ca0f540d 100644
--- a/src/TNL/Operators/fdm/ForwardFiniteDifference.h
+++ b/src/TNL/Operators/fdm/ForwardFiniteDifference.h
@@ -23,7 +23,7 @@ template< typename Mesh,
int YDifference = 0,
int ZDifference = 0,
typename RealType = typename Mesh::RealType,
- typename IndexType = typename Mesh::IndexType >
+ typename IndexType = typename Mesh::GlobalIndexType >
class ForwardFiniteDifference
{
};
diff --git a/src/TNL/Operators/geometric/CoFVMGradientNorm.h b/src/TNL/Operators/geometric/CoFVMGradientNorm.h
index 47f168656a1def8d48aa8ff56f6531964a4ef98f..0380ecc223f8b1556cf379bfb2dd64a1197ae576 100644
--- a/src/TNL/Operators/geometric/CoFVMGradientNorm.h
+++ b/src/TNL/Operators/geometric/CoFVMGradientNorm.h
@@ -22,7 +22,7 @@ namespace Operators {
template< typename Mesh,
int MeshEntityDimension = Mesh::getMeshDimension(),
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class CoFVMGradientNorm
{
};
@@ -118,11 +118,11 @@ class CoFVMGradientNorm< Meshes::Grid< 1,MeshReal, Device, MeshIndex >, 0, Real,
"The mesh function u must be stored on mesh cells.." );
static_assert( MeshEntity::getMeshDimension() == 0,
"The complementary finite volume gradient norm may be evaluated only on faces." );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.template getNeighbourEntities< 1 >();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.template getNeighborEntities< 1 >();
const RealType& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1 >();
- const RealType& u_x = ( u[ neighbourEntities.template getEntityIndex< 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1 >() ] ) * hxDiv;
+ const RealType& u_x = ( u[ neighborEntities.template getEntityIndex< 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1 >() ] ) * hxDiv;
return ::sqrt( this->epsSquare + ( u_x * u_x ) );
}
@@ -180,37 +180,37 @@ class CoFVMGradientNorm< Meshes::Grid< 2, MeshReal, Device, MeshIndex >, 1, Real
"The mesh function u must be stored on mesh cells.." );
static_assert( MeshEntity::getMeshDimension() == 1,
"The complementary finite volume gradient norm may be evaluated only on faces." );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.template getNeighbourEntities< 2 >();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.template getNeighborEntities< 2 >();
const RealType& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1, 0 >();
const RealType& hyDiv = entity.getMesh().template getSpaceStepsProducts< 0, -1 >();
if( entity.getOrientation().x() != 0.0 )
{
const RealType u_x =
- ( u[ neighbourEntities.template getEntityIndex< 1, 0 >()] -
- u[ neighbourEntities.template getEntityIndex< -1, 0 >()] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 0 >()] -
+ u[ neighborEntities.template getEntityIndex< -1, 0 >()] ) * hxDiv;
RealType u_y;
if( entity.getCoordinates().y() > 0 )
{
if( entity.getCoordinates().y() < entity.getMesh().getDimensions().y() - 1 )
u_y = 0.25 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, -1 >() ] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, -1 >() ] ) * hyDiv;
else // if( entity.getCoordinates().y() < entity.getMesh().getDimensions().y() - 1 )
u_y = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, -1 >() ] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, -1 >() ] ) * hyDiv;
}
else // if( entity.getCoordinates().y() > 0 )
{
u_y = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0 >() ] ) * hyDiv;
}
return ::sqrt( this->epsSquare + u_x * u_x + u_y * u_y );
}
@@ -219,28 +219,28 @@ class CoFVMGradientNorm< Meshes::Grid< 2, MeshReal, Device, MeshIndex >, 1, Real
{
if( entity.getCoordinates().x() < entity.getMesh().getDimensions().x() - 1 )
u_x = 0.25 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, -1 >() ] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, -1 >() ] ) * hxDiv;
else // if( entity.getCoordinates().x() < entity.getMesh().getDimensions().x() - 1 )
u_x = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, -1 >() ] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, -1 >() ] ) * hxDiv;
}
else // if( entity.getCoordinates().x() > 0 )
{
u_x = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1 >() ] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1 >() ] ) * hxDiv;
}
const RealType u_y =
- ( u[ neighbourEntities.template getEntityIndex< 0, 1 >()] -
- u[ neighbourEntities.template getEntityIndex< 0, -1 >()] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 1 >()] -
+ u[ neighborEntities.template getEntityIndex< 0, -1 >()] ) * hyDiv;
return ::sqrt( this->epsSquare + u_x * u_x + u_y * u_y );
}
@@ -296,43 +296,43 @@ class CoFVMGradientNorm< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, 2, Real
"The mesh function u must be stored on mesh cells.." );
static_assert( MeshEntity::getMeshDimension() == 2,
"The complementary finite volume gradient norm may be evaluated only on faces." );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.template getNeighbourEntities< 3 >();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.template getNeighborEntities< 3 >();
const RealType& hxDiv = entity.getMesh().template getSpaceStepsProducts< -1, 0, 0 >();
const RealType& hyDiv = entity.getMesh().template getSpaceStepsProducts< 0, -1, 0 >();
const RealType& hzDiv = entity.getMesh().template getSpaceStepsProducts< 0, 0, -1 >();
if( entity.getOrientation().x() != 0.0 )
{
const RealType u_x =
- ( u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >()] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >()] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 0, 0 >()] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, 0 >()] ) * hxDiv;
RealType u_y;
if( entity.getCoordinates().y() > 0 )
{
if( entity.getCoordinates().y() < entity.getMesh().getDimensions().y() - 1 )
{
u_y = 0.25 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, 1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1, -1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, -1, 0 >() ] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, 1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1, -1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, -1, 0 >() ] ) * hyDiv;
}
else // if( entity.getCoordinates().y() < entity.getMesh().getDimensions().y() - 1 )
{
u_y = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1, -1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, -1, 0 >() ] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1, -1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, -1, 0 >() ] ) * hyDiv;
}
}
else // if( entity.getCoordinates().y() > 0 )
{
u_y = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, 1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, 1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] ) * hyDiv;
}
RealType u_z;
@@ -341,27 +341,27 @@ class CoFVMGradientNorm< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, 2, Real
if( entity.getCoordinates().z() < entity.getMesh().getDimensions().z() - 1 )
{
u_z = 0.25 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 0, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, 0, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1, 0, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, -1 >() ] ) * hzDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 0, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, 0, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1, 0, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, -1 >() ] ) * hzDiv;
}
else //if( entity.getCoordinates().z() < entity.getMesh().getDimensions().z() - 1 )
{
u_z = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1, 0, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, -1 >() ] ) * hzDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1, 0, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, -1 >() ] ) * hzDiv;
}
}
else //if( entity.getCoordinates().z() > 0 )
{
u_z = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 0, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, 0, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] ) * hzDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 0, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, 0, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] ) * hzDiv;
}
return ::sqrt( this->epsSquare + u_x * u_x + u_y * u_y + u_z * u_z );
}
@@ -373,58 +373,58 @@ class CoFVMGradientNorm< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, 2, Real
if( entity.getCoordinates().x() < entity.getMesh().getDimensions().x() - 1 )
{
u_x = 0.25 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, -1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, -1, 0 >() ] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, -1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, -1, 0 >() ] ) * hxDiv;
}
else // if( entity.getCoordinates().x() < entity.getMesh().getDimensions().x() - 1 )
{
u_x = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, -1, 0 >() ] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, -1, 0 >() ] ) * hxDiv;
}
}
else // if( entity.getCoordinates().x() > 0 )
{
u_x = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, -1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, -1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] ) * hxDiv;
}
const RealType u_y =
- ( u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >()] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >()] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 1, 0 >()] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, 0 >()] ) * hyDiv;
RealType u_z;
if( entity.getCoordinates().z() > 0 )
{
if( entity.getCoordinates().z() < entity.getMesh().getDimensions().z() - 1 )
{
u_z = 0.25 *
- ( u[ neighbourEntities.template getEntityIndex< 0, 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, -1, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 1, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, -1 >() ] ) * hzDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, -1, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 1, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, -1 >() ] ) * hzDiv;
}
else // if( entity.getCoordinates().z() < entity.getMesh().getDimensions().z() - 1 )
{
u_z = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 1, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, -1 >() ] ) * hzDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 1, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, -1 >() ] ) * hzDiv;
}
}
else // if( entity.getCoordinates().z() > 0 )
{
u_z = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 0, 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, -1, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] ) * hzDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, -1, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] ) * hzDiv;
}
return ::sqrt( this->epsSquare + u_x * u_x + u_y * u_y + u_z * u_z );
}
@@ -434,28 +434,28 @@ class CoFVMGradientNorm< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, 2, Real
if( entity.getCoordinates().x() < entity.getMesh().getDimensions().x() - 1 )
{
u_x = 0.25 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 0, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, 0, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, -1 >() ] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 0, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, 0, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, -1 >() ] ) * hxDiv;
}
else // if( entity.getCoordinates().x() < entity.getMesh().getDimensions().x() - 1 )
{
u_x = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1, 0, -1 >() ] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1, 0, -1 >() ] ) * hxDiv;
}
}
else // if( entity.getCoordinates().x() > 0 )
{
u_x = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 1, 0, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, 0, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] ) * hxDiv;
+ ( u[ neighborEntities.template getEntityIndex< 1, 0, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, 0, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] ) * hxDiv;
}
RealType u_y;
if( entity.getCoordinates().y() > 0 )
@@ -463,31 +463,31 @@ class CoFVMGradientNorm< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, 2, Real
if( entity.getCoordinates().y() < entity.getMesh().getDimensions().y() - 1 )
{
u_y = 0.25 *
- ( u[ neighbourEntities.template getEntityIndex< 0, 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 1, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, -1 >() ] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 1, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, -1 >() ] ) * hyDiv;
}
else //if( entity.getCoordinates().y() < entity.getMesh().getDimensions().y() - 1 )
{
u_y = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, -1, -1 >() ] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, -1, -1 >() ] ) * hyDiv;
}
}
else //if( entity.getCoordinates().y() > 0 )
{
u_y = 0.5 *
- ( u[ neighbourEntities.template getEntityIndex< 0, 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 1, -1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] ) * hyDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 1, -1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] ) * hyDiv;
}
const RealType u_z =
- ( u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >()] -
- u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >()] ) * hzDiv;
+ ( u[ neighborEntities.template getEntityIndex< 0, 0, 1 >()] -
+ u[ neighborEntities.template getEntityIndex< 0, 0, -1 >()] ) * hzDiv;
return ::sqrt( this->epsSquare + u_x * u_x + u_y * u_y + u_z * u_z );
}
diff --git a/src/TNL/Operators/geometric/ExactGradientNorm.h b/src/TNL/Operators/geometric/ExactGradientNorm.h
index 48d67a12a8392f3838a5bbc19751325e0e651528..cf7e3384820be1c2e8628fe655211037c6dde1dd 100644
--- a/src/TNL/Operators/geometric/ExactGradientNorm.h
+++ b/src/TNL/Operators/geometric/ExactGradientNorm.h
@@ -10,8 +10,6 @@
#pragma once
-#include <TNL/Containers/Vector.h>
-#include <TNL/Containers/SharedVector.h>
#include <TNL/Meshes/Grid.h>
#include <TNL/Functions/Domain.h>
diff --git a/src/TNL/Operators/geometric/FDMGradientNorm.h b/src/TNL/Operators/geometric/FDMGradientNorm.h
index d574b90123835169e4e390d7a93c04f758f114ad..a5eb4536317a0ff5258a681585f67557ff029d59 100644
--- a/src/TNL/Operators/geometric/FDMGradientNorm.h
+++ b/src/TNL/Operators/geometric/FDMGradientNorm.h
@@ -20,7 +20,7 @@ namespace Operators {
template< typename Mesh,
template< typename, int, int, int, typename, typename > class DifferenceOperatorTemplate = ForwardFiniteDifference,
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class FDMGradientNorm
{
};
diff --git a/src/TNL/Operators/geometric/TwoSidedGradientNorm.h b/src/TNL/Operators/geometric/TwoSidedGradientNorm.h
index 9d78daa5da338ec2e435ec0afc8656e819a3920d..2d86167b1c1ee3a9466635c2605931248d80eb56 100644
--- a/src/TNL/Operators/geometric/TwoSidedGradientNorm.h
+++ b/src/TNL/Operators/geometric/TwoSidedGradientNorm.h
@@ -20,7 +20,7 @@ namespace Operators {
template< typename Mesh,
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class TwoSidedGradientNorm
{
};
diff --git a/src/TNL/Operators/interpolants/MeshEntitiesInterpolants.h b/src/TNL/Operators/interpolants/MeshEntitiesInterpolants.h
index bb171716806a69daca7a81f7c83ab321c00f6e40..ca9381d1f418e1a033d39c4619678b94856afc63 100644
--- a/src/TNL/Operators/interpolants/MeshEntitiesInterpolants.h
+++ b/src/TNL/Operators/interpolants/MeshEntitiesInterpolants.h
@@ -48,10 +48,10 @@ class MeshEntitiesInterpolants< Meshes::Grid< 1, Real, Device, Index >, 1, 0 >
static_assert( std::is_same< typename MeshEntity::MeshType, MeshType >::value,
"The mesh entity belongs to other mesh type then the interpolants." );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
- return 0.5 * ( u[ neighbourEntities.template getEntityIndex< -1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1 >() ] );
+ return 0.5 * ( u[ neighborEntities.template getEntityIndex< -1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1 >() ] );
}
};
@@ -80,10 +80,10 @@ class MeshEntitiesInterpolants< Meshes::Grid< 1, Real, Device, Index >, 0, 1 >
static_assert( std::is_same< typename MeshEntity::MeshType, MeshType >::value,
"The mesh entity belongs to other mesh type then the interpolants." );
- const typename MeshEntity::template NeighbourEntities< 0 >& neighbourEntities = entity.template getNeighbourEntities< 0 >();
+ const typename MeshEntity::template NeighborEntities< 0 >& neighborEntities = entity.template getNeighborEntities< 0 >();
- return 0.5 * ( u[ neighbourEntities.template getEntityIndex< -1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1 >() ] );
+ return 0.5 * ( u[ neighborEntities.template getEntityIndex< -1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1 >() ] );
}
};
@@ -112,14 +112,14 @@ class MeshEntitiesInterpolants< Meshes::Grid< 2, Real, Device, Index >, 2, 1 >
static_assert( std::is_same< typename MeshEntity::MeshType, MeshType >::value,
"The mesh entity belongs to other mesh type then the interpolants." );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
if( entity.getOrientation().x() == 1.0 )
- return 0.5 * ( u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] );
+ return 0.5 * ( u[ neighborEntities.template getEntityIndex< -1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, 0 >() ] );
else
- return 0.5 * ( u[ neighbourEntities.template getEntityIndex< 0, -1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] );
+ return 0.5 * ( u[ neighborEntities.template getEntityIndex< 0, -1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 1 >() ] );
}
};
@@ -148,12 +148,12 @@ class MeshEntitiesInterpolants< Meshes::Grid< 2, Real, Device, Index >, 2, 0 >
static_assert( std::is_same< typename MeshEntity::MeshType, MeshType >::value,
"The mesh entity belongs to other mesh type then the interpolants." );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
- return 0.25 * ( u[ neighbourEntities.template getEntityIndex< -1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, -1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, -1 >() ] );
+ return 0.25 * ( u[ neighborEntities.template getEntityIndex< -1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, -1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, -1 >() ] );
}
};
@@ -182,12 +182,12 @@ class MeshEntitiesInterpolants< Meshes::Grid< 2, Real, Device, Index >, 1, 2 >
static_assert( std::is_same< typename MeshEntity::MeshType, MeshType >::value,
"The mesh entity belongs to other mesh type then the interpolants." );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.template getNeighbourEntities< 1 >();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.template getNeighborEntities< 1 >();
- return 0.25 * ( u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, -1 >() ] );
+ return 0.25 * ( u[ neighborEntities.template getEntityIndex< -1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, -1 >() ] );
}
};
@@ -216,12 +216,12 @@ class MeshEntitiesInterpolants< Meshes::Grid< 2, Real, Device, Index >, 0, 2 >
static_assert( std::is_same< typename MeshEntity::MeshType, MeshType >::value,
"The mesh entity belongs to other mesh type then the interpolants." );
- const typename MeshEntity::template NeighbourEntities< 0 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 0 >& neighborEntities = entity.getNeighborEntities();
- return 0.25 * ( u[ neighbourEntities.template getEntityIndex< -1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, 1 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1, -1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, -1 >() ] );
+ return 0.25 * ( u[ neighborEntities.template getEntityIndex< -1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, 1 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1, -1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, -1 >() ] );
}
};
@@ -250,17 +250,17 @@ class MeshEntitiesInterpolants< Meshes::Grid< 3, Real, Device, Index >, 3, 2 >
static_assert( std::is_same< typename MeshEntity::MeshType, MeshType >::value,
"The mesh entity belongs to other mesh type then the interpolants." );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
if( entity.getOrientation().x() == 1.0 )
- return 0.5 * ( u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] );
+ return 0.5 * ( u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] );
if( entity.getOrientation().y() == 1.0 )
- return 0.5 * ( u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] );
+ return 0.5 * ( u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] );
else
- return 0.5 * ( u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] );
+ return 0.5 * ( u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] );
}
};
@@ -289,14 +289,14 @@ class MeshEntitiesInterpolants< Meshes::Grid< 3, Real, Device, Index >, 2, 3 >
static_assert( std::is_same< typename MeshEntity::MeshType, MeshType >::value,
"The mesh entity belongs to other mesh type then the interpolants." );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.template getNeighbourEntities< 2 >();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.template getNeighborEntities< 2 >();
- return 1.0 / 6.0 * ( u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] );
+ return 1.0 / 6.0 * ( u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] );
}
};
diff --git a/src/TNL/Operators/operator-Q/CMakeLists.txt b/src/TNL/Operators/operator-Q/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Operators/operator-Q/tnlFiniteVolumeOperatorQ.h b/src/TNL/Operators/operator-Q/tnlFiniteVolumeOperatorQ.h
index 5cf59f1766f8accd727b173dd1dc134bd166f06b..7f145198f049c2062e0e53d33b215e0fe8c0a3b8 100644
--- a/src/TNL/Operators/operator-Q/tnlFiniteVolumeOperatorQ.h
+++ b/src/TNL/Operators/operator-Q/tnlFiniteVolumeOperatorQ.h
@@ -19,7 +19,7 @@ namespace Operators {
template< typename Mesh,
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType,
+ typename Index = typename Mesh::GlobalIndexType,
int Precomputation = 0 >
class tnlFiniteVolumeOperatorQ
{
diff --git a/src/TNL/Operators/operator-Q/tnlFiniteVolumeOperatorQ_impl.h b/src/TNL/Operators/operator-Q/tnlFiniteVolumeOperatorQ_impl.h
index e27927f97d0c833529c705a15b6e5e0ba845d257..0fae70006b3b9f69c8157cfcf2ad5538489b8f1d 100644
--- a/src/TNL/Operators/operator-Q/tnlFiniteVolumeOperatorQ_impl.h
+++ b/src/TNL/Operators/operator-Q/tnlFiniteVolumeOperatorQ_impl.h
@@ -270,37 +270,37 @@ boundaryDerivative(
const IndexType& dy,
const IndexType& dz ) const
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const IndexType& cellIndex = entity.getIndex();
if ( ( AxeX == 1 ) && ( AxeY == 0 ) && ( AxeZ == 0 ) )
{
if ( ( dx == 1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< -1, 0 >() * ( u[ neighbourEntities.template getEntityIndex< 1,0 >() ] - u[ cellIndex ] );
+ return mesh.template getSpaceStepsProducts< -1, 0 >() * ( u[ neighborEntities.template getEntityIndex< 1,0 >() ] - u[ cellIndex ] );
if ( ( dx == -1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< -1, 0 >() * ( u[ cellIndex ] - u[ neighbourEntities.template getEntityIndex< -1,0 >() ] );
+ return mesh.template getSpaceStepsProducts< -1, 0 >() * ( u[ cellIndex ] - u[ neighborEntities.template getEntityIndex< -1,0 >() ] );
if ( ( dx == 0 ) && ( dy == 1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< -1, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 1,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1,1 >() ] - u[ neighbourEntities.template getEntityIndex< -1,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1,1 >() ] );
+ return mesh.template getSpaceStepsProducts< -1, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 1,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1,1 >() ] - u[ neighborEntities.template getEntityIndex< -1,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1,1 >() ] );
if ( ( dx == 0 ) && ( dy == -1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< -1, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 1,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1,-1 >() ] - u[ neighbourEntities.template getEntityIndex< -1,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< -1, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 1,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1,-1 >() ] - u[ neighborEntities.template getEntityIndex< -1,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1,-1 >() ] );
}
if ( ( AxeX == 0 ) && ( AxeY == 1 ) && ( AxeZ == 0 ) )
{
if ( ( dx == 0 ) && ( dy == 1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, -1 >() * ( u[ neighbourEntities.template getEntityIndex< 0,1 >() ] - u[ cellIndex ] );
+ return mesh.template getSpaceStepsProducts< 0, -1 >() * ( u[ neighborEntities.template getEntityIndex< 0,1 >() ] - u[ cellIndex ] );
if ( ( dx == 0 ) && ( dy == -1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, -1 >() * ( u[ cellIndex ] - u[ neighbourEntities.template getEntityIndex< 0,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, -1 >() * ( u[ cellIndex ] - u[ neighborEntities.template getEntityIndex< 0,-1 >() ] );
if ( ( dx == 1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, -1 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1,1 >() ] - u[ neighbourEntities.template getEntityIndex< 0,-1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, -1 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1,1 >() ] - u[ neighborEntities.template getEntityIndex< 0,-1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1,-1 >() ] );
if ( ( dx == -1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, -1 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,1 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1,1 >() ] - u[ neighbourEntities.template getEntityIndex< 0,-1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, -1 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,1 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1,1 >() ] - u[ neighborEntities.template getEntityIndex< 0,-1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1,-1 >() ] );
}
return 0.0;
}
@@ -321,14 +321,14 @@ operator()( const MeshEntity& entity,
const IndexType& dy,
const IndexType& dz ) const
{
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const IndexType& cellIndex = entity.getIndex();
if ( ( dx == 0 ) && ( dy == 0 ) && ( dz == 0 ) )
- return ::sqrt( this->eps + ( u[ neighbourEntities.template getEntityIndex< 0,1 >() ] - u[ cellIndex ] ) *
- ( u[ neighbourEntities.template getEntityIndex< 0,1 >() ] - u[ cellIndex ] )
- * mesh.template getSpaceStepsProducts< 0, -1 >() * mesh.template getSpaceStepsProducts< 0, -1 >() + ( u[ neighbourEntities.template getEntityIndex< 1,0 >() ] - u[ cellIndex ] )
- * ( u[ neighbourEntities.template getEntityIndex< 1,0 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< -1, 0 >() * mesh.template getSpaceStepsProducts< -1, 0 >() );
+ return ::sqrt( this->eps + ( u[ neighborEntities.template getEntityIndex< 0,1 >() ] - u[ cellIndex ] ) *
+ ( u[ neighborEntities.template getEntityIndex< 0,1 >() ] - u[ cellIndex ] )
+ * mesh.template getSpaceStepsProducts< 0, -1 >() * mesh.template getSpaceStepsProducts< 0, -1 >() + ( u[ neighborEntities.template getEntityIndex< 1,0 >() ] - u[ cellIndex ] )
+ * ( u[ neighborEntities.template getEntityIndex< 1,0 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< -1, 0 >() * mesh.template getSpaceStepsProducts< -1, 0 >() );
if ( ( dx == 1 ) && ( dy == 0 ) && ( dz == 0 ) )
return ::sqrt( this->eps + this->template boundaryDerivative< MeshEntity, Vector,1,0 >( mesh, entity, u, time, 1, 0 ) *
this->template boundaryDerivative< MeshEntity, Vector,1,0 >( mesh, entity, u, time, 1, 0 ) +
@@ -437,8 +437,7 @@ tnlFiniteVolumeOperatorQ< Meshes::Grid< 3, MeshReal, Device, MeshIndex >, Real,
bind( Vector& u)
{
this->u.bind(u);
- if(q.setSize(u.getSize()))
- return 1;
+ q.setSize(u.getSize());
q.setValue(0);
return 0;
}
@@ -480,76 +479,76 @@ boundaryDerivative(
const IndexType& dy,
const IndexType& dz ) const
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const IndexType& cellIndex = entity.getIndex();
if ( ( AxeX == 1 ) && ( AxeY == 0 ) && ( AxeZ == 0 ) )
{
if ( ( dx == 1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * ( u[ neighbourEntities.template getEntityIndex< 1,0,0 >() ] - u[ cellIndex ] );
+ return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * ( u[ neighborEntities.template getEntityIndex< 1,0,0 >() ] - u[ cellIndex ] );
if ( ( dx == -1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * ( u[ cellIndex ] - u[ neighbourEntities.template getEntityIndex< -1,0,0 >() ] );
+ return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * ( u[ cellIndex ] - u[ neighborEntities.template getEntityIndex< -1,0,0 >() ] );
if ( ( dx == 0 ) && ( dy == 1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 1,0,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1,1,0 >() ] - u[ neighbourEntities.template getEntityIndex< -1,0,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1,1,0 >() ] );
+ return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 1,0,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1,1,0 >() ] - u[ neighborEntities.template getEntityIndex< -1,0,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1,1,0 >() ] );
if ( ( dx == 0 ) && ( dy == -1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 1,0,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1,-1,0 >() ] - u[ neighbourEntities.template getEntityIndex< -1,0,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1,-1,0 >() ] );
+ return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 1,0,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1,-1,0 >() ] - u[ neighborEntities.template getEntityIndex< -1,0,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1,-1,0 >() ] );
if ( ( dx == 0 ) && ( dy == 0 ) && ( dz == 1 ) )
- return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 1,0,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1,0,1 >() ] - u[ neighbourEntities.template getEntityIndex< -1,0,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1,0,1 >() ] );
+ return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 1,0,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1,0,1 >() ] - u[ neighborEntities.template getEntityIndex< -1,0,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1,0,1 >() ] );
if ( ( dx == 0 ) && ( dy == 0 ) && ( dz == -1 ) )
- return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 1,0,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1,0,-1 >() ] - u[ neighbourEntities.template getEntityIndex< -1,0,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1,0,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< -1, 0, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 1,0,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1,0,-1 >() ] - u[ neighborEntities.template getEntityIndex< -1,0,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1,0,-1 >() ] );
}
if ( ( AxeX == 0 ) && ( AxeY == 1 ) && ( AxeZ == 0 ) )
{
if ( ( dx == 0 ) && ( dy == 1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * ( u[ neighbourEntities.template getEntityIndex< 0,1,0 >() ] - u[ cellIndex ] );
+ return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * ( u[ neighborEntities.template getEntityIndex< 0,1,0 >() ] - u[ cellIndex ] );
if ( ( dx == 0 ) && ( dy == -1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * ( u[ cellIndex ] - u[ neighbourEntities.template getEntityIndex< 0,-1,0 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * ( u[ cellIndex ] - u[ neighborEntities.template getEntityIndex< 0,-1,0 >() ] );
if ( ( dx == 1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,1,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1,1,0 >() ] - u[ neighbourEntities.template getEntityIndex< 0,-1,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1,-1,0 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,1,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1,1,0 >() ] - u[ neighborEntities.template getEntityIndex< 0,-1,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1,-1,0 >() ] );
if ( ( dx == -1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,1,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1,1,0 >() ] - u[ neighbourEntities.template getEntityIndex< 0,-1,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1,-1,0 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,1,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1,1,0 >() ] - u[ neighborEntities.template getEntityIndex< 0,-1,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1,-1,0 >() ] );
if ( ( dx == 0 ) && ( dy == 0 ) && ( dz == 1 ) )
- return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,1,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0,1,1 >() ] - u[ neighbourEntities.template getEntityIndex< 0,-1,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0,-1,1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,1,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0,1,1 >() ] - u[ neighborEntities.template getEntityIndex< 0,-1,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0,-1,1 >() ] );
if ( ( dx == 0 ) && ( dy == 0 ) && ( dz == -1 ) )
- return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,1,0 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0,1,-1 >() ] - u[ neighbourEntities.template getEntityIndex< 0,-1,0 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0,-1,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, -1, 0 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,1,0 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0,1,-1 >() ] - u[ neighborEntities.template getEntityIndex< 0,-1,0 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0,-1,-1 >() ] );
}
if ( ( AxeX == 0 ) && ( AxeY == 0 ) && ( AxeZ == 1 ) )
{
if ( ( dx == 0 ) && ( dy == 0 ) && ( dz == 1 ) )
- return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * ( u[ neighbourEntities.template getEntityIndex< 0,0,1 >() ] - u[ cellIndex ] );
+ return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * ( u[ neighborEntities.template getEntityIndex< 0,0,1 >() ] - u[ cellIndex ] );
if ( ( dx == 0 ) && ( dy == 0 ) && ( dz == -1 ) )
- return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * ( u[ cellIndex ] - u[ neighbourEntities.template getEntityIndex< 0,0,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * ( u[ cellIndex ] - u[ neighborEntities.template getEntityIndex< 0,0,-1 >() ] );
if ( ( dx == 1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,0,1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 1,0,1 >() ] - u[ neighbourEntities.template getEntityIndex< 0,0,-1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 1,0,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,0,1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 1,0,1 >() ] - u[ neighborEntities.template getEntityIndex< 0,0,-1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 1,0,-1 >() ] );
if ( ( dx == -1 ) && ( dy == 0 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,0,1 >() ] +
- u[ neighbourEntities.template getEntityIndex< -1,0,1 >() ] - u[ neighbourEntities.template getEntityIndex< 0,0,-1 >() ] -
- u[ neighbourEntities.template getEntityIndex< -1,0,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,0,1 >() ] +
+ u[ neighborEntities.template getEntityIndex< -1,0,1 >() ] - u[ neighborEntities.template getEntityIndex< 0,0,-1 >() ] -
+ u[ neighborEntities.template getEntityIndex< -1,0,-1 >() ] );
if ( ( dx == 0 ) && ( dy == 1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,0,1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0,1,1 >() ] - u[ neighbourEntities.template getEntityIndex< 0,0,-1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0,1,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,0,1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0,1,1 >() ] - u[ neighborEntities.template getEntityIndex< 0,0,-1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0,1,-1 >() ] );
if ( ( dx == 0 ) && ( dy == -1 ) && ( dz == 0 ) )
- return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * 0.25 * ( u[ neighbourEntities.template getEntityIndex< 0,0,1 >() ] +
- u[ neighbourEntities.template getEntityIndex< 0,-1,1 >() ] - u[ neighbourEntities.template getEntityIndex< 0,0,-1 >() ] -
- u[ neighbourEntities.template getEntityIndex< 0,-1,-1 >() ] );
+ return mesh.template getSpaceStepsProducts< 0, 0, -1 >() * 0.25 * ( u[ neighborEntities.template getEntityIndex< 0,0,1 >() ] +
+ u[ neighborEntities.template getEntityIndex< 0,-1,1 >() ] - u[ neighborEntities.template getEntityIndex< 0,0,-1 >() ] -
+ u[ neighborEntities.template getEntityIndex< 0,-1,-1 >() ] );
}
return 0.0;
}
@@ -571,16 +570,16 @@ operator()(
const IndexType& dy,
const IndexType& dz ) const
{
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const IndexType& cellIndex = entity.getIndex();
if ( ( dx == 0 ) && ( dy == 0 ) && ( dz == 0 ) )
- return ::sqrt( this->eps + ( u[ neighbourEntities.template getEntityIndex< 0,1,0 >() ] - u[ cellIndex ] ) *
- ( u[ neighbourEntities.template getEntityIndex< 0,1,0 >() ] - u[ cellIndex ] )
- * mesh.template getSpaceStepsProducts< 0, -1, 0 >() * mesh.template getSpaceStepsProducts< 0, -1, 0 >() + ( u[ neighbourEntities.template getEntityIndex< 1,0,0 >() ] - u[ cellIndex ] )
- * ( u[ neighbourEntities.template getEntityIndex< 1,0,0 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< -1, 0, 0 >() * mesh.template getSpaceStepsProducts< -1, 0, 0 >()
- + ( u[ neighbourEntities.template getEntityIndex< 0,0,1 >() ] - u[ cellIndex ] )
- * ( u[ neighbourEntities.template getEntityIndex< 0,0,1 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< 0, 0, -1 >() * mesh.template getSpaceStepsProducts< 0, 0, -1 >() );
+ return ::sqrt( this->eps + ( u[ neighborEntities.template getEntityIndex< 0,1,0 >() ] - u[ cellIndex ] ) *
+ ( u[ neighborEntities.template getEntityIndex< 0,1,0 >() ] - u[ cellIndex ] )
+ * mesh.template getSpaceStepsProducts< 0, -1, 0 >() * mesh.template getSpaceStepsProducts< 0, -1, 0 >() + ( u[ neighborEntities.template getEntityIndex< 1,0,0 >() ] - u[ cellIndex ] )
+ * ( u[ neighborEntities.template getEntityIndex< 1,0,0 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< -1, 0, 0 >() * mesh.template getSpaceStepsProducts< -1, 0, 0 >()
+ + ( u[ neighborEntities.template getEntityIndex< 0,0,1 >() ] - u[ cellIndex ] )
+ * ( u[ neighborEntities.template getEntityIndex< 0,0,1 >() ] - u[ cellIndex ] ) * mesh.template getSpaceStepsProducts< 0, 0, -1 >() * mesh.template getSpaceStepsProducts< 0, 0, -1 >() );
if ( ( dx == 1 ) && ( dy == 0 ) && ( dz == 0 ) )
return ::sqrt( this->eps + this->template boundaryDerivative< MeshEntity, Vector,1,0,0 >( mesh, entity, u, time, 1, 0, 0 ) *
this->template boundaryDerivative< MeshEntity, Vector,1,0,0 >( mesh, entity, u, time, 1, 0, 0 ) +
diff --git a/src/TNL/Operators/operator-Q/tnlOneSideDiffOperatorQ.h b/src/TNL/Operators/operator-Q/tnlOneSideDiffOperatorQ.h
index 29b2ac29091f8490adba7dc91c9b6075aea026b3..a96d22f5134029fb9686150713696da47ff05bfc 100644
--- a/src/TNL/Operators/operator-Q/tnlOneSideDiffOperatorQ.h
+++ b/src/TNL/Operators/operator-Q/tnlOneSideDiffOperatorQ.h
@@ -10,8 +10,6 @@
#pragma once
-#include <TNL/Containers/Vector.h>
-#include <TNL/Containers/SharedVector.h>
#include <TNL/Meshes/Grid.h>
namespace TNL {
@@ -19,7 +17,7 @@ namespace Operators {
template< typename Mesh,
typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class tnlOneSideDiffOperatorQ
{
};
diff --git a/src/TNL/Operators/operator-Q/tnlOneSideDiffOperatorQ_impl.h b/src/TNL/Operators/operator-Q/tnlOneSideDiffOperatorQ_impl.h
index 483680f5a174b11c917fa34b22bc8fdbc47cb788..21f5e44f08ec29fe365de11cfcbb5fb898f9af26 100644
--- a/src/TNL/Operators/operator-Q/tnlOneSideDiffOperatorQ_impl.h
+++ b/src/TNL/Operators/operator-Q/tnlOneSideDiffOperatorQ_impl.h
@@ -58,9 +58,9 @@ operator()( const MeshFunction& u,
const Real& time ) const
{
const IndexType& cellIndex = entity.getIndex();
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
- const RealType& u_x = ( u[ neighbourEntities.template getEntityIndex< 1 >() ] - u[ cellIndex ] ) *
+ const RealType& u_x = ( u[ neighborEntities.template getEntityIndex< 1 >() ] - u[ cellIndex ] ) *
mesh.template getSpaceStepsProducts< -1 >();
return ::sqrt( this->epsSquare + u_x * u_x );
}
@@ -79,12 +79,12 @@ getValueStriped( const MeshFunction& u,
const Real& time ) const
{
const IndexType& cellIndex = entity.getIndex();
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const RealType& u_c = u[ cellIndex ];
- const RealType& u_x_f = ( u[ neighbourEntities.template getEntityIndex< 1 >() ] - u_c ) *
+ const RealType& u_x_f = ( u[ neighborEntities.template getEntityIndex< 1 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< -1 >();
- const RealType& u_x_b = ( u_c - u[ neighbourEntities.template getEntityIndex< -1 >() ] ) *
+ const RealType& u_x_b = ( u_c - u[ neighborEntities.template getEntityIndex< -1 >() ] ) *
mesh.template getSpaceStepsProducts< -1 >();
return ::sqrt( this->epsSquare + 0.5 * ( u_x_f * u_x_f + u_x_b * u_x_b ) );
}
@@ -134,12 +134,12 @@ operator()( const MeshFunction& u,
const Real& time ) const
{
const IndexType& cellIndex = entity.getIndex();
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const RealType& u_c = u[ cellIndex ];
- const RealType u_x = ( u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] - u_c ) *
+ const RealType u_x = ( u[ neighborEntities.template getEntityIndex< 1, 0 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< -1, 0 >();
- const RealType u_y = ( u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] - u_c ) *
+ const RealType u_y = ( u[ neighborEntities.template getEntityIndex< 0, 1 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< 0, -1 >();
return ::sqrt( this->epsSquare + u_x * u_x + u_y * u_y );
}
@@ -158,16 +158,16 @@ getValueStriped( const MeshFunction& u,
const Real& time ) const
{
const IndexType& cellIndex = entity.getIndex();
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const RealType& u_c = u[ cellIndex ];
- const RealType u_x_f = ( u[ neighbourEntities.template getEntityIndex< 1, 0 >() ] - u_c ) *
+ const RealType u_x_f = ( u[ neighborEntities.template getEntityIndex< 1, 0 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< -1, 0 >();
- const RealType u_y_f = ( u[ neighbourEntities.template getEntityIndex< 0, 1 >() ] - u_c ) *
+ const RealType u_y_f = ( u[ neighborEntities.template getEntityIndex< 0, 1 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< 0, -1 >();
- const RealType u_x_b = ( u_c - u[ neighbourEntities.template getEntityIndex< -1, 0 >() ] ) *
+ const RealType u_x_b = ( u_c - u[ neighborEntities.template getEntityIndex< -1, 0 >() ] ) *
mesh.template getSpaceStepsProducts< -1, 0 >();
- const RealType u_y_b = ( u_c - u[ neighbourEntities.template getEntityIndex< 0, -1 >() ] ) *
+ const RealType u_y_b = ( u_c - u[ neighborEntities.template getEntityIndex< 0, -1 >() ] ) *
mesh.template getSpaceStepsProducts< 0, -1 >();
return ::sqrt( this->epsSquare +
@@ -219,15 +219,15 @@ operator()( const MeshFunction& u,
const Real& time ) const
{
const IndexType& cellIndex = entity.getIndex();
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const RealType& u_c =u[ cellIndex ];
- const RealType u_x = ( u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] - u_c ) *
+ const RealType u_x = ( u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< -1, 0, 0 >();
- const RealType u_y = ( u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] - u_c ) *
+ const RealType u_y = ( u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< 0, -1, 0 >();
- const RealType u_z = ( u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] - u_c ) *
+ const RealType u_z = ( u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< 0, 0, -1 >();
return ::sqrt( this->epsSquare + u_x * u_x + u_y * u_y + u_z * u_z );
}
@@ -246,21 +246,21 @@ getValueStriped( const MeshFunction& u,
const Real& time ) const
{
const IndexType& cellIndex = entity.getIndex();
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const typename MeshEntity::MeshType& mesh = entity.getMesh();
const RealType& u_c = u[ cellIndex ];
- const RealType u_x_f = ( u[ neighbourEntities.template getEntityIndex< 1, 0, 0 >() ] - u_c ) *
+ const RealType u_x_f = ( u[ neighborEntities.template getEntityIndex< 1, 0, 0 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< -1, 0, 0 >();
- const RealType u_y_f = ( u[ neighbourEntities.template getEntityIndex< 0, 1, 0 >() ] - u_c ) *
+ const RealType u_y_f = ( u[ neighborEntities.template getEntityIndex< 0, 1, 0 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< 0, -1, 0 >();
- const RealType u_z_f = ( u[ neighbourEntities.template getEntityIndex< 0, 0, 1 >() ] - u_c ) *
+ const RealType u_z_f = ( u[ neighborEntities.template getEntityIndex< 0, 0, 1 >() ] - u_c ) *
mesh.template getSpaceStepsProducts< 0, 0, -1 >();
- const RealType u_x_b = ( u_c - u[ neighbourEntities.template getEntityIndex< -1, 0, 0 >() ] ) *
+ const RealType u_x_b = ( u_c - u[ neighborEntities.template getEntityIndex< -1, 0, 0 >() ] ) *
mesh.template getSpaceStepsProducts< -1, 0, 0 >();
- const RealType u_y_b = ( u_c - u[ neighbourEntities.template getEntityIndex< 0, -1, 0 >() ] ) *
+ const RealType u_y_b = ( u_c - u[ neighborEntities.template getEntityIndex< 0, -1, 0 >() ] ) *
mesh.template getSpaceStepsProducts< 0, -1, 0 >();
- const RealType u_z_b = ( u_c - u[ neighbourEntities.template getEntityIndex< 0, 0, -1 >() ] ) *
+ const RealType u_z_b = ( u_c - u[ neighborEntities.template getEntityIndex< 0, 0, -1 >() ] ) *
mesh.template getSpaceStepsProducts< 0, 0, -1 >();
return ::sqrt( this->epsSquare +
diff --git a/src/TNL/Operators/operator-curvature/CMakeLists.txt b/src/TNL/Operators/operator-curvature/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Operators/operator-curvature/ExactOperatorCurvature.h b/src/TNL/Operators/operator-curvature/ExactOperatorCurvature.h
index 2c7d15fe99ab715a6a5ea8a98710f3cceddc29ad..33a20e255ce950bfa9714504185c1f236ad4abed 100644
--- a/src/TNL/Operators/operator-curvature/ExactOperatorCurvature.h
+++ b/src/TNL/Operators/operator-curvature/ExactOperatorCurvature.h
@@ -31,14 +31,8 @@ class ExactOperatorCurvature< OperatorQ, 1 >
static String getType();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder = 0, int YDiffOrder = 0, int ZDiffOrder = 0, typename Function, typename Point, typename Real >
-#else
template< int XDiffOrder = 0, int YDiffOrder = 0, int ZDiffOrder = 0, typename Function, typename Point, typename Real = typename Point::RealType >
-#endif
-#ifdef HAVE_CUDA
- __device__ __host__
-#endif
+ __cuda_callable__
static Real getValue( const Function& function,
const Point& v,
const Real& time = 0.0, const Real& eps = 1.0 );
@@ -54,14 +48,8 @@ class ExactOperatorCurvature< ExactOperatorQ, 2 >
static String getType();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder = 0, int YDiffOrder = 0, int ZDiffOrder = 0, typename Function, typename Point, typename Real >
-#else
template< int XDiffOrder = 0, int YDiffOrder = 0, int ZDiffOrder = 0, typename Function, typename Point, typename Real = typename Point::RealType >
-#endif
-#ifdef HAVE_CUDA
- __device__ __host__
-#endif
+ __cuda_callable__
static Real getValue( const Function& function,
const Point& v,
const Real& time = 0.0, const Real& eps = 1.0 );
@@ -76,14 +64,8 @@ class ExactOperatorCurvature< ExactOperatorQ, 3 >
static String getType();
-#ifdef HAVE_NOT_CXX11
- template< int XDiffOrder = 0, int YDiffOrder = 0, int ZDiffOrder = 0, typename Function, typename Point, typename Real >
-#else
template< int XDiffOrder = 0, int YDiffOrder = 0, int ZDiffOrder = 0, typename Function, typename Point, typename Real = typename Point::RealType >
-#endif
-#ifdef HAVE_CUDA
- __device__ __host__
-#endif
+ __cuda_callable__
static Real getValue( const Function& function,
const Point& v,
const Real& time = 0.0, const Real& eps = 1.0 )
diff --git a/src/TNL/ParallelFor.h b/src/TNL/ParallelFor.h
new file mode 100644
index 0000000000000000000000000000000000000000..b1e81ca97bf66576875aa1d0dcb34ecc0298d5df
--- /dev/null
+++ b/src/TNL/ParallelFor.h
@@ -0,0 +1,84 @@
+/***************************************************************************
+ ParallelFor.h - description
+ -------------------
+ begin : Mar 10, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#pragma once
+
+#include <TNL/Devices/Host.h>
+#include <TNL/Devices/Cuda.h>
+#include <TNL/Devices/CudaDeviceInfo.h>
+
+/*
+ * The implementation of ParallelFor is not meant to provide maximum performance
+ * at every cost, but maximum flexibility for operating with data stored on the
+ * device.
+ *
+ * The grid-stride loop for CUDA has been inspired by Nvidia's blog post:
+ * https://devblogs.nvidia.com/parallelforall/cuda-pro-tip-write-flexible-kernels-grid-stride-loops/
+ *
+ * Implemented by: Jakub Klinkovsky
+ */
+
+namespace TNL {
+
+template< typename Device = Devices::Host >
+struct ParallelFor
+{
+ template< typename Index,
+ typename Function,
+ typename... FunctionArgs >
+ static void exec( Index start, Index end, Function f, FunctionArgs... args )
+ {
+#ifdef HAVE_OPENMP
+ #pragma omp parallel for if( TNL::Devices::Host::isOMPEnabled() && end - start > 512 )
+#endif
+ for( Index i = start; i < end; i++ )
+ f( i, args... );
+ }
+};
+
+#ifdef HAVE_CUDA
+template< typename Index,
+ typename Function,
+ typename... FunctionArgs >
+__global__ void
+ParallelForKernel( Index start, Index end, Function f, FunctionArgs... args )
+{
+ for( Index i = start + blockIdx.x * blockDim.x + threadIdx.x;
+ i < end;
+ i += blockDim.x * gridDim.x )
+ {
+ f( i, args... );
+ }
+}
+#endif
+
+template<>
+struct ParallelFor< Devices::Cuda >
+{
+ template< typename Index,
+ typename Function,
+ typename... FunctionArgs >
+ static void exec( Index start, Index end, Function f, FunctionArgs... args )
+ {
+#ifdef HAVE_CUDA
+ if( end > start ) {
+ dim3 blockSize( 256 );
+ dim3 gridSize;
+ const int desGridSize = 32 * Devices::CudaDeviceInfo::getCudaMultiprocessors( Devices::CudaDeviceInfo::getActiveDevice() );
+ gridSize.x = min( desGridSize, Devices::Cuda::getNumberOfBlocks( end - start, blockSize.x ) );
+
+ Devices::Cuda::synchronizeDevice();
+ ParallelForKernel<<< gridSize, blockSize >>>( start, end, f, args... );
+ }
+#endif
+ }
+};
+
+} // namespace TNL
diff --git a/src/TNL/Problems/CMakeLists.txt b/src/TNL/Problems/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Problems/HeatEquationProblem.h b/src/TNL/Problems/HeatEquationProblem.h
index 1d5e95c9b70af4b91ea96666670ba737734f5d4c..cb01045e6cd00aec1d91f19357cc555144f4848e 100644
--- a/src/TNL/Problems/HeatEquationProblem.h
+++ b/src/TNL/Problems/HeatEquationProblem.h
@@ -108,10 +108,15 @@ class HeatEquationProblem : public PDEProblem< Mesh,
DofVectorPointer& rightHandSidePointer,
MeshDependentDataPointer& meshDependentData );
+ template< typename Matrix >
+ void saveFailedLinearSystem( const Matrix& matrix,
+ const DofVectorType& dofs,
+ const DofVectorType& rightHandSide ) const;
protected:
MeshFunctionPointer uPointer;
+ MeshFunctionPointer fuPointer;
DifferentialOperatorPointer differentialOperatorPointer;
diff --git a/src/TNL/Problems/HeatEquationProblem_impl.h b/src/TNL/Problems/HeatEquationProblem_impl.h
index 182e2fb0726249041fc839d7c842b601ac9be1b3..39255752baa2c81112a9db5238ef30fc2bec4c90 100644
--- a/src/TNL/Problems/HeatEquationProblem_impl.h
+++ b/src/TNL/Problems/HeatEquationProblem_impl.h
@@ -154,8 +154,7 @@ setupLinearSystem( const MeshPointer& meshPointer,
const IndexType dofs = this->getDofs( meshPointer );
typedef typename MatrixPointer::ObjectType::CompressedRowLengthsVector CompressedRowLengthsVectorType;
SharedPointer< CompressedRowLengthsVectorType > rowLengthsPointer;
- if( ! rowLengthsPointer->setSize( dofs ) )
- return false;
+ rowLengthsPointer->setSize( dofs );
Matrices::MatrixSetter< MeshType, DifferentialOperator, BoundaryCondition, CompressedRowLengthsVectorType > matrixSetter;
matrixSetter.template getCompressedRowLengths< typename Mesh::Cell >(
meshPointer,
@@ -163,8 +162,7 @@ setupLinearSystem( const MeshPointer& meshPointer,
boundaryConditionPointer,
rowLengthsPointer );
matrixPointer->setDimensions( dofs, dofs );
- if( ! matrixPointer->setCompressedRowLengths( *rowLengthsPointer ) )
- return false;
+ matrixPointer->setCompressedRowLengths( *rowLengthsPointer );
return true;
//return MultidiagonalMatrixSetter< Mesh >::setupMatrix( mesh, matrix );
}
@@ -252,5 +250,18 @@ assemblyLinearSystem( const RealType& time,
bPointer );
}
+template< typename Mesh,
+ typename BoundaryCondition,
+ typename RightHandSide,
+ typename DifferentialOperator >
+ template< typename Matrix >
+void
+HeatEquationProblem< Mesh, BoundaryCondition, RightHandSide, DifferentialOperator >::
+saveFailedLinearSystem( const Matrix& matrix,
+ const DofVectorType& dofs,
+ const DofVectorType& rightHandSide ) const
+{
+}
+
} // namespace Problems
} // namespace TNL
diff --git a/src/TNL/Problems/MeanCurvatureFlowProblem.h b/src/TNL/Problems/MeanCurvatureFlowProblem.h
index 0e7a88df3312b5fde78ec4964b4984ed95500f8d..a06049647dcc0ef45ef7540fb3e9fd1ae39f479c 100644
--- a/src/TNL/Problems/MeanCurvatureFlowProblem.h
+++ b/src/TNL/Problems/MeanCurvatureFlowProblem.h
@@ -31,7 +31,7 @@ template< typename Mesh,
typename DifferentialOperator =
OneSidedMeanCurvature< Mesh,
typename Mesh::RealType,
- typename Mesh::IndexType,
+ typename Mesh::GlobalIndexType,
false > >
class MeanCurvatureFlowProblem : public PDEProblem< Mesh,
typename DifferentialOperator::RealType,
@@ -97,6 +97,10 @@ class MeanCurvatureFlowProblem : public PDEProblem< Mesh,
DofVectorType& rightHandSide,
MeshDependentDataPointer& meshDependentData );
+ template< typename Matrix >
+ void saveFailedLinearSystem( const Matrix& matrix,
+ const DofVectorType& dofs,
+ const DofVectorType& rightHandSide ) const;
protected:
diff --git a/src/TNL/Problems/MeanCurvatureFlowProblem_impl.h b/src/TNL/Problems/MeanCurvatureFlowProblem_impl.h
index 79c730732fff1fea884d623076e08ce32dc3bf96..4dcba17fcd9ca6f78ce16aba0df8b6128ea8f3f0 100644
--- a/src/TNL/Problems/MeanCurvatureFlowProblem_impl.h
+++ b/src/TNL/Problems/MeanCurvatureFlowProblem_impl.h
@@ -140,8 +140,7 @@ setupLinearSystem( const MeshType& mesh,
const IndexType dofs = this->getDofs( mesh );
typedef typename MatrixType::CompressedRowLengthsVector CompressedRowLengthsVectorType;
CompressedRowLengthsVectorType rowLengths;
- if( ! rowLengths.setSize( dofs ) )
- return false;
+ rowLengths.setSize( dofs );
MatrixSetter< MeshType, DifferentialOperator, BoundaryCondition, CompressedRowLengthsVectorType > matrixSetter;
matrixSetter.template getCompressedRowLengths< typename Mesh::Cell >(
mesh,
@@ -150,8 +149,7 @@ setupLinearSystem( const MeshType& mesh,
rowLengths
);
matrix.setDimensions( dofs, dofs );
- if( ! matrix.setCompressedRowLengths( rowLengths ) )
- return false;
+ matrix.setCompressedRowLengths( rowLengths );
return true;
}
@@ -262,5 +260,18 @@ assemblyLinearSystem( const RealType& time,
//abort();*/
}
+template< typename Mesh,
+ typename BoundaryCondition,
+ typename RightHandSide,
+ typename DifferentialOperator >
+ template< typename Matrix >
+void
+MeanCurvatureFlowProblem< Mesh, BoundaryCondition, RightHandSide, DifferentialOperator >::
+saveFailedLinearSystem( const Matrix& matrix,
+ const DofVectorType& dofs,
+ const DofVectorType& rightHandSide ) const
+{
+}
+
} // namespace Problems
} // namespace TNL
diff --git a/src/TNL/Problems/PDEProblem.h b/src/TNL/Problems/PDEProblem.h
index 431f6588828d2166f42bbc297eb533ee00dd260c..2da2f9981f1bc12673096d278c20a4b3b1b7eedc 100644
--- a/src/TNL/Problems/PDEProblem.h
+++ b/src/TNL/Problems/PDEProblem.h
@@ -20,7 +20,7 @@ namespace Problems {
template< typename Mesh,
typename Real = typename Mesh::RealType,
typename Device = typename Mesh::DeviceType,
- typename Index = typename Mesh::IndexType >
+ typename Index = typename Mesh::GlobalIndexType >
class PDEProblem : public Problem< Real, Device, Index >
{
public:
diff --git a/src/TNL/Problems/cfd/CMakeLists.txt b/src/TNL/Problems/cfd/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Problems/cfd/navier-stokes/CMakeLists.txt b/src/TNL/Problems/cfd/navier-stokes/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Problems/cfd/navier-stokes/NavierStokesSolver_impl.h b/src/TNL/Problems/cfd/navier-stokes/NavierStokesSolver_impl.h
index 50f2138dd01edf6112ecd6d151d1776aa2910edf..229ff4667b4e6d0c84e1ba27dd5464f40dd94b7f 100644
--- a/src/TNL/Problems/cfd/navier-stokes/NavierStokesSolver_impl.h
+++ b/src/TNL/Problems/cfd/navier-stokes/NavierStokesSolver_impl.h
@@ -345,10 +345,10 @@ void NavierStokesSolver< AdvectionScheme,
SolverVectorType& u,
SolverVectorType& fu )
{
- TNL_ASSERT( this->advection, );
- TNL_ASSERT( this->u1Viscosity, );
- TNL_ASSERT( this->u2Viscosity, );
- TNL_ASSERT( this->boundaryConditions, );
+ TNL_ASSERT_TRUE( this->advection, "advection scheme was not set" );
+ TNL_ASSERT_TRUE( this->u1Viscosity, "diffusion scheme was not set" );
+ TNL_ASSERT_TRUE( this->u2Viscosity, "diffusion scheme was not set" );
+ TNL_ASSERT_TRUE( this->boundaryConditions, "boundary conditions were not set" );
SharedVector< RealType, DeviceType, IndexType > dofs_rho, dofs_rho_u1, dofs_rho_u2, dofs_e,
rho_t, rho_u1_t, rho_u2_t, e_t;
diff --git a/src/TNL/SharedPointer.h b/src/TNL/SharedPointer.h
index a0308daf19dce085093ef75633f13083d91cb03f..c9e71427518d373f26aec6ff57e386f5809d7365 100644
--- a/src/TNL/SharedPointer.h
+++ b/src/TNL/SharedPointer.h
@@ -14,11 +14,13 @@
#include <TNL/Devices/Host.h>
#include <TNL/Devices/Cuda.h>
+#include <TNL/Devices/MIC.h>
#include <TNL/SmartPointer.h>
#include <cstring>
+
//#define TNL_DEBUG_SHARED_POINTERS
#ifdef TNL_DEBUG_SHARED_POINTERS
@@ -161,11 +163,18 @@ class SharedPointer< Object, Devices::Host > : public SmartPointer
return this->pd->data;
}
- operator bool()
+ __cuda_callable__
+ operator bool() const
{
return this->pd;
}
+ __cuda_callable__
+ bool operator!() const
+ {
+ return ! this->pd;
+ }
+
template< typename Device = Devices::Host >
__cuda_callable__
const Object& getData() const
@@ -395,11 +404,18 @@ class SharedPointer< Object, Devices::Cuda > : public SmartPointer
return this->pd->data;
}
- operator bool()
+ __cuda_callable__
+ operator bool() const
{
return this->pd;
}
+ __cuda_callable__
+ bool operator!() const
+ {
+ return ! this->pd;
+ }
+
template< typename Device = Devices::Host >
__cuda_callable__
const Object& getData() const
@@ -500,7 +516,7 @@ class SharedPointer< Object, Devices::Cuda > : public SmartPointer
#endif
TNL_ASSERT( this->cuda_pointer, );
cudaMemcpy( (void*) this->cuda_pointer, (void*) &this->pd->data, sizeof( Object ), cudaMemcpyHostToDevice );
- if( ! checkCudaDevice ) {
+ if( ! TNL_CHECK_CUDA_DEVICE ) {
return false;
}
this->set_last_sync_state();
@@ -544,12 +560,8 @@ class SharedPointer< Object, Devices::Cuda > : public SmartPointer
bool allocate( Args... args )
{
this->pd = new PointerData( args... );
- if( ! this->pd )
- return false;
// pass to device
this->cuda_pointer = Devices::Cuda::passToDevice( this->pd->data );
- if( ! this->cuda_pointer )
- return false;
// set last-sync state
this->set_last_sync_state();
#ifdef TNL_DEBUG_SHARED_POINTERS
@@ -602,4 +614,355 @@ class SharedPointer< Object, Devices::Cuda > : public SmartPointer
Object* cuda_pointer;
};
+#ifdef HAVE_MIC
+/****
+ * Specialization for MIC
+ */
+template< typename Object>
+class SharedPointer< Object, Devices::MIC > : public SmartPointer
+{
+ private:
+ // Convenient template alias for controlling the selection of copy- and
+ // move-constructors and assignment operators using SFINAE.
+ // The type Object_ is "enabled" iff Object_ and Object are not the same,
+ // but after removing const and volatile qualifiers they are the same.
+ template< typename Object_ >
+ using Enabler = std::enable_if< ! std::is_same< Object_, Object >::value &&
+ std::is_same< typename std::remove_cv< Object >::type, Object_ >::value >;
+
+ // friend class will be needed for templated assignment operators
+ template< typename Object_, typename Device_>
+ friend class SharedPointer;
+
+ public:
+
+ typedef Object ObjectType;
+ typedef Devices::MIC DeviceType;
+ typedef SharedPointer< Object, Devices::MIC> ThisType;
+
+ template< typename... Args >
+ explicit SharedPointer( Args... args )
+ : pd( nullptr ),
+ mic_pointer( nullptr )
+ {
+ this->allocate( args... );
+ }
+
+ // this is needed only to avoid the default compiler-generated constructor
+ SharedPointer( const ThisType& pointer )
+ : pd( (PointerData*) pointer.pd ),
+ mic_pointer( pointer.mic_pointer )
+ {
+ this->pd->counter += 1;
+ }
+
+ // conditional constructor for non-const -> const data
+ template< typename Object_,
+ typename = typename Enabler< Object_ >::type >
+ SharedPointer( const SharedPointer< Object_, DeviceType >& pointer )
+ : pd( (PointerData*) pointer.pd ),
+ mic_pointer( pointer.mic_pointer )
+ {
+ this->pd->counter += 1;
+ }
+
+ // this is needed only to avoid the default compiler-generated constructor
+ SharedPointer( ThisType&& pointer )
+ : pd( (PointerData*) pointer.pd ),
+ mic_pointer( pointer.mic_pointer )
+ {
+ pointer.pd = nullptr;
+ pointer.mic_pointer = nullptr;
+ }
+
+ // conditional constructor for non-const -> const data
+ template< typename Object_,
+ typename = typename Enabler< Object_ >::type >
+ SharedPointer( SharedPointer< Object_, DeviceType >&& pointer )
+ : pd( (PointerData*) pointer.pd ),
+ mic_pointer( pointer.mic_pointer )
+ {
+ pointer.pd = nullptr;
+ pointer.mic_pointer = nullptr;
+ }
+
+ template< typename... Args >
+ bool recreate( Args... args )
+ {
+#ifdef TNL_DEBUG_SHARED_POINTERS
+ std::cerr << "Recreating shared pointer to " << demangle(typeid(ObjectType).name()) << std::endl;
+#endif
+ if( ! this->pd )
+ return this->allocate( args... );
+
+ if( this->pd->counter == 1 )
+ {
+ /****
+ * The object is not shared -> recreate it in-place, without reallocation
+ */
+ this->pd->data.~Object();
+ new ( &this->pd->data ) Object( args... );
+ Devices::MIC::CopyToMIC(this->mic_pointer,(void*) &this->pd->data,sizeof(Object));
+ this->set_last_sync_state();
+ return true;
+ }
+
+ // free will just decrement the counter
+ this->free();
+
+ return this->allocate( args... );
+ }
+
+ const Object* operator->() const
+ {
+ return &this->pd->data;
+ }
+
+ Object* operator->()
+ {
+ this->pd->maybe_modified = true;
+ return &this->pd->data;
+ }
+
+ const Object& operator *() const
+ {
+ return this->pd->data;
+ }
+
+ Object& operator *()
+ {
+ this->pd->maybe_modified = true;
+ return this->pd->data;
+ }
+
+ operator bool()
+ {
+ return this->pd;
+ }
+
+ template< typename Device = Devices::Host >
+ __cuda_callable__
+ const Object& getData() const
+ {
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::MIC >::value, "Only Devices::Host or Devices::MIC devices are accepted here." );
+ TNL_ASSERT( this->pd, );
+ TNL_ASSERT( this->mic_pointer, );
+ if( std::is_same< Device, Devices::Host >::value )
+ return this->pd->data;
+ if( std::is_same< Device, Devices::MIC >::value )
+ return *( this->mic_pointer );
+
+ }
+
+ template< typename Device = Devices::Host >
+ __cuda_callable__
+ Object& modifyData()
+ {
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::MIC >::value, "Only Devices::Host or Devices::MIC devices are accepted here." );
+ TNL_ASSERT( this->pd, );
+ TNL_ASSERT( this->mic_pointer, );
+ if( std::is_same< Device, Devices::Host >::value )
+ {
+ this->pd->maybe_modified = true;
+ return this->pd->data;
+ }
+ if( std::is_same< Device, Devices::MIC >::value )
+ return *( this->mic_pointer );
+
+ }
+
+ // this is needed only to avoid the default compiler-generated operator
+ const ThisType& operator=( const ThisType& ptr )
+ {
+ this->free();
+ this->pd = (PointerData*) ptr.pd;
+ this->mic_pointer = ptr.mic_pointer;
+ this->pd->counter += 1;
+#ifdef TNL_DEBUG_SHARED_POINTERS
+ std::cerr << "Copy-assigned shared pointer: counter = " << this->pd->counter << ", type: " << demangle(typeid(ObjectType).name()) << std::endl;
+#endif
+ return *this;
+ }
+
+ // conditional operator for non-const -> const data
+ template< typename Object_,
+ typename = typename Enabler< Object_ >::type >
+ const ThisType& operator=( const SharedPointer< Object_, DeviceType >& ptr )
+ {
+ this->free();
+ this->pd = (PointerData*) ptr.pd;
+ this->mic_pointer = ptr.mic_pointer;
+ this->pd->counter += 1;
+#ifdef TNL_DEBUG_SHARED_POINTERS
+ std::cerr << "Copy-assigned shared pointer: counter = " << this->pd->counter << ", type: " << demangle(typeid(ObjectType).name()) << std::endl;
+#endif
+ return *this;
+ }
+
+ // this is needed only to avoid the default compiler-generated operator
+ const ThisType& operator=( ThisType&& ptr )
+ {
+ this->free();
+ this->pd = (PointerData*) ptr.pd;
+ this->mic_pointer = ptr.mic_pointer;
+ ptr.pd = nullptr;
+ ptr.mic_pointer = nullptr;
+#ifdef TNL_DEBUG_SHARED_POINTERS
+ std::cerr << "Move-assigned shared pointer: counter = " << this->pd->counter << ", type: " << demangle(typeid(ObjectType).name()) << std::endl;
+#endif
+ return *this;
+ }
+
+ // conditional operator for non-const -> const data
+ template< typename Object_,
+ typename = typename Enabler< Object_ >::type >
+ const ThisType& operator=( SharedPointer< Object_, DeviceType >&& ptr )
+ {
+ this->free();
+ this->pd = (PointerData*) ptr.pd;
+ this->mic_pointer = ptr.mic_pointer;
+ ptr.pd = nullptr;
+ ptr.mic_pointer = nullptr;
+#ifdef TNL_DEBUG_SHARED_POINTERS
+ std::cerr << "Move-assigned shared pointer: counter = " << this->pd->counter << ", type: " << demangle(typeid(ObjectType).name()) << std::endl;
+#endif
+ return *this;
+ }
+
+ bool synchronize()
+ {
+ if( ! this->pd )
+ return true;
+
+ if( this->modified() )
+ {
+#ifdef TNL_DEBUG_SHARED_POINTERS
+ std::cerr << "Synchronizing shared pointer: counter = " << this->pd->counter << ", type: " << demangle(typeid(Object).name()) << std::endl;
+ std::cerr << " ( " << sizeof( Object ) << " bytes, MIC adress " << this->mic_pointer << " )" << std::endl;
+#endif
+ TNL_ASSERT( this->mic_pointer, );
+
+ Devices::MIC::CopyToMIC((void*)this->mic_pointer,(void*) &this->pd->data,sizeof(Object));
+ this->set_last_sync_state();
+ return true;
+ }
+ return false; //??
+ }
+
+ void clear()
+ {
+ this->free();
+ }
+
+ ~SharedPointer()
+ {
+ this->free();
+ Devices::MIC::removeSmartPointer( this );
+ }
+
+ protected:
+
+ struct PointerData
+ {
+ Object data;
+ uint8_t data_image[ sizeof(Object) ];
+ int counter;
+ bool maybe_modified;
+
+ template< typename... Args >
+ explicit PointerData( Args... args )
+ : data( args... ),
+ counter( 1 ),
+ maybe_modified( false )
+ {}
+ };
+
+ template< typename... Args >
+ bool allocate( Args... args )
+ {
+ this->pd = new PointerData( args... );
+ if( ! this->pd )
+ return false;
+
+ mic_pointer=(Object*)Devices::MIC::AllocMIC(sizeof(Object));
+ Devices::MIC::CopyToMIC((void*)this->mic_pointer,(void*) &this->pd->data,sizeof(Object));
+
+ if( ! this->mic_pointer )
+ return false;
+ // set last-sync state
+ this->set_last_sync_state();
+#ifdef TNL_DEBUG_SHARED_POINTERS
+ std::cerr << "Created shared pointer to " << demangle(typeid(ObjectType).name()) << " (mic_pointer = " << this->mic_pointer << ")" << std::endl;
+#endif
+ Devices::MIC::insertSmartPointer( this );
+ return true;
+ }
+
+ void set_last_sync_state()
+ {
+ TNL_ASSERT( this->pd, );
+ std::memcpy( (void*) &this->pd->data_image, (void*) &this->pd->data, sizeof( Object ) );
+ this->pd->maybe_modified = false;
+ }
+
+ bool modified()
+ {
+ TNL_ASSERT( this->pd, );
+ // optimization: skip bitwise comparison if we're sure that the data is the same
+ if( ! this->pd->maybe_modified )
+ return false;
+ return std::memcmp( (void*) &this->pd->data_image, (void*) &this->pd->data, sizeof( Object ) ) != 0;
+ }
+
+ void free()
+ {
+ if( this->pd )
+ {
+#ifdef TNL_DEBUG_SHARED_POINTERS
+ std::cerr << "Freeing shared pointer: counter = " << this->pd->counter << ", mic_pointer = " << this->mic_pointer << ", type: " << demangle(typeid(ObjectType).name()) << std::endl;
+#endif
+ if( ! --this->pd->counter )
+ {
+ delete this->pd;
+ this->pd = nullptr;
+ if( this->mic_pointer )
+ {
+ Devices::MIC::FreeMIC((void*)mic_pointer);
+ mic_pointer=nullptr;
+ }
+#ifdef TNL_DEBUG_SHARED_POINTERS
+ std::cerr << "...deleted data." << std::endl;
+#endif
+ }
+ }
+ }
+
+ PointerData* pd;
+
+ // cuda_pointer can't be part of PointerData structure, since we would be
+ // unable to dereference this-pd on the device -- NevÃÂm zda to platàpro MIC, asi jo
+ Object* mic_pointer;
+};
+#endif
+
+
+#if (!defined(NDEBUG)) && (!defined(HAVE_MIC))
+namespace Assert {
+
+template< typename Object, typename Device >
+struct Formatter< SharedPointer< Object, Device > >
+{
+ static std::string
+ printToString( const SharedPointer< Object, Device >& value )
+ {
+ ::std::stringstream ss;
+ ss << "(SharedPointer< " << Object::getType() << ", " << Device::getDeviceType()
+ << " > object at " << &value << ")";
+ return ss.str();
+ }
+};
+
+} // namespace Assert
+#endif
+
} // namespace TNL
diff --git a/src/TNL/SmartPointersRegister.cpp b/src/TNL/SmartPointersRegister.cpp
index 7fe8a654f4a6ac6c435b4d8ccb02aadc84df37fb..03d3e058f004c201c61d7c307cce7327d5e106c9 100644
--- a/src/TNL/SmartPointersRegister.cpp
+++ b/src/TNL/SmartPointersRegister.cpp
@@ -44,7 +44,7 @@ bool SmartPointersRegister::synchronizeDevice( int deviceId )
const auto & set = pointersOnDevices.at( deviceId );
for( auto&& it : set )
( *it ).synchronize();
- return checkCudaDevice;
+ return TNL_CHECK_CUDA_DEVICE;
}
catch( std::out_of_range ) {
return false;
diff --git a/src/TNL/Solvers/BuildConfigTags.h b/src/TNL/Solvers/BuildConfigTags.h
index 003974c3b39c3fae2ce7645803e70a0b5312ea05..e89aa9d925c7bfe748ec8ab1f98bbc8408787d18 100644
--- a/src/TNL/Solvers/BuildConfigTags.h
+++ b/src/TNL/Solvers/BuildConfigTags.h
@@ -16,6 +16,7 @@
#include <TNL/Solvers/Linear/SOR.h>
#include <TNL/Solvers/Linear/CG.h>
#include <TNL/Solvers/Linear/BICGStab.h>
+#include <TNL/Solvers/Linear/BICGStabL.h>
#include <TNL/Solvers/Linear/CWYGMRES.h>
#include <TNL/Solvers/Linear/GMRES.h>
#include <TNL/Solvers/Linear/TFQMR.h>
@@ -36,6 +37,10 @@ template< typename ConfigTag, typename Device > struct ConfigTagDevice{ enum { e
template< typename ConfigTag > struct ConfigTagDevice< ConfigTag, Devices::Cuda >{ enum { enabled = false }; };
#endif
+#ifndef HAVE_MIC
+template< typename ConfigTag > struct ConfigTagDevice< ConfigTag, Devices::MIC >{ enum { enabled = false }; };
+#endif
+
/****
* All real types are enabled by default.
*/
@@ -133,6 +138,16 @@ public:
using Template = Linear::BICGStab< Matrix, Preconditioner >;
};
+class SemiImplicitBICGStabLSolverTag
+{
+public:
+ template< typename Matrix,
+ typename Preconditioner = Linear::Preconditioners::Dummy< typename Matrix::RealType,
+ typename Matrix::DeviceType,
+ typename Matrix::IndexType > >
+ using Template = Linear::BICGStabL< Matrix, Preconditioner >;
+};
+
class SemiImplicitCWYGMRESSolverTag
{
public:
diff --git a/src/TNL/Solvers/CMakeLists.txt b/src/TNL/Solvers/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Solvers/Linear/BICGStab.h b/src/TNL/Solvers/Linear/BICGStab.h
index c7ba294d02100c386250fadbc4050336bcd46ebc..02d17965e890123968bf6059082cba0164e16f71 100644
--- a/src/TNL/Solvers/Linear/BICGStab.h
+++ b/src/TNL/Solvers/Linear/BICGStab.h
@@ -14,7 +14,6 @@
#include <TNL/Object.h>
#include <TNL/SharedPointer.h>
#include <TNL/Containers/Vector.h>
-#include <TNL/Containers/SharedVector.h>
#include <TNL/Solvers/Linear/Preconditioners/Dummy.h>
#include <TNL/Solvers/IterativeSolver.h>
#include <TNL/Solvers/Linear/LinearResidueGetter.h>
@@ -62,7 +61,7 @@ class BICGStab : public Object,
protected:
- bool setSize( IndexType size );
+ void setSize( IndexType size );
bool exact_residue;
diff --git a/src/TNL/Solvers/Linear/BICGStabL.h b/src/TNL/Solvers/Linear/BICGStabL.h
new file mode 100644
index 0000000000000000000000000000000000000000..124f70839950e5565a5fb9c0a6931c08abce4509
--- /dev/null
+++ b/src/TNL/Solvers/Linear/BICGStabL.h
@@ -0,0 +1,119 @@
+/***************************************************************************
+ BICGStabL.h - description
+ -------------------
+ begin : Jul 4, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+/*
+ * BICGStabL implements an iterative solver for non-symmetric linear systems,
+ * using the BiCGstab(l) algorithm described in [1] and [2]. It is a
+ * generalization of the stabilized biconjugate-gradient (BiCGstab) algorithm
+ * proposed by van der Vorst [3]. BiCGstab(1) is equivalent to BiCGstab, and
+ * BiCGstab(2) is a slightly more efficient version of the BiCGstab2 algorithm
+ * by Gutknecht [4], while BiCGstab(l>2) is a further generalization.
+ *
+ * This code was implemented by: Jakub Klinkovsky <klinkjak@fjfi.cvut.cz>
+ *
+ * [1] Gerard L. G. Sleijpen and Diederik R. Fokkema, "BiCGstab(l) for linear
+ * equations involving unsymmetric matrices with complex spectrum",
+ * Electronic Trans. on Numerical Analysis 1, 11-32 (1993).
+ * [2] Gerard L. G. Sleijpen, Henk A. van der Vorst, and Diederik R. Fokkema,
+ * "BiCGstab(l) and other Hybrid Bi-CG Methods", Numerical Algorithms 7,
+ * 75-109 (1994).
+ * [3] Henk A. van der Vorst, "Bi-CGSTAB: A fast and smoothly converging variant
+ * of Bi-CG for the solution of nonsymmetric linear systems, SIAM Journal on
+ * scientific and Statistical Computing 13.2, 631-644 (1992).
+ * [4] Martin H. Gutknecht, "Variants of BiCGStab for matrices with complex
+ * spectrum", IPS Research Report No. 91-14 (1991).
+ *
+ * TODO: further variations to explore:
+ *
+ * [5] Gerard L. G. Sleijpen and Henk A. van der Vorst, "Reliable updated
+ * residuals in hybrid Bi-CG methods", Computing 56 (2), 141-163 (1996).
+ * [6] Gerard L. G. Sleijpen and Henk A. van der Vorst, "Maintaining convergence
+ * properties of BiCGstab methods in finite precision arithmetic", Numerical
+ * Algorithms 10, 203-223 (1995).
+ */
+
+#pragma once
+
+#include <math.h>
+#include <TNL/Object.h>
+#include <TNL/SharedPointer.h>
+#include <TNL/Containers/Vector.h>
+#include <TNL/Solvers/Linear/Preconditioners/Dummy.h>
+#include <TNL/Solvers/IterativeSolver.h>
+#include <TNL/Solvers/Linear/LinearResidueGetter.h>
+
+namespace TNL {
+namespace Solvers {
+namespace Linear {
+
+template< typename Matrix,
+ typename Preconditioner = Preconditioners::Dummy< typename Matrix :: RealType,
+ typename Matrix :: DeviceType,
+ typename Matrix :: IndexType> >
+
+class BICGStabL
+ : public Object,
+ public IterativeSolver< typename Matrix :: RealType,
+ typename Matrix :: IndexType >
+{
+public:
+ typedef typename Matrix::RealType RealType;
+ typedef typename Matrix::IndexType IndexType;
+ typedef typename Matrix::DeviceType DeviceType;
+ typedef Matrix MatrixType;
+ typedef Preconditioner PreconditionerType;
+ typedef SharedPointer< const MatrixType, DeviceType > MatrixPointer;
+ typedef SharedPointer< const PreconditionerType, DeviceType > PreconditionerPointer;
+ typedef Containers::Vector< RealType, DeviceType, IndexType > DeviceVector;
+ typedef Containers::Vector< RealType, Devices::Host, IndexType > HostVector;
+
+ BICGStabL();
+
+ String getType() const;
+
+ static void configSetup( Config::ConfigDescription& config,
+ const String& prefix = "" );
+
+ bool setup( const Config::ParameterContainer& parameters,
+ const String& prefix = "" );
+
+ void setMatrix( const MatrixPointer& matrix );
+
+ void setPreconditioner( const PreconditionerPointer& preconditioner );
+
+ template< typename Vector,
+ typename ResidueGetter = LinearResidueGetter< Matrix, Vector > >
+ bool solve( const Vector& b, Vector& x );
+
+protected:
+ void setSize( IndexType size );
+
+ int ell = 1;
+
+ bool exact_residue = false;
+
+ // matrices (in column-major format)
+ DeviceVector R, U;
+ // single vectors
+ DeviceVector r_ast, M_tmp, res_tmp;
+ // host-only storage
+ HostVector T, sigma, g_0, g_1, g_2;
+
+ IndexType size, ldSize;
+
+ MatrixPointer matrix;
+ PreconditionerPointer preconditioner;
+};
+
+} // namespace Linear
+} // namespace Solvers
+} // namespace TNL
+
+#include <TNL/Solvers/Linear/BICGStabL_impl.h>
diff --git a/src/TNL/Solvers/Linear/BICGStabL_impl.h b/src/TNL/Solvers/Linear/BICGStabL_impl.h
new file mode 100644
index 0000000000000000000000000000000000000000..f448e7c8835c49dd72a31e4b8e689dd09d825f85
--- /dev/null
+++ b/src/TNL/Solvers/Linear/BICGStabL_impl.h
@@ -0,0 +1,323 @@
+/***************************************************************************
+ BICGStabL.h - description
+ -------------------
+ begin : Jul 4, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Jakub Klinkovsky
+
+#pragma once
+
+#include "BICGStabL.h"
+
+#include <TNL/Matrices/MatrixOperations.h>
+
+namespace TNL {
+namespace Solvers {
+namespace Linear {
+
+template< typename Matrix,
+ typename Preconditioner >
+BICGStabL< Matrix, Preconditioner >::BICGStabL()
+{
+ /****
+ * Clearing the shared pointer means that there is no
+ * preconditioner set.
+ */
+ this->preconditioner.clear();
+}
+
+template< typename Matrix,
+ typename Preconditioner >
+String
+BICGStabL< Matrix, Preconditioner >::getType() const
+{
+ return String( "BICGStabL< " ) +
+ this->matrix -> getType() + ", " +
+ this->preconditioner -> getType() + " >";
+}
+
+template< typename Matrix,
+ typename Preconditioner >
+void
+BICGStabL< Matrix, Preconditioner >::
+configSetup( Config::ConfigDescription& config,
+ const String& prefix )
+{
+ //IterativeSolver< RealType, IndexType >::configSetup( config, prefix );
+ config.addEntry< int >( prefix + "bicgstab-ell", "Number of Bi-CG iterations before the MR part starts.", 1 );
+ config.addEntry< bool >( prefix + "bicgstab-exact-residue", "Whether the BiCGstab should compute the exact residue in each step (true) or to use a cheap approximation (false).", false );
+}
+
+template< typename Matrix,
+ typename Preconditioner >
+bool
+BICGStabL< Matrix, Preconditioner >::
+setup( const Config::ParameterContainer& parameters,
+ const String& prefix )
+{
+ ell = parameters.getParameter< int >( "bicgstab-ell" );
+ exact_residue = parameters.getParameter< bool >( "bicgstab-exact-residue" );
+ return IterativeSolver< RealType, IndexType >::setup( parameters, prefix );
+}
+
+template< typename Matrix,
+ typename Preconditioner >
+void
+BICGStabL< Matrix, Preconditioner >::setMatrix( const MatrixPointer& matrix )
+{
+ this->matrix = matrix;
+}
+
+template< typename Matrix,
+ typename Preconditioner >
+void
+BICGStabL< Matrix, Preconditioner >::setPreconditioner( const PreconditionerPointer& preconditioner )
+{
+ this->preconditioner = preconditioner;
+}
+
+template< typename Matrix,
+ typename Preconditioner >
+ template< typename Vector, typename ResidueGetter >
+bool
+BICGStabL< Matrix, Preconditioner >::solve( const Vector& b, Vector& x )
+{
+ this->setSize( matrix->getRows() );
+
+ RealType alpha, beta, gamma, rho_0, rho_1, omega, b_norm;
+ DeviceVector r_0, r_j, r_i, u_0, Au, u;
+ r_0.bind( R.getData(), size );
+ u_0.bind( U.getData(), size );
+
+ auto matvec = [this]( const DeviceVector& src, DeviceVector& dst )
+ {
+ if( preconditioner ) {
+ matrix->vectorProduct( src, M_tmp );
+ preconditioner->solve( M_tmp, dst );
+ }
+ else {
+ matrix->vectorProduct( src, dst );
+ }
+ };
+
+ if( preconditioner ) {
+ preconditioner->solve( b, M_tmp );
+ b_norm = M_tmp.lpNorm( ( RealType ) 2.0 );
+
+ matrix->vectorProduct( x, M_tmp );
+ M_tmp.addVector( b, 1.0, -1.0 );
+ preconditioner->solve( M_tmp, r_0 );
+ }
+ else {
+ b_norm = b.lpNorm( 2.0 );
+ matrix->vectorProduct( x, r_0 );
+ r_0.addVector( b, 1.0, -1.0 );
+ }
+
+ sigma[ 0 ] = r_0.lpNorm( 2.0 );
+ if( std::isnan( sigma[ 0 ] ) )
+ throw std::runtime_error( "BiCGstab(ell): initial residue is NAN" );
+
+ r_ast = r_0;
+ r_ast /= sigma[ 0 ];
+ rho_0 = 1.0;
+ alpha = 0.0;
+ omega = 1.0;
+ u_0.setValue( 0.0 );
+
+ if( b_norm == 0.0 )
+ b_norm = 1.0;
+
+ this->resetIterations();
+ this->setResidue( sigma[ 0 ] / b_norm );
+
+ while( this->checkNextIteration() )
+ {
+ rho_0 = - omega * rho_0;
+
+ /****
+ * Bi-CG part
+ */
+ for( int j = 0; j < ell; j++ ) {
+ this->nextIteration();
+ r_j.bind( &R.getData()[ j * ldSize ], size );
+
+ rho_1 = r_ast.scalarProduct( r_j );
+ beta = alpha * rho_1 / rho_0;
+ rho_0 = rho_1;
+
+ /****
+ * U_[0:j] := R_[0:j] - beta * U_[0:j]
+ */
+ MatrixOperations< DeviceType >::
+ geam( size, j + 1,
+ 1.0, R.getData(), ldSize,
+ -beta, U.getData(), ldSize,
+ U.getData(), ldSize );
+
+ /****
+ * u_{j+1} = A u_j
+ */
+ u.bind( &U.getData()[ j * ldSize ], size );
+ Au.bind( &U.getData()[ (j + 1) * ldSize ], size );
+ matvec( u, Au );
+
+ gamma = r_ast.scalarProduct( Au );
+ alpha = rho_0 / gamma;
+
+ /****
+ * R_[0:j] := R_[0:j] - alpha * U_[1:j+1]
+ */
+ MatrixOperations< DeviceType >::
+ geam( size, j + 1,
+ 1.0, R.getData(), ldSize,
+ -alpha, U.getData() + ldSize, ldSize,
+ R.getData(), ldSize );
+
+ /****
+ * r_{j+1} = A r_j
+ */
+ r_j.bind( &R.getData()[ j * ldSize ], size );
+ r_i.bind( &R.getData()[ (j + 1) * ldSize ], size );
+ matvec( r_j, r_i );
+
+ /****
+ * x_0 := x_0 + alpha * u_0
+ */
+ x.addVector( u_0, alpha );
+ }
+
+ /****
+ * MGS part
+ */
+ for( int j = 1; j <= ell; j++ ) {
+ r_j.bind( &R.getData()[ j * ldSize ], size );
+
+ // MGS without reorthogonalization
+ for( int i = 1; i < j; i++ ) {
+ r_i.bind( &R.getData()[ i * ldSize ], size );
+ /****
+ * T_{i,j} = (r_i, r_j) / sigma_i
+ * r_j := r_j - T_{i,j} * r_i
+ */
+ const int ij = (i-1) + (j-1) * ell;
+ T[ ij ] = r_i.scalarProduct( r_j ) / sigma[ i ];
+ r_j.addVector( r_i, -T[ ij ] );
+ }
+
+ // MGS with reorthogonalization
+// for( int i = 1; i < j; i++ ) {
+// const int ij = (i-1) + (j-1) * ell;
+// T[ ij ] = 0.0;
+// }
+// for( int l = 0; l < 2; l++ )
+// for( int i = 1; i < j; i++ ) {
+// r_i.bind( &R.getData()[ i * ldSize ], size );
+// /****
+// * T_{i,j} = (r_i, r_j) / sigma_i
+// * r_j := r_j - T_{i,j} * r_i
+// */
+// const int ij = (i-1) + (j-1) * ell;
+// const RealType T_ij = r_i.scalarProduct( r_j ) / sigma[ i ];
+// T[ ij ] += T_ij;
+// r_j.addVector( r_i, -T_ij );
+// }
+
+ sigma[ j ] = r_j.scalarProduct( r_j );
+ g_1[ j ] = r_0.scalarProduct( r_j ) / sigma[ j ];
+ }
+
+ omega = g_1[ ell ];
+
+ /****
+ * g_0 = T^{-1} g_1
+ */
+ for( int j = ell; j >= 1; j-- ) {
+ g_0[ j ] = g_1[ j ];
+ for( int i = j + 1; i <= ell; i++ )
+ g_0[ j ] -= T[ (j-1) + (i-1) * ell ] * g_0[ i ];
+ }
+
+ /****
+ * g_2 = T * S * g_0,
+ * where S e_1 = 0, S e_j = e_{j-1} for j = 2, ... ell
+ */
+ for( int j = 1; j < ell; j++ ) {
+ g_2[ j ] = g_0[ j + 1 ];
+ for( int i = j + 1; i < ell; i++ )
+ g_2[ j ] += T[ (j-1) + (i-1) * ell ] * g_0[ i + 1 ];
+ }
+
+ /****
+ * Final updates
+ */
+ // x := x + R_[0:ell-1] * g_2
+ g_2[ 0 ] = g_0[ 1 ];
+ MatrixOperations< DeviceType >::gemv( size, ell,
+ 1.0, R.getData(), ldSize, g_2.getData(),
+ 1.0, x.getData() );
+ // r_0 := r_0 - R_[1:ell] * g_1_[1:ell]
+ MatrixOperations< DeviceType >::gemv( size, ell,
+ -1.0, R.getData() + ldSize, ldSize, &g_1[ 1 ],
+ 1.0, r_0.getData() );
+ // u_0 := u_0 - U_[1:ell] * g_0_[1:ell]
+ MatrixOperations< DeviceType >::gemv( size, ell,
+ -1.0, U.getData() + ldSize, ldSize, &g_0[ 1 ],
+ 1.0, u_0.getData() );
+
+ if( exact_residue ) {
+ /****
+ * Compute the exact preconditioned residue into the 's' vector.
+ */
+ if( preconditioner ) {
+ matrix->vectorProduct( x, M_tmp );
+ M_tmp.addVector( b, 1.0, -1.0 );
+ preconditioner->solve( M_tmp, res_tmp );
+ }
+ else {
+ matrix->vectorProduct( x, res_tmp );
+ res_tmp.addVector( b, 1.0, -1.0 );
+ }
+ sigma[ 0 ] = res_tmp.lpNorm( 2.0 );
+ this->setResidue( sigma[ 0 ] / b_norm );
+ }
+ else {
+ /****
+ * Use the "orthogonal residue vector" for stopping.
+ */
+ sigma[ 0 ] = r_0.lpNorm( 2.0 );
+ this->setResidue( sigma[ 0 ] / b_norm );
+ }
+ }
+
+ this->refreshSolverMonitor( true );
+ return this->checkConvergence();
+}
+
+template< typename Matrix,
+ typename Preconditioner >
+void
+BICGStabL< Matrix, Preconditioner >::setSize( IndexType size )
+{
+ this->size = ldSize = size;
+ R.setSize( (ell + 1) * ldSize );
+ U.setSize( (ell + 1) * ldSize );
+ r_ast.setSize( size );
+ M_tmp.setSize( size );
+ if( exact_residue )
+ res_tmp.setSize( size );
+ T.setSize( ell * ell );
+ sigma.setSize( ell + 1 );
+ g_0.setSize( ell + 1 );
+ g_1.setSize( ell + 1 );
+ g_2.setSize( ell + 1 );
+}
+
+} // namespace Linear
+} // namespace Solvers
+} // namespace TNL
diff --git a/src/TNL/Solvers/Linear/BICGStab_impl.h b/src/TNL/Solvers/Linear/BICGStab_impl.h
index a86af13ca545d6da4cf88cd7f967177028da9f21..adf744b50da033cd102b778189eacd68a22f9a38 100644
--- a/src/TNL/Solvers/Linear/BICGStab_impl.h
+++ b/src/TNL/Solvers/Linear/BICGStab_impl.h
@@ -55,7 +55,7 @@ BICGStab< Matrix, Preconditioner >::
setup( const Config::ParameterContainer& parameters,
const String& prefix )
{
- exact_residue = parameters.getParameter< int >( "bicgstab-exact-residue" );
+ exact_residue = parameters.getParameter< bool >( "bicgstab-exact-residue" );
return IterativeSolver< RealType, IndexType >::setup( parameters, prefix );
}
@@ -78,8 +78,7 @@ template< typename Matrix,
template< typename Vector, typename ResidueGetter >
bool BICGStab< Matrix, Preconditioner >::solve( const Vector& b, Vector& x )
{
- if( ! this->setSize( matrix->getRows() ) )
- return false;
+ this->setSize( matrix->getRows() );
RealType alpha, beta, omega, aux, rho, rho_old, b_norm;
@@ -193,21 +192,15 @@ bool BICGStab< Matrix, Preconditioner >::solve( const Vector& b, Vector& x )
template< typename Matrix,
typename Preconditioner >
-bool BICGStab< Matrix, Preconditioner > :: setSize( IndexType size )
+void BICGStab< Matrix, Preconditioner > :: setSize( IndexType size )
{
- if( ! r.setSize( size ) ||
- ! r_ast.setSize( size ) ||
- ! p.setSize( size ) ||
- ! s.setSize( size ) ||
- ! Ap.setSize( size ) ||
- ! As.setSize( size ) ||
- ! M_tmp.setSize( size ) )
- {
- std::cerr << "I am not able to allocate all supporting arrays for the BICGStab solver." << std::endl;
- return false;
- }
- return true;
-
+ r.setSize( size );
+ r_ast.setSize( size );
+ p.setSize( size );
+ s.setSize( size );
+ Ap.setSize( size );
+ As.setSize( size );
+ M_tmp.setSize( size );
}
} // namespace Linear
diff --git a/src/TNL/Solvers/Linear/CG.h b/src/TNL/Solvers/Linear/CG.h
index 2a7b2403da80dc9f0ad77d2ef3936a1fed8fb4a5..670303873d401e489ae55740b35905ab1914371b 100644
--- a/src/TNL/Solvers/Linear/CG.h
+++ b/src/TNL/Solvers/Linear/CG.h
@@ -14,7 +14,6 @@
#include <TNL/Object.h>
#include <TNL/SharedPointer.h>
#include <TNL/Containers/Vector.h>
-#include <TNL/Containers/SharedVector.h>
#include <TNL/Solvers/Linear/Preconditioners/Dummy.h>
#include <TNL/Solvers/IterativeSolver.h>
#include <TNL/Solvers/Linear/LinearResidueGetter.h>
@@ -60,11 +59,9 @@ class CG : public Object,
typename ResidueGetter = LinearResidueGetter< Matrix, Vector > >
bool solve( const Vector& b, Vector& x );
- ~CG();
-
protected:
- bool setSize( IndexType size );
+ void setSize( IndexType size );
Containers::Vector< RealType, DeviceType, IndexType > r, new_r, p, Ap;
diff --git a/src/TNL/Solvers/Linear/CG_impl.h b/src/TNL/Solvers/Linear/CG_impl.h
index e8b0f40056b37ee56e61329c075889b2372d0843..4c11c224e5c0b4f73c45a82146862473b9fc6488 100644
--- a/src/TNL/Solvers/Linear/CG_impl.h
+++ b/src/TNL/Solvers/Linear/CG_impl.h
@@ -77,7 +77,7 @@ bool
CG< Matrix, Preconditioner >::
solve( const Vector& b, Vector& x )
{
- if( ! this->setSize( matrix->getRows() ) ) return false;
+ this->setSize( matrix->getRows() );
this->resetIterations();
this->setResidue( this->getConvergenceResidue() + 1.0 );
@@ -147,28 +147,17 @@ solve( const Vector& b, Vector& x )
this->setResidue( ResidueGetter::getResidue( *matrix, x, b, bNorm ) );
this->refreshSolverMonitor( true );
return this->checkConvergence();
-};
-
-template< typename Matrix,
- typename Preconditioner >
-CG< Matrix, Preconditioner > :: ~CG()
-{
-};
+}
template< typename Matrix,
typename Preconditioner >
-bool CG< Matrix, Preconditioner > :: setSize( IndexType size )
+void CG< Matrix, Preconditioner > :: setSize( IndexType size )
{
- if( ! r. setSize( size ) ||
- ! new_r. setSize( size ) ||
- ! p. setSize( size ) ||
- ! Ap. setSize( size ) )
- {
- std::cerr << "I am not able to allocated all supporting arrays for the CG solver." << std::endl;
- return false;
- }
- return true;
-};
+ r.setSize( size );
+ new_r.setSize( size );
+ p.setSize( size );
+ Ap.setSize( size );
+}
} // namespace Linear
} // namespace Solvers
diff --git a/src/TNL/Solvers/Linear/CMakeLists.txt b/src/TNL/Solvers/Linear/CMakeLists.txt
old mode 100755
new mode 100644
index 01e22b367b23872fcddb76ba6d6c7a42d1af36a6..2321264a86c29551aeefffddcf930c9b3eac8077
--- a/src/TNL/Solvers/Linear/CMakeLists.txt
+++ b/src/TNL/Solvers/Linear/CMakeLists.txt
@@ -2,6 +2,8 @@ ADD_SUBDIRECTORY( Preconditioners )
SET( headers BICGStab.h
BICGStab_impl.h
+ BICGStabL.h
+ BICGStabL_impl.h
CG.h
CG_impl.h
CWYGMRES.h
diff --git a/src/TNL/Solvers/Linear/CWYGMRES.h b/src/TNL/Solvers/Linear/CWYGMRES.h
index 13f8e514e537f27e9070020f23d530cf5f350a82..fd7b4dbef3af1fe71202d6942403ba244a2d43d3 100644
--- a/src/TNL/Solvers/Linear/CWYGMRES.h
+++ b/src/TNL/Solvers/Linear/CWYGMRES.h
@@ -108,7 +108,7 @@ protected:
RealType& sn );
- bool setSize( IndexType _size, IndexType m );
+ void setSize( IndexType _size, IndexType m );
// single vectors
DeviceVector r, z, w, _M_tmp;
@@ -119,7 +119,7 @@ protected:
// host-only storage for Givens rotations and the least squares problem
HostVector cs, sn, H, s;
- IndexType size, ldSize, restarting;
+ IndexType size, ldSize, restarting_min, restarting_max, restarting_step_min, restarting_step_max;
MatrixPointer matrix;
PreconditionerPointer preconditioner;
diff --git a/src/TNL/Solvers/Linear/CWYGMRES_impl.h b/src/TNL/Solvers/Linear/CWYGMRES_impl.h
index f15f4ab83532737e8fb712b9ce3dce626feb6d41..01f966faa5b1df7298bdd603db893b79210ec310 100644
--- a/src/TNL/Solvers/Linear/CWYGMRES_impl.h
+++ b/src/TNL/Solvers/Linear/CWYGMRES_impl.h
@@ -14,6 +14,7 @@
#include <type_traits>
+#include <TNL/Exceptions/CudaSupportMissing.h>
#include <TNL/Containers/Algorithms/Multireduction.h>
#include <TNL/Matrices/MatrixOperations.h>
@@ -29,7 +30,10 @@ CWYGMRES< Matrix, Preconditioner >::
CWYGMRES()
: size( 0 ),
ldSize( 0 ),
- restarting( 10 )
+ restarting_min( 10 ),
+ restarting_max( 10 ),
+ restarting_step_min( 3 ),
+ restarting_step_max( 3 )
{
/****
* Clearing the shared pointer means that there is no
@@ -64,7 +68,10 @@ configSetup( Config::ConfigDescription& config,
const String& prefix )
{
//IterativeSolver< RealType, IndexType >::configSetup( config, prefix );
- config.addEntry< int >( prefix + "gmres-restarting", "Number of iterations after which the CWYGMRES restarts.", 10 );
+ config.addEntry< int >( prefix + "gmres-restarting-min", "Minimal number of iterations after which the GMRES restarts.", 10 );
+ config.addEntry< int >( prefix + "gmres-restarting-max", "Maximal number of iterations after which the GMRES restarts.", 10 );
+ config.addEntry< int >( prefix + "gmres-restarting-step-min", "Minimal adjusting step for the adaptivity of the GMRES restarting parameter.", 3 );
+ config.addEntry< int >( prefix + "gmres-restarting-step-max", "Maximal adjusting step for the adaptivity of the GMRES restarting parameter.", 3 );
}
template< typename Matrix,
@@ -75,7 +82,10 @@ setup( const Config::ParameterContainer& parameters,
const String& prefix )
{
IterativeSolver< RealType, IndexType >::setup( parameters, prefix );
- this->setRestarting( parameters.getParameter< int >( "gmres-restarting" ) );
+ restarting_min = parameters.getParameter< int >( "gmres-restarting-min" );
+ this->setRestarting( parameters.getParameter< int >( "gmres-restarting-max" ) );
+ restarting_step_min = parameters.getParameter< int >( "gmres-restarting-step-min" );
+ restarting_step_max = parameters.getParameter< int >( "gmres-restarting-step-max" );
return true;
}
@@ -87,7 +97,7 @@ setRestarting( IndexType rest )
{
if( size != 0 )
setSize( size, rest );
- restarting = rest;
+ restarting_max = rest;
}
template< typename Matrix,
@@ -115,18 +125,20 @@ bool
CWYGMRES< Matrix, Preconditioner >::
solve( const Vector& b, Vector& x )
{
- TNL_ASSERT( matrix, std::cerr << "No matrix was set in CWYGMRES. Call setMatrix() before solve()." << std::endl );
- if( restarting <= 0 )
+ TNL_ASSERT_TRUE( matrix, "No matrix was set in CWYGMRES. Call setMatrix() before solve()." );
+ if( restarting_min <= 0 || restarting_max <= 0 || restarting_min > restarting_max )
{
- std::cerr << "I have wrong value for the restarting of the CWYGMRES solver. It is set to " << restarting
- << ". Please set some positive value using the SetRestarting method." << std::endl;
+ std::cerr << "Wrong value for the GMRES restarting parameters: r_min = " << restarting_min
+ << ", r_max = " << restarting_max << std::endl;
return false;
}
- if( ! setSize( matrix -> getRows(), restarting ) )
+ if( restarting_step_min < 0 || restarting_step_max < 0 || restarting_step_min > restarting_step_max )
{
- std::cerr << "I am not able to allocate enough memory for the CWYGMRES solver. You may try to decrease the restarting parameter." << std::endl;
- return false;
+ std::cerr << "Wrong value for the GMRES restarting adjustment parameters: d_min = " << restarting_step_min
+ << ", d_max = " << restarting_step_max << std::endl;
+ return false;
}
+ setSize( matrix -> getRows(), restarting_max );
RealType normb( 0.0 ), beta( 0.0 );
/****
@@ -160,10 +172,37 @@ solve( const Vector& b, Vector& x )
this->resetIterations();
this->setResidue( beta / normb );
+ // parameters for the adaptivity of the restarting parameter
+ RealType beta_ratio = 1; // = beta / beta_ratio (small value indicates good convergence rate)
+ const RealType max_beta_ratio = 0.99; // = cos(8°) \approx 0.99
+ const RealType min_beta_ratio = 0.175; // = cos(80°) \approx 0.175
+ int restart_cycles = 0; // counter of restart cycles
+ int m = restarting_max; // current restarting parameter
+
DeviceVector vi, vk;
while( this->checkNextIteration() )
{
- const IndexType m = restarting;
+ // adaptivity of the restarting parameter
+ // reference: A.H. Baker, E.R. Jessup, Tz.V. Kolev - A simple strategy for varying the restart parameter in GMRES(m)
+ // http://www.sciencedirect.com/science/article/pii/S0377042709000132
+ if( restart_cycles > 0 ) {
+ if( beta_ratio > max_beta_ratio )
+ // near stagnation -> set maximum
+ m = restarting_max;
+ else if( beta_ratio >= min_beta_ratio ) {
+ // the step size is determined based on current m using linear interpolation
+ // between restarting_step_min and restarting_step_max
+ const int step = restarting_step_min + (float) ( restarting_step_max - restarting_step_min ) /
+ ( restarting_max - restarting_min ) *
+ ( m - restarting_min );
+ if( m - step >= restarting_min )
+ m -= step;
+ else
+ // set restarting_max when we hit restarting_min (see Baker et al. (2009))
+ m = restarting_max;
+ }
+// std::cerr << "restarting: cycle = " << restart_cycles << ", beta_ratio = " << beta_ratio << ", m = " << m << " " << std::endl;
+ }
/***
* z = r / | r | = 1.0 / beta * r
@@ -295,6 +334,7 @@ solve( const Vector& b, Vector& x )
/****
* r = M.solve(b - A * x);
*/
+ const RealType beta_old = beta;
if( preconditioner )
{
matrix->vectorProduct( x, _M_tmp );
@@ -313,6 +353,9 @@ solve( const Vector& b, Vector& x )
// cout << " beta = " << beta << endl;
// cout << "residue = " << beta / normb << endl;
+ // update parameters for the adaptivity of the restarting parameter
+ ++restart_cycles;
+ beta_ratio = beta / beta_old;
}
this->refreshSolverMonitor( true );
return this->checkConvergence();
@@ -375,9 +418,9 @@ hauseholder_generate( DeviceVector& Y,
z.getData(),
i,
size );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#else
- CudaSupportMissingMessage;
+ throw Exceptions::CudaSupportMissing();
#endif
}
@@ -401,7 +444,7 @@ hauseholder_generate( DeviceVector& Y,
// assuming it's stable enough...
const RealType t_i = 2.0 / (norm_yi * norm_yi);
- T[ i + i * (restarting + 1) ] = t_i;
+ T[ i + i * (restarting_max + 1) ] = t_i;
if( i > 0 ) {
// aux = Y_{i-1}^T * y_i
RealType aux[ i ];
@@ -421,9 +464,9 @@ hauseholder_generate( DeviceVector& Y,
// [T_i]_{0..i-1} = - T_{i-1} * t_i * aux
for( int k = 0; k < i; k++ ) {
- T[ k + i * (restarting + 1) ] = 0.0;
+ T[ k + i * (restarting_max + 1) ] = 0.0;
for( int j = k; j < i; j++ )
- T[ k + i * (restarting + 1) ] -= T[ k + j * (restarting + 1) ] * (t_i * aux[ j ]);
+ T[ k + i * (restarting_max + 1) ] -= T[ k + j * (restarting_max + 1) ] * (t_i * aux[ j ]);
}
}
}
@@ -441,7 +484,7 @@ hauseholder_apply_trunc( HostVector& out,
DeviceVector y_i;
y_i.bind( &Y.getData()[ i * ldSize ], size );
- const RealType aux = T[ i + i * (restarting + 1) ] * y_i.scalarProduct( z );
+ const RealType aux = T[ i + i * (restarting_max + 1) ] * y_i.scalarProduct( z );
if( std::is_same< DeviceType, Devices::Host >::value ) {
for( int k = 0; k <= i; k++ )
out[ k ] = z[ k ] - y_i[ k ] * aux;
@@ -449,9 +492,9 @@ hauseholder_apply_trunc( HostVector& out,
if( std::is_same< DeviceType, Devices::Cuda >::value ) {
// copy part of y_i to buffer on host
// here we duplicate the upper (m+1)x(m+1) submatrix of Y on host for fast access
- RealType* host_yi = &YL[ i * (restarting + 1) ];
+ RealType* host_yi = &YL[ i * (restarting_max + 1) ];
RealType host_z[ i + 1 ];
- if( ! Containers::Algorithms::ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< RealType, RealType, IndexType >( host_yi, y_i.getData(), restarting + 1 ) ||
+ if( ! Containers::Algorithms::ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< RealType, RealType, IndexType >( host_yi, y_i.getData(), restarting_max + 1 ) ||
! Containers::Algorithms::ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< RealType, RealType, IndexType >( host_z, z.getData(), i + 1 ) )
{
std::cerr << "Failed to copy part of device vectors y_i or z to host buffer." << std::endl;
@@ -480,7 +523,7 @@ hauseholder_cwy( DeviceVector& v,
if( std::is_same< DeviceType, Devices::Cuda >::value ) {
// the upper (m+1)x(m+1) submatrix of Y is duplicated on host for fast access
for( int k = 0; k <= i; k++ )
- aux[ k ] = YL[ i + k * (restarting + 1) ];
+ aux[ k ] = YL[ i + k * (restarting_max + 1) ];
}
// aux = T_i * aux
@@ -488,7 +531,7 @@ hauseholder_cwy( DeviceVector& v,
for( int k = 0; k <= i; k++ ) {
RealType aux2 = 0.0;
for( int j = k; j <= i; j++ )
- aux2 += T[ k + j * (restarting + 1) ] * aux[ j ];
+ aux2 += T[ k + j * (restarting_max + 1) ] * aux[ j ];
aux[ k ] = aux2;
}
@@ -530,7 +573,7 @@ hauseholder_cwy_transposed( DeviceVector& z,
for( int k = i; k >= 0; k-- ) {
RealType aux2 = 0.0;
for( int j = 0; j <= k; j++ )
- aux2 += T[ j + k * (restarting + 1) ] * aux[ j ];
+ aux2 += T[ j + k * (restarting_max + 1) ] * aux[ j ];
aux[ k ] = aux2;
}
@@ -629,9 +672,10 @@ applyPlaneRotation( RealType& dx,
template< typename Matrix,
typename Preconditioner >
-bool CWYGMRES< Matrix, Preconditioner > :: setSize( IndexType _size, IndexType m )
+void CWYGMRES< Matrix, Preconditioner > :: setSize( IndexType _size, IndexType m )
{
- if( size == _size && restarting == m ) return true;
+ if( size == _size && restarting_max == m )
+ return;
size = _size;
if( std::is_same< DeviceType, Devices::Cuda >::value )
// align each column to 256 bytes - optimal for CUDA
@@ -639,24 +683,19 @@ bool CWYGMRES< Matrix, Preconditioner > :: setSize( IndexType _size, IndexType m
else
// on the host, we add 1 to disrupt the cache false-sharing pattern
ldSize = roundToMultiple( size, 256 / sizeof( RealType ) ) + 1;
- restarting = m;
- if( ! r.setSize( size ) ||
- ! z.setSize( size ) ||
- ! w.setSize( size ) ||
- ! V.setSize( ldSize * ( restarting + 1 ) ) ||
- ! Y.setSize( ldSize * ( restarting + 1 ) ) ||
- ! T.setSize( (restarting + 1) * (restarting + 1) ) ||
- ! YL.setSize( (restarting + 1) * (restarting + 1) ) ||
- ! cs.setSize( restarting + 1 ) ||
- ! sn.setSize( restarting + 1 ) ||
- ! H.setSize( ( restarting + 1 ) * restarting ) ||
- ! s.setSize( restarting + 1 ) ||
- ! _M_tmp.setSize( size ) )
- {
- std::cerr << "I could not allocate all supporting arrays for the CWYGMRES solver." << std::endl;
- return false;
- }
- return true;
+ restarting_max = m;
+ r.setSize( size );
+ z.setSize( size );
+ w.setSize( size );
+ V.setSize( ldSize * ( m + 1 ) );
+ Y.setSize( ldSize * ( m + 1 ) );
+ T.setSize( (m + 1) * (m + 1) );
+ YL.setSize( (m + 1) * (m + 1) );
+ cs.setSize( m + 1 );
+ sn.setSize( m + 1 );
+ H.setSize( ( m + 1 ) * m );
+ s.setSize( m + 1 );
+ _M_tmp.setSize( size );
}
} // namespace Linear
diff --git a/src/TNL/Solvers/Linear/GMRES.h b/src/TNL/Solvers/Linear/GMRES.h
index 95fdbd9e3f9ceb80792ff7e6c90d537c13257a09..845f9399503c0981f0f608e9b685ce8be508b594 100644
--- a/src/TNL/Solvers/Linear/GMRES.h
+++ b/src/TNL/Solvers/Linear/GMRES.h
@@ -82,12 +82,12 @@ protected:
RealType& sn );
- bool setSize( IndexType _size, IndexType m );
+ void setSize( IndexType _size, IndexType m );
Containers::Vector< RealType, DeviceType, IndexType > _r, w, _v, _M_tmp;
Containers::Vector< RealType, Devices::Host, IndexType > _s, _cs, _sn, _H;
- IndexType size, restarting;
+ IndexType size, restarting_min, restarting_max, restarting_step_min, restarting_step_max;
MatrixPointer matrix;
diff --git a/src/TNL/Solvers/Linear/GMRES_impl.cpp b/src/TNL/Solvers/Linear/GMRES_impl.cpp
index 3bf16b3ee99df843aaa1f402497c377c7ca81f02..de639a5c219d0743951e0ebac8abb2a28b726864 100644
--- a/src/TNL/Solvers/Linear/GMRES_impl.cpp
+++ b/src/TNL/Solvers/Linear/GMRES_impl.cpp
@@ -8,6 +8,8 @@
/* See Copyright Notice in tnl/Copyright */
+#ifdef TEMPLATE_EXPLICIT_INSTANTIATION
+
#include <TNL/Solvers/Linear/GMRES.h>
#include <TNL/Matrices/CSR.h>
#include <TNL/Matrices/Ellpack.h>
@@ -53,3 +55,5 @@ template class GMRES< Matrices::Multidiagonal< double, Devices::Cuda, long int >
} // namespace Linear
} // namespace Solvers
} // namespace TNL
+
+#endif // #ifdef TEMPLATE_EXPLICIT_INSTANTIATION
diff --git a/src/TNL/Solvers/Linear/GMRES_impl.h b/src/TNL/Solvers/Linear/GMRES_impl.h
index 15c887c16d1aa9665fe0102e3317fbbd6b7d67a1..d3a20175926d9023228152f91a24ac04154fc80c 100644
--- a/src/TNL/Solvers/Linear/GMRES_impl.h
+++ b/src/TNL/Solvers/Linear/GMRES_impl.h
@@ -21,7 +21,10 @@ template< typename Matrix,
GMRES< Matrix, Preconditioner >::
GMRES()
: size( 0 ),
- restarting( 10 )
+ restarting_min( 10 ),
+ restarting_max( 10 ),
+ restarting_step_min( 3 ),
+ restarting_step_max( 3 )
{
/****
* Clearing the shared pointer means that there is no
@@ -56,7 +59,10 @@ configSetup( Config::ConfigDescription& config,
const String& prefix )
{
//IterativeSolver< RealType, IndexType >::configSetup( config, prefix );
- config.addEntry< int >( prefix + "gmres-restarting", "Number of iterations after which the GMRES restarts.", 10 );
+ config.addEntry< int >( prefix + "gmres-restarting-min", "Minimal number of iterations after which the GMRES restarts.", 10 );
+ config.addEntry< int >( prefix + "gmres-restarting-max", "Maximal number of iterations after which the GMRES restarts.", 10 );
+ config.addEntry< int >( prefix + "gmres-restarting-step-min", "Minimal adjusting step for the adaptivity of the GMRES restarting parameter.", 3 );
+ config.addEntry< int >( prefix + "gmres-restarting-step-max", "Maximal adjusting step for the adaptivity of the GMRES restarting parameter.", 3 );
}
template< typename Matrix,
@@ -67,7 +73,10 @@ setup( const Config::ParameterContainer& parameters,
const String& prefix )
{
IterativeSolver< RealType, IndexType >::setup( parameters, prefix );
- this->setRestarting( parameters.getParameter< int >( "gmres-restarting" ) );
+ restarting_min = parameters.getParameter< int >( "gmres-restarting-min" );
+ this->setRestarting( parameters.getParameter< int >( "gmres-restarting-max" ) );
+ restarting_step_min = parameters.getParameter< int >( "gmres-restarting-step-min" );
+ restarting_step_max = parameters.getParameter< int >( "gmres-restarting-step-max" );
return true;
}
@@ -79,7 +88,7 @@ setRestarting( IndexType rest )
{
if( size != 0 )
setSize( size, rest );
- restarting = rest;
+ restarting_max = rest;
}
template< typename Matrix,
@@ -107,18 +116,20 @@ bool
GMRES< Matrix, Preconditioner >::
solve( const Vector& b, Vector& x )
{
- TNL_ASSERT( matrix, std::cerr << "No matrix was set in GMRES. Call setMatrix() before solve()." << std::endl );
- if( restarting <= 0 )
+ TNL_ASSERT_TRUE( matrix, "No matrix was set in GMRES. Call setMatrix() before solve()." );
+ if( restarting_min <= 0 || restarting_max <= 0 || restarting_min > restarting_max )
{
- std::cerr << "I have wrong value for the restarting of the GMRES solver. It is set to " << restarting
- << ". Please set some positive value using the SetRestarting method." << std::endl;
+ std::cerr << "Wrong value for the GMRES restarting parameters: r_min = " << restarting_min
+ << ", r_max = " << restarting_max << std::endl;
return false;
}
- if( ! setSize( matrix -> getRows(), restarting ) )
+ if( restarting_step_min < 0 || restarting_step_max < 0 || restarting_step_min > restarting_step_max )
{
- std::cerr << "I am not able to allocate enough memory for the GMRES solver. You may try to decrease the restarting parameter." << std::endl;
- return false;
+ std::cerr << "Wrong value for the GMRES restarting adjustment parameters: d_min = " << restarting_step_min
+ << ", d_max = " << restarting_step_max << std::endl;
+ return false;
}
+ setSize( matrix -> getRows(), restarting_max );
IndexType _size = size;
@@ -161,10 +172,38 @@ solve( const Vector& b, Vector& x )
this->resetIterations();
this->setResidue( beta / normb );
+ // parameters for the adaptivity of the restarting parameter
+ RealType beta_ratio = 1; // = beta / beta_ratio (small value indicates good convergence rate)
+ const RealType max_beta_ratio = 0.99; // = cos(8°) \approx 0.99
+ const RealType min_beta_ratio = 0.175; // = cos(80°) \approx 0.175
+ int restart_cycles = 0; // counter of restart cycles
+ int m = restarting_max; // current restarting parameter
+
Containers::Vector< RealType, DeviceType, IndexType > vi, vk;
while( this->checkNextIteration() )
{
- const IndexType m = restarting;
+ // adaptivity of the restarting parameter
+ // reference: A.H. Baker, E.R. Jessup, Tz.V. Kolev - A simple strategy for varying the restart parameter in GMRES(m)
+ // http://www.sciencedirect.com/science/article/pii/S0377042709000132
+ if( restarting_max > restarting_min && restart_cycles > 0 ) {
+ if( beta_ratio > max_beta_ratio )
+ // near stagnation -> set maximum
+ m = restarting_max;
+ else if( beta_ratio >= min_beta_ratio ) {
+ // the step size is determined based on current m using linear interpolation
+ // between restarting_step_min and restarting_step_max
+ const int step = restarting_step_min + (float) ( restarting_step_max - restarting_step_min ) /
+ ( restarting_max - restarting_min ) *
+ ( m - restarting_min );
+ if( m - step >= restarting_min )
+ m -= step;
+ else
+ // set restarting_max when we hit restarting_min (see Baker et al. (2009))
+ m = restarting_max;
+ }
+// std::cerr << "restarting: cycle = " << restart_cycles << ", beta_ratio = " << beta_ratio << ", m = " << m << " " << std::endl;
+ }
+
for( IndexType i = 0; i < m + 1; i ++ )
H[ i ] = s[ i ] = cs[ i ] = sn[ i ] = 0.0;
@@ -279,6 +318,7 @@ solve( const Vector& b, Vector& x )
/****
* r = M.solve(b - A * x);
*/
+ const RealType beta_old = beta;
beta = 0.0;
if( preconditioner )
{
@@ -299,6 +339,9 @@ solve( const Vector& b, Vector& x )
//cout << " beta = " << beta << std::endl;
//cout << "residue = " << beta / normb << std::endl;
+ // update parameters for the adaptivity of the restarting parameter
+ ++restart_cycles;
+ beta_ratio = beta / beta_old;
}
this->refreshSolverMonitor( true );
return this->checkConvergence();
@@ -385,26 +428,22 @@ applyPlaneRotation( RealType& dx,
template< typename Matrix,
typename Preconditioner >
-bool
+void
GMRES< Matrix, Preconditioner >::
setSize( IndexType _size, IndexType m )
{
- if( size == _size && restarting == m ) return true;
+ if( size == _size && restarting_max == m )
+ return;
size = _size;
- restarting = m;
- if( ! _r.setSize( size ) ||
- ! w.setSize( size ) ||
- ! _s.setSize( restarting + 1 ) ||
- ! _cs.setSize( restarting + 1 ) ||
- ! _sn.setSize( restarting + 1 ) ||
- ! _v.setSize( size * ( restarting + 1 ) ) ||
- ! _H.setSize( ( restarting + 1 ) * restarting ) ||
- ! _M_tmp.setSize( size ) )
- {
- std::cerr << "I could not allocate all supporting arrays for the GMRES solver." << std::endl;
- return false;
- }
- return true;
+ restarting_max = m;
+ _r.setSize( size );
+ w.setSize( size );
+ _s.setSize( m + 1 );
+ _cs.setSize( m + 1 );
+ _sn.setSize( m + 1 );
+ _v.setSize( size * ( m + 1 ) );
+ _H.setSize( ( m + 1 ) * m );
+ _M_tmp.setSize( size );
}
} // namespace Linear
diff --git a/src/TNL/Solvers/Linear/LinearResidueGetter_impl.h b/src/TNL/Solvers/Linear/LinearResidueGetter_impl.h
index ccd70971c4a9937043f87a2c3d3005f049098b25..db81e861f7cd850ed151e305eb81c6915def7f4f 100644
--- a/src/TNL/Solvers/Linear/LinearResidueGetter_impl.h
+++ b/src/TNL/Solvers/Linear/LinearResidueGetter_impl.h
@@ -10,6 +10,10 @@
#pragma once
+#include <cmath>
+
+#include <TNL/Solvers/Linear/LinearResidueGetter.h>
+
namespace TNL {
namespace Solvers {
namespace Linear {
diff --git a/src/TNL/Solvers/Linear/Preconditioners/CMakeLists.txt b/src/TNL/Solvers/Linear/Preconditioners/CMakeLists.txt
old mode 100755
new mode 100644
index d14c10ad596625ba7f069f1697ec0737b8622212..43541c7ec74936497212c8436d95a2cbe03c3fcd
--- a/src/TNL/Solvers/Linear/Preconditioners/CMakeLists.txt
+++ b/src/TNL/Solvers/Linear/Preconditioners/CMakeLists.txt
@@ -1,6 +1,8 @@
SET( headers Dummy.h
Diagonal.h
Diagonal_impl.h
+ ILU0.h
+ ILU0_impl.h
)
INSTALL( FILES ${headers} DESTINATION include/tnl-${tnlVersion}/TNL/Solvers/Linear/Preconditioners )
diff --git a/src/TNL/Solvers/Linear/Preconditioners/Diagonal_impl.h b/src/TNL/Solvers/Linear/Preconditioners/Diagonal_impl.h
index 991b6a4858f9dcb7f58359835964d12d91c09c43..0c77239443c054c4429416c5c654ab53d8419a9e 100644
--- a/src/TNL/Solvers/Linear/Preconditioners/Diagonal_impl.h
+++ b/src/TNL/Solvers/Linear/Preconditioners/Diagonal_impl.h
@@ -57,7 +57,8 @@ update( const MatrixPointer& matrix )
{
// std::cout << getType() << "->setMatrix()" << std::endl;
- TNL_ASSERT( matrix->getRows() > 0 && matrix->getRows() == matrix->getColumns(), );
+ TNL_ASSERT_GT( matrix->getRows(), 0, "empty matrix" );
+ TNL_ASSERT_EQ( matrix->getRows(), matrix->getColumns(), "matrix must be square" );
if( diagonal.getSize() != matrix->getRows() )
diagonal.setSize( matrix->getRows() );
@@ -81,7 +82,7 @@ update( const MatrixPointer& matrix )
&matrix.template getData< Devices::Cuda >(),
diagonal.getData(),
size );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
}
@@ -112,7 +113,7 @@ solve( const Vector1& b, Vector2& x ) const
x.getData(),
size );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
return true;
diff --git a/src/TNL/Solvers/Linear/Preconditioners/Dummy.h b/src/TNL/Solvers/Linear/Preconditioners/Dummy.h
index 5629568c78b0bf1571c6a2d25a46d4bdc0f3199e..2a7283b22bd31f881b2d8a320b09ddc50b3bbe8a 100644
--- a/src/TNL/Solvers/Linear/Preconditioners/Dummy.h
+++ b/src/TNL/Solvers/Linear/Preconditioners/Dummy.h
@@ -29,8 +29,7 @@ class Dummy
template< typename Vector1, typename Vector2 >
bool solve( const Vector1& b, Vector2& x ) const
{
- TNL_ASSERT( false,
- std::cerr << "The solve() method of a dummy preconditioner should not be called." << std::endl; );
+ TNL_ASSERT_TRUE( false, "The solve() method of a dummy preconditioner should not be called." );
return true;
}
diff --git a/src/TNL/Solvers/Linear/Preconditioners/ILU0.h b/src/TNL/Solvers/Linear/Preconditioners/ILU0.h
new file mode 100644
index 0000000000000000000000000000000000000000..651f36b73fc3f8773d2063ae7d0dec53f96f82f5
--- /dev/null
+++ b/src/TNL/Solvers/Linear/Preconditioners/ILU0.h
@@ -0,0 +1,206 @@
+/***************************************************************************
+ ILU0.h - description
+ -------------------
+ begin : Dec 24, 2016
+ copyright : (C) 2016 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Jakub Klinkovsky
+
+#pragma once
+
+#include <type_traits>
+
+#include <TNL/Object.h>
+#include <TNL/Containers/Vector.h>
+#include <TNL/Matrices/CSR.h>
+
+#ifdef HAVE_CUDA
+#include <cusparse.h>
+#endif
+
+namespace TNL {
+namespace Solvers {
+namespace Linear {
+namespace Preconditioners {
+
+template< typename Real, typename Device, typename Index >
+class ILU0
+{};
+
+template< typename Real, typename Index >
+class ILU0< Real, Devices::Host, Index >
+{
+public:
+ typedef Real RealType;
+ typedef Devices::Host DeviceType;
+ typedef Index IndexType;
+
+ template< typename MatrixPointer >
+ void update( const MatrixPointer& matrixPointer );
+
+ template< typename Vector1, typename Vector2 >
+ bool solve( const Vector1& b, Vector2& x ) const;
+
+ String getType() const
+ {
+ return String( "ILU0" );
+ }
+
+protected:
+// Matrices::CSR< RealType, DeviceType, IndexType > A;
+ Matrices::CSR< RealType, DeviceType, IndexType > L;
+ Matrices::CSR< RealType, DeviceType, IndexType > U;
+};
+
+template<>
+class ILU0< double, Devices::Cuda, int >
+{
+public:
+ using RealType = double;
+ using DeviceType = Devices::Cuda;
+ using IndexType = int;
+
+ ILU0()
+ {
+#ifdef HAVE_CUDA
+ cusparseCreate( &handle );
+#endif
+ }
+
+ template< typename MatrixPointer >
+ void update( const MatrixPointer& matrixPointer );
+
+ template< typename Vector1, typename Vector2 >
+ bool solve( const Vector1& b, Vector2& x ) const;
+
+ String getType() const
+ {
+ return String( "ILU0" );
+ }
+
+ ~ILU0()
+ {
+#ifdef HAVE_CUDA
+ resetMatrices();
+ cusparseDestroy( handle );
+#endif
+ }
+
+protected:
+#ifdef HAVE_CUDA
+ Matrices::CSR< RealType, DeviceType, IndexType > A;
+ Containers::Vector< RealType, DeviceType, IndexType > y;
+
+ cusparseHandle_t handle;
+
+ cusparseMatDescr_t descr_A = 0;
+ cusparseMatDescr_t descr_L = 0;
+ cusparseMatDescr_t descr_U = 0;
+ csrilu02Info_t info_A = 0;
+ csrsv2Info_t info_L = 0;
+ csrsv2Info_t info_U = 0;
+
+ const cusparseSolvePolicy_t policy_A = CUSPARSE_SOLVE_POLICY_USE_LEVEL;
+ const cusparseSolvePolicy_t policy_L = CUSPARSE_SOLVE_POLICY_USE_LEVEL;
+ const cusparseSolvePolicy_t policy_U = CUSPARSE_SOLVE_POLICY_USE_LEVEL;
+ const cusparseOperation_t trans_L = CUSPARSE_OPERATION_NON_TRANSPOSE;
+ const cusparseOperation_t trans_U = CUSPARSE_OPERATION_NON_TRANSPOSE;
+
+ Containers::Array< char, DeviceType, int > pBuffer;
+
+ // scaling factor for triangular solves
+ const double alpha = 1.0;
+
+ void resetMatrices()
+ {
+ if( descr_A ) {
+ cusparseDestroyMatDescr( descr_A );
+ descr_A = 0;
+ }
+ if( descr_L ) {
+ cusparseDestroyMatDescr( descr_L );
+ descr_L = 0;
+ }
+ if( descr_U ) {
+ cusparseDestroyMatDescr( descr_U );
+ descr_U = 0;
+ }
+ if( info_A ) {
+ cusparseDestroyCsrilu02Info( info_A );
+ info_A = 0;
+ }
+ if( info_L ) {
+ cusparseDestroyCsrsv2Info( info_L );
+ info_L = 0;
+ }
+ if( info_U ) {
+ cusparseDestroyCsrsv2Info( info_U );
+ info_U = 0;
+ }
+ pBuffer.reset();
+ }
+
+ // TODO: extend Matrices::copySparseMatrix accordingly
+ template< typename Matrix,
+ typename = typename std::enable_if< ! std::is_same< DeviceType, typename Matrix::DeviceType >::value >::type >
+ void copyMatrix( const Matrix& matrix )
+ {
+ typename Matrix::CudaType A_tmp;
+ A_tmp = matrix;
+ Matrices::copySparseMatrix( A, A_tmp );
+ }
+
+ template< typename Matrix,
+ typename = typename std::enable_if< std::is_same< DeviceType, typename Matrix::DeviceType >::value >::type,
+ typename = void >
+ void copyMatrix( const Matrix& matrix )
+ {
+ Matrices::copySparseMatrix( A, matrix );
+ }
+#endif
+};
+
+#ifdef HAVE_MIC
+template< typename Real, typename Index >
+class ILU0< Real, Devices::MIC, Index >
+{
+public:
+ typedef Real RealType;
+ typedef Devices::MIC DeviceType;
+ typedef Index IndexType;
+
+ template< typename MatrixPointer >
+ void update( const MatrixPointer& matrixPointer )
+ {
+ throw std::runtime_error("Not Iplemented yet for MIC");
+ }
+
+ template< typename Vector1, typename Vector2 >
+ bool solve( const Vector1& b, Vector2& x ) const
+ {
+ throw std::runtime_error("Not Iplemented yet for MIC");
+ }
+
+ String getType() const
+ {
+ return String( "ILU0" );
+ }
+
+protected:
+// Matrices::CSR< RealType, DeviceType, IndexType > A;
+ Matrices::CSR< RealType, DeviceType, IndexType > L;
+ Matrices::CSR< RealType, DeviceType, IndexType > U;
+};
+
+#endif
+
+} // namespace Preconditioners
+} // namespace Linear
+} // namespace Solvers
+} // namespace TNL
+
+#include <TNL/Solvers/Linear/Preconditioners/ILU0_impl.h>
diff --git a/src/TNL/Solvers/Linear/Preconditioners/ILU0_impl.h b/src/TNL/Solvers/Linear/Preconditioners/ILU0_impl.h
new file mode 100644
index 0000000000000000000000000000000000000000..58d1c4ad8c9cee8572e05e83de81105b5a96086f
--- /dev/null
+++ b/src/TNL/Solvers/Linear/Preconditioners/ILU0_impl.h
@@ -0,0 +1,300 @@
+/***************************************************************************
+ ILU0_impl.h - description
+ -------------------
+ begin : Dec 24, 2016
+ copyright : (C) 2016 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+// Implemented by: Jakub Klinkovsky
+
+#pragma once
+
+#include "ILU0.h"
+
+#include <TNL/Exceptions/CudaSupportMissing.h>
+
+namespace TNL {
+namespace Solvers {
+namespace Linear {
+namespace Preconditioners {
+
+template< typename Real, typename Index >
+ template< typename MatrixPointer >
+void
+ILU0< Real, Devices::Host, Index >::
+update( const MatrixPointer& matrixPointer )
+{
+ TNL_ASSERT_GT( matrixPointer->getRows(), 0, "empty matrix" );
+ TNL_ASSERT_EQ( matrixPointer->getRows(), matrixPointer->getColumns(), "matrix must be square" );
+
+ const IndexType N = matrixPointer->getRows();
+
+ L.setDimensions( N, N );
+ U.setDimensions( N, N );
+
+ // copy row lengths
+ typename decltype(L)::CompressedRowLengthsVector L_rowLengths;
+ typename decltype(U)::CompressedRowLengthsVector U_rowLengths;
+ L_rowLengths.setSize( N );
+ U_rowLengths.setSize( N );
+ for( IndexType i = 0; i < N; i++ ) {
+ const auto row = matrixPointer->getRow( i );
+ const auto max_length = matrixPointer->getRowLength( i );
+ IndexType L_entries = 0;
+ IndexType U_entries = 0;
+ for( IndexType j = 0; j < max_length; j++ ) {
+ const auto column = row.getElementColumn( j );
+ if( column < i )
+ L_entries++;
+ else if( column < N )
+ U_entries++;
+ else
+ break;
+ }
+ L_rowLengths[ i ] = L_entries;
+ U_rowLengths[ N - 1 - i ] = U_entries;
+ }
+ L.setCompressedRowLengths( L_rowLengths );
+ U.setCompressedRowLengths( U_rowLengths );
+
+ // Incomplete LU factorization
+ // The factors L and U are stored separately and the rows of U are reversed.
+ for( IndexType i = 0; i < N; i++ ) {
+ // copy all non-zero entries from A into L and U
+ const auto max_length = matrixPointer->getRowLength( i );
+ IndexType columns[ max_length ];
+ RealType values[ max_length ];
+ matrixPointer->getRowFast( i, columns, values );
+
+ const auto L_entries = L_rowLengths[ i ];
+ const auto U_entries = U_rowLengths[ N - 1 - i ];
+ L.setRow( i, columns, values, L_entries );
+ U.setRow( N - 1 - i, &columns[ L_entries ], &values[ L_entries ], U_entries );
+
+ // this condition is to avoid segfaults on empty L.getRow( i )
+ if( L_entries > 0 ) {
+ const auto L_i = L.getRow( i );
+ const auto U_i = U.getRow( N - 1 - i );
+
+ // loop for k = 0, ..., i - 2; but only over the non-zero entries
+ for( IndexType c_k = 0; c_k < L_entries; c_k++ ) {
+ const auto k = L_i.getElementColumn( c_k );
+
+ auto L_ik = L.getElementFast( i, k ) / U.getElementFast( N - 1 - k, k );
+ L.setElement( i, k, L_ik );
+
+ // loop for j = k+1, ..., N-1; but only over the non-zero entries
+ // and split into two loops over L and U separately
+ for( IndexType c_j = c_k + 1; c_j < L_entries; c_j++ ) {
+ const auto j = L_i.getElementColumn( c_j );
+ const auto L_ij = L.getElementFast( i, j ) - L_ik * U.getElementFast( N - 1 - k, j );
+ L.setElement( i, j, L_ij );
+ }
+ for( IndexType c_j = 0; c_j < U_entries; c_j++ ) {
+ const auto j = U_i.getElementColumn( c_j );
+ const auto U_ij = U.getElementFast( N - 1 - i, j ) - L_ik * U.getElementFast( N - 1 - k, j );
+ U.setElement( N - 1 - i, j, U_ij );
+ }
+ }
+ }
+ }
+}
+
+template< typename Real, typename Index >
+ template< typename Vector1, typename Vector2 >
+bool
+ILU0< Real, Devices::Host, Index >::
+solve( const Vector1& b, Vector2& x ) const
+{
+ TNL_ASSERT_EQ( b.getSize(), L.getRows(), "wrong size of the right hand side" );
+ TNL_ASSERT_EQ( x.getSize(), L.getRows(), "wrong size of the solution vector" );
+
+ const IndexType N = x.getSize();
+
+ // Step 1: solve y from Ly = b
+ for( IndexType i = 0; i < N; i++ ) {
+ x[ i ] = b[ i ];
+
+ const auto L_entries = L.getRowLength( i );
+
+ // this condition is to avoid segfaults on empty L.getRow( i )
+ if( L_entries > 0 ) {
+ const auto L_i = L.getRow( i );
+
+ // loop for j = 0, ..., i - 1; but only over the non-zero entries
+ for( IndexType c_j = 0; c_j < L_entries; c_j++ ) {
+ const auto j = L_i.getElementColumn( c_j );
+ x[ i ] -= L_i.getElementValue( c_j ) * x[ j ];
+ }
+ }
+ }
+
+ // Step 2: solve x from Ux = y
+ for( IndexType i = N - 1; i >= 0; i-- ) {
+ const IndexType U_idx = N - 1 - i;
+
+ const auto U_entries = U.getRowLength( U_idx );
+ const auto U_i = U.getRow( U_idx );
+
+ const auto U_ii = U_i.getElementValue( 0 );
+
+ // loop for j = i+1, ..., N-1; but only over the non-zero entries
+ for( IndexType c_j = 1; c_j < U_entries ; c_j++ ) {
+ const auto j = U_i.getElementColumn( c_j );
+ x[ i ] -= U_i.getElementValue( c_j ) * x[ j ];
+ }
+
+ x[ i ] /= U_ii;
+ }
+
+ return true;
+}
+
+
+ template< typename MatrixPointer >
+void
+ILU0< double, Devices::Cuda, int >::
+update( const MatrixPointer& matrixPointer )
+{
+#ifdef HAVE_CUDA
+ // TODO: only numerical factorization has to be done every time, split the rest into separate "setup" method which is called less often
+ resetMatrices();
+
+ // Note: the decomposition will be in-place, matrices L and U will have the
+ // storage of A
+ copyMatrix( *matrixPointer );
+
+ const int m = A.getRows();
+ const int nnz = A.getValues().getSize();
+
+ y.setSize( m );
+
+ // create matrix descriptors
+ cusparseCreateMatDescr( &descr_A );
+ cusparseSetMatIndexBase( descr_A, CUSPARSE_INDEX_BASE_ZERO );
+ cusparseSetMatType( descr_A, CUSPARSE_MATRIX_TYPE_GENERAL );
+
+ cusparseCreateMatDescr( &descr_L );
+ cusparseSetMatIndexBase( descr_L, CUSPARSE_INDEX_BASE_ZERO );
+ cusparseSetMatType( descr_L, CUSPARSE_MATRIX_TYPE_GENERAL );
+ cusparseSetMatFillMode( descr_L, CUSPARSE_FILL_MODE_LOWER );
+ cusparseSetMatDiagType( descr_L, CUSPARSE_DIAG_TYPE_UNIT );
+
+ cusparseCreateMatDescr( &descr_U);
+ cusparseSetMatIndexBase( descr_U, CUSPARSE_INDEX_BASE_ZERO );
+ cusparseSetMatType( descr_U, CUSPARSE_MATRIX_TYPE_GENERAL );
+ cusparseSetMatFillMode( descr_U, CUSPARSE_FILL_MODE_UPPER );
+ cusparseSetMatDiagType( descr_U, CUSPARSE_DIAG_TYPE_NON_UNIT );
+
+ // create info structures
+ cusparseCreateCsrilu02Info( &info_A );
+ cusparseCreateCsrsv2Info( &info_L );
+ cusparseCreateCsrsv2Info( &info_U );
+
+ // query how much memory will be needed in csrilu02 and csrsv2, and allocate the buffer
+ int pBufferSize_A, pBufferSize_L, pBufferSize_U;
+ cusparseDcsrilu02_bufferSize( handle, m, nnz, descr_A,
+ A.getValues().getData(),
+ A.getRowPointers().getData(),
+ A.getColumnIndexes().getData(),
+ info_A, &pBufferSize_A );
+ cusparseDcsrsv2_bufferSize( handle, trans_L, m, nnz, descr_L,
+ A.getValues().getData(),
+ A.getRowPointers().getData(),
+ A.getColumnIndexes().getData(),
+ info_L, &pBufferSize_L );
+ cusparseDcsrsv2_bufferSize( handle, trans_U, m, nnz, descr_U,
+ A.getValues().getData(),
+ A.getRowPointers().getData(),
+ A.getColumnIndexes().getData(),
+ info_U, &pBufferSize_U );
+ const int pBufferSize = max( pBufferSize_A, max( pBufferSize_L, pBufferSize_U ) );
+ pBuffer.setSize( pBufferSize );
+
+ // Symbolic analysis of the incomplete LU decomposition
+ cusparseDcsrilu02_analysis( handle, m, nnz, descr_A,
+ A.getValues().getData(),
+ A.getRowPointers().getData(),
+ A.getColumnIndexes().getData(),
+ info_A, policy_A, pBuffer.getData() );
+ int structural_zero;
+ cusparseStatus_t
+ status = cusparseXcsrilu02_zeroPivot( handle, info_A, &structural_zero );
+ if( CUSPARSE_STATUS_ZERO_PIVOT == status ) {
+ std::cerr << "A(" << structural_zero << ", " << structural_zero << ") is missing." << std::endl;
+ throw 1;
+ }
+
+ // Analysis for the triangular solves for L and U
+ // Trick: the lower (upper) triangular part of A has the same sparsity
+ // pattern as L (U), so we can do the analysis for csrsv2 on the matrix A.
+ cusparseDcsrsv2_analysis( handle, trans_L, m, nnz, descr_L,
+ A.getValues().getData(),
+ A.getRowPointers().getData(),
+ A.getColumnIndexes().getData(),
+ info_L, policy_L, pBuffer.getData() );
+ cusparseDcsrsv2_analysis( handle, trans_U, m, nnz, descr_U,
+ A.getValues().getData(),
+ A.getRowPointers().getData(),
+ A.getColumnIndexes().getData(),
+ info_U, policy_U, pBuffer.getData() );
+
+ // Numerical incomplete LU decomposition
+ cusparseDcsrilu02( handle, m, nnz, descr_A,
+ A.getValues().getData(),
+ A.getRowPointers().getData(),
+ A.getColumnIndexes().getData(),
+ info_A, policy_A, pBuffer.getData() );
+ int numerical_zero;
+ status = cusparseXcsrilu02_zeroPivot( handle, info_A, &numerical_zero );
+ if( CUSPARSE_STATUS_ZERO_PIVOT == status ) {
+ std::cerr << "A(" << numerical_zero << ", " << numerical_zero << ") is zero." << std::endl;
+ throw 1;
+ }
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
+}
+
+ template< typename Vector1, typename Vector2 >
+bool
+ILU0< double, Devices::Cuda, int >::
+solve( const Vector1& b, Vector2& x ) const
+{
+#ifdef HAVE_CUDA
+ const int m = A.getRows();
+ const int nnz = A.getValues().getSize();
+
+ // Step 1: solve y from Ly = b
+ cusparseDcsrsv2_solve( handle, trans_L, m, nnz, &alpha, descr_L,
+ A.getValues().getData(),
+ A.getRowPointers().getData(),
+ A.getColumnIndexes().getData(),
+ info_L,
+ b.getData(),
+ (RealType*) y.getData(),
+ policy_L, (void*) pBuffer.getData() );
+
+ // Step 2: solve x from Ux = y
+ cusparseDcsrsv2_solve( handle, trans_U, m, nnz, &alpha, descr_U,
+ A.getValues().getData(),
+ A.getRowPointers().getData(),
+ A.getColumnIndexes().getData(),
+ info_U,
+ y.getData(),
+ x.getData(),
+ policy_U, (void*) pBuffer.getData() );
+
+ return true;
+#else
+ throw Exceptions::CudaSupportMissing();
+#endif
+}
+
+} // namespace Preconditioners
+} // namespace Linear
+} // namespace Solvers
+} // namespace TNL
diff --git a/src/TNL/Solvers/Linear/SOR.h b/src/TNL/Solvers/Linear/SOR.h
index d1159884c1dc98883d782db832c936da71e03186..fe36b65932aa0e4eed149a3e1f46a6acc9560c73 100644
--- a/src/TNL/Solvers/Linear/SOR.h
+++ b/src/TNL/Solvers/Linear/SOR.h
@@ -10,7 +10,6 @@
#pragma once
-#include <math.h>
#include <TNL/Object.h>
#include <TNL/SharedPointer.h>
#include <TNL/Solvers/Linear/Preconditioners/Dummy.h>
@@ -61,8 +60,6 @@ class SOR : public Object,
typename ResidueGetter = LinearResidueGetter< Matrix, Vector > >
bool solve( const Vector& b, Vector& x );
- ~SOR();
-
protected:
RealType omega;
@@ -76,4 +73,3 @@ class SOR : public Object,
} // namespace TNL
#include <TNL/Solvers/Linear/SOR_impl.h>
-
diff --git a/src/TNL/Solvers/Linear/SOR_impl.cpp b/src/TNL/Solvers/Linear/SOR_impl.cpp
index 294b819bcaa73724f0b289f7914fa7e1e2171084..4f4231e18cb7bb517ef09abfe57213e3fe924eda 100644
--- a/src/TNL/Solvers/Linear/SOR_impl.cpp
+++ b/src/TNL/Solvers/Linear/SOR_impl.cpp
@@ -8,6 +8,8 @@
/* See Copyright Notice in tnl/Copyright */
+#ifdef TEMPLATE_EXPLICIT_INSTANTIATION
+
#include <TNL/Solvers/Linear/SOR.h>
#include <TNL/Matrices/CSR.h>
#include <TNL/Matrices/Ellpack.h>
@@ -54,4 +56,4 @@ template class SOR< Matrices::Multidiagonal< double, Devices::Cuda, long int > >
} // namespace Solvers
} // namespace TNL
-
+#endif // #ifdef TEMPLATE_EXPLICIT_INSTANTIATION
diff --git a/src/TNL/Solvers/Linear/SOR_impl.h b/src/TNL/Solvers/Linear/SOR_impl.h
index d7d883f161342a4311caf8a7a99411fc36683d17..509221475dfc2b69c33198cb361fe51f7542a566 100644
--- a/src/TNL/Solvers/Linear/SOR_impl.h
+++ b/src/TNL/Solvers/Linear/SOR_impl.h
@@ -10,6 +10,8 @@
#pragma once
+#include <TNL/Solvers/Linear/SOR.h>
+
namespace TNL {
namespace Solvers {
namespace Linear {
@@ -106,6 +108,7 @@ bool SOR< Matrix, Preconditioner > :: solve( const Vector& b, Vector& x )
row,
x,
this->getOmega() );
+ // FIXME: the LinearResidueGetter works only on the host
this->setResidue( ResidueGetter::getResidue( *matrix, x, b, bNorm ) );
this->refreshSolverMonitor();
}
@@ -114,15 +117,13 @@ bool SOR< Matrix, Preconditioner > :: solve( const Vector& b, Vector& x )
return this->checkConvergence();
};
-template< typename Matrix, typename Preconditioner >
-SOR< Matrix, Preconditioner > :: ~SOR()
-{
-}
-
} // namespace Linear
} // namespace Solvers
} // namespace TNL
+
+#ifdef TEMPLATE_EXPLICIT_INSTANTIATION
+
#include <TNL/Matrices/CSR.h>
#include <TNL/Matrices/Ellpack.h>
#include <TNL/Matrices/Multidiagonal.h>
@@ -175,3 +176,5 @@ extern template class SOR< tnlMutliDiagonalMatrix< double, Devices::Cuda, long i
} // namespace Linear
} // namespace Solvers
} // namespace TNL
+
+#endif // #ifdef TEMPLATE_EXPLICIT_INSTANTIATION
diff --git a/src/TNL/Solvers/Linear/TFQMR.h b/src/TNL/Solvers/Linear/TFQMR.h
index ec2ae65eeb927323f73c33e60057be73e6d8785b..abf37d737298e27395c1dbb4015f398582cf17a2 100644
--- a/src/TNL/Solvers/Linear/TFQMR.h
+++ b/src/TNL/Solvers/Linear/TFQMR.h
@@ -14,7 +14,6 @@
#include <TNL/Object.h>
#include <TNL/SharedPointer.h>
#include <TNL/Containers/Vector.h>
-#include <TNL/Containers/SharedVector.h>
#include <TNL/Solvers/Linear/Preconditioners/Dummy.h>
#include <TNL/Solvers/IterativeSolver.h>
#include <TNL/Solvers/Linear/LinearResidueGetter.h>
@@ -62,7 +61,7 @@ class TFQMR : public Object,
protected:
- bool setSize( IndexType size );
+ void setSize( IndexType size );
Containers::Vector< RealType, DeviceType, IndexType > d, r, w, u, v, r_ast, Au, M_tmp;
diff --git a/src/TNL/Solvers/Linear/TFQMR_impl.h b/src/TNL/Solvers/Linear/TFQMR_impl.h
index 7fb2b7aa98b1d02a519e248a6a6ec6f35eef9a0c..7f6546a75caa6fe1dc4c09de47a694b46a4f02df 100644
--- a/src/TNL/Solvers/Linear/TFQMR_impl.h
+++ b/src/TNL/Solvers/Linear/TFQMR_impl.h
@@ -76,8 +76,7 @@ template< typename Matrix,
template< typename Vector, typename ResidueGetter >
bool TFQMR< Matrix, Preconditioner >::solve( const Vector& b, Vector& x )
{
- if( ! this->setSize( matrix -> getRows() ) )
- return false;
+ this->setSize( matrix -> getRows() );
RealType tau, theta, eta, rho, alpha, b_norm, w_norm;
@@ -174,32 +173,25 @@ bool TFQMR< Matrix, Preconditioner >::solve( const Vector& b, Vector& x )
this->refreshSolverMonitor( true );
return this->checkConvergence();
-};
+}
template< typename Matrix,
typename Preconditioner >
-bool TFQMR< Matrix, Preconditioner > :: setSize( IndexType size )
+void TFQMR< Matrix, Preconditioner > :: setSize( IndexType size )
{
if( this->size == size )
- return true;
+ return;
this->size = size;
- if( ! d. setSize( size ) ||
- ! r. setSize( size ) ||
- ! w. setSize( size ) ||
- ! u. setSize( size ) ||
- ! v. setSize( size ) ||
- ! r_ast. setSize( size ) ||
- ! Au. setSize( size ) ||
- ! M_tmp. setSize( size ) )
- {
- std::cerr << "I am not able to allocate all supporting vectors for the TFQMR solver." << std::endl;
- return false;
- }
- return true;
-
-};
+ d.setSize( size );
+ r.setSize( size );
+ w.setSize( size );
+ u.setSize( size );
+ v.setSize( size );
+ r_ast.setSize( size );
+ Au.setSize( size );
+ M_tmp.setSize( size );
+}
} // namespace Linear
} // namespace Solvers
} // namespace TNL
-
diff --git a/src/TNL/Solvers/Linear/UmfpackWrapper_impl.h b/src/TNL/Solvers/Linear/UmfpackWrapper_impl.h
index a07143a573900b4369cae396e43ea4b1a0797cc4..ebc401d633e9e2792bcff1b16ebcb9bc88646a09 100644
--- a/src/TNL/Solvers/Linear/UmfpackWrapper_impl.h
+++ b/src/TNL/Solvers/Linear/UmfpackWrapper_impl.h
@@ -63,8 +63,9 @@ bool UmfpackWrapper< Matrices::CSR< double, Devices::Host, int >, Preconditioner
solve( const Vector& b,
Vector& x )
{
- TNL_ASSERT( matrix->getRows() == matrix->getColumns(), );
- TNL_ASSERT( matrix->getColumns() == x.getSize() && matrix->getColumns() == b.getSize(), );
+ TNL_ASSERT_EQ( matrix->getRows(), matrix->getColumns(), "matrix must be square" );
+ TNL_ASSERT_EQ( matrix->getColumns(), x.getSize(), "wrong size of the solution vector" );
+ TNL_ASSERT_EQ( matrix->getColumns(), b.getSize(), "wrong size of the right hand side" );
const IndexType size = matrix -> getRows();
@@ -87,9 +88,9 @@ solve( const Vector& b,
// symbolic reordering of the sparse matrix
status = umfpack_di_symbolic( size, size,
- matrix->getRowPointers(),
- matrix->getColumnIndexes(),
- matrix->getValues(),
+ matrix->getRowPointers().getData(),
+ matrix->getColumnIndexes().getData(),
+ matrix->getValues().getData(),
&Symbolic, Control, Info );
if( status != UMFPACK_OK ) {
std::cerr << "error: symbolic reordering failed" << std::endl;
@@ -97,9 +98,9 @@ solve( const Vector& b,
}
// numeric factorization
- status = umfpack_di_numeric( matrix->getRowPointers(),
- matrix->getColumnIndexes(),
- matrix->getValues(),
+ status = umfpack_di_numeric( matrix->getRowPointers().getData(),
+ matrix->getColumnIndexes().getData(),
+ matrix->getValues().getData(),
Symbolic, &Numeric, Control, Info );
if( status != UMFPACK_OK ) {
std::cerr << "error: numeric factorization failed" << std::endl;
@@ -108,9 +109,9 @@ solve( const Vector& b,
// solve with specified right-hand-side
status = umfpack_di_solve( system_type,
- matrix->getRowPointers(),
- matrix->getColumnIndexes(),
- matrix->getValues(),
+ matrix->getRowPointers().getData(),
+ matrix->getColumnIndexes().getData(),
+ matrix->getValues().getData(),
x.getData(),
b.getData(),
Numeric, Control, Info );
diff --git a/src/TNL/Solvers/ODE/CMakeLists.txt b/src/TNL/Solvers/ODE/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Solvers/ODE/Euler_impl.h b/src/TNL/Solvers/ODE/Euler_impl.h
index ce90e16ec77286be4ca6c8187d14ebc51e6031e0..42f563b11438bce4cc7c625a011e134d59d27c1b 100644
--- a/src/TNL/Solvers/ODE/Euler_impl.h
+++ b/src/TNL/Solvers/ODE/Euler_impl.h
@@ -10,6 +10,8 @@
#pragma once
+#include <TNL/Devices/MIC.h>
+
namespace TNL {
namespace Solvers {
namespace ODE {
@@ -74,11 +76,7 @@ bool Euler< Problem > :: solve( DofVectorPointer& u )
* First setup the supporting meshes k1...k5 and k_tmp.
*/
//timer.start();
- if( ! k1->setLike( *u ) )
- {
- std::cerr << "I do not have enough memory to allocate a supporting grid for the Euler explicit solver." << std::endl;
- return false;
- }
+ k1->setLike( *u );
k1->setValue( 0.0 );
@@ -203,12 +201,38 @@ void Euler< Problem > :: computeNewTimeLevel( DofVectorPointer& u,
this->cudaBlockResidue.getData() );
localResidue += this->cudaBlockResidue.sum();
cudaThreadSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
#endif
}
- localResidue /= tau * ( RealType ) size;
+
+ //MIC
+ if( std::is_same< DeviceType, Devices::MIC >::value )
+ {
+
+#ifdef HAVE_MIC
+ Devices::MICHider<RealType> mu;
+ mu.pointer=_u;
+ Devices::MICHider<RealType> mk1;
+ mk1.pointer=_k1;
+ #pragma offload target(mic) in(mu,mk1,size) inout(localResidue)
+ {
+ #pragma omp parallel for reduction(+:localResidue) firstprivate( mu, mk1 )
+ for( IndexType i = 0; i < size; i ++ )
+ {
+ const RealType add = tau * mk1.pointer[ i ];
+ mu.pointer[ i ] += add;
+ localResidue += std::fabs( add );
+ }
+ }
+#endif
+ }
+
+
+
+ localResidue /= tau * ( RealType ) size;
MPIAllreduce( localResidue, currentResidue, 1, MPI_SUM, this->solver_comm );
+
}
#ifdef HAVE_CUDA
@@ -220,7 +244,7 @@ __global__ void updateUEuler( const Index size,
RealType* cudaBlockResidue )
{
extern __shared__ RealType du[];
- const Index blockOffset = blockIdx. x * blockDim. x;
+ const Index blockOffset = blockIdx. x * blockDim.x;
const Index i = blockOffset + threadIdx. x;
if( i < size )
u[ i ] += du[ threadIdx.x ] = tau * k1[ i ];
diff --git a/src/TNL/Solvers/ODE/Merson_impl.h b/src/TNL/Solvers/ODE/Merson_impl.h
index cb8a7808bfa71e3ec42d5c57d43827f76e376abd..d206087f1cc3d5ba78a888ef916b06f40e76235a 100644
--- a/src/TNL/Solvers/ODE/Merson_impl.h
+++ b/src/TNL/Solvers/ODE/Merson_impl.h
@@ -133,19 +133,20 @@ bool Merson< Problem > :: solve( DofVectorPointer& u )
std::cerr << "No problem was set for the Merson ODE solver." << std::endl;
return false;
}
- /****
- * First setup the supporting meshes k1...k5 and kAux.
- */
- if( ! k1->setLike( *u ) ||
- ! k2->setLike( *u ) ||
- ! k3->setLike( *u ) ||
- ! k4->setLike( *u ) ||
- ! k5->setLike( *u ) ||
- ! kAux->setLike( *u ) )
+ if( this->getTau() == 0.0 )
{
- std::cerr << "I do not have enough memory to allocate supporting grids for the Merson explicit solver." << std::endl;
+ std::cerr << "The time step for the Merson ODE solver is zero." << std::endl;
return false;
}
+ /****
+ * First setup the supporting meshes k1...k5 and kAux.
+ */
+ k1->setLike( *u );
+ k2->setLike( *u );
+ k3->setLike( *u );
+ k4->setLike( *u );
+ k5->setLike( *u );
+ kAux->setLike( *u );
k1->setValue( 0.0 );
k2->setValue( 0.0 );
k3->setValue( 0.0 );
@@ -254,21 +255,12 @@ void Merson< Problem >::computeKFunctions( DofVectorPointer& u,
/****
* Compute data transfers statistics
*/
-#ifdef HAVE_NOT_CXX11
- k1->template touch< IndexType >( 4 );
- k2->template touch< IndexType >( 1 );
- k3->template touch< IndexType >( 2 );
- k4->template touch< IndexType >( 1 );
- kAux->template touch< IndexType >( 4 );
- u->template touch< IndexType >( 4 );
-#else
k1->touch( 4 );
k2->touch( 1 );
k3->touch( 2 );
k4->touch( 1 );
kAux->touch( 4 );
u->touch( 4 );
-#endif
RealType tau_3 = tau / 3.0;
@@ -372,17 +364,10 @@ typename Problem :: RealType Merson< Problem > :: computeError( const RealType t
/****
* Compute data transfers statistics
*/
-#ifdef HAVE_NOT_CXX11
- k1->template touch< IndexType >();
- k3->template touch< IndexType >();
- k4->template touch< IndexType >();
- k5->template touch< IndexType >();
-#else
k1->touch();
k3->touch();
k4->touch();
k5->touch();
-#endif
RealType eps( 0.0 ), maxEps( 0.0 );
if( std::is_same< DeviceType, Devices::Host >::value )
@@ -453,17 +438,10 @@ void Merson< Problem >::computeNewTimeLevel( DofVectorPointer& u,
/****
* Compute data transfers statistics
*/
-#ifdef HAVE_NOT_CXX11
- u->template touch< IndexType >();
- k1->template touch< IndexType >();
- k4->template touch< IndexType >();
- k5->template touch< IndexType >();
-#else
u->touch();
k1->touch();
k4->touch();
k5->touch();
-#endif
if( std::is_same< DeviceType, Devices::Host >::value )
{
diff --git a/src/TNL/Solvers/PDE/CMakeLists.txt b/src/TNL/Solvers/PDE/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/Solvers/PDE/ExplicitTimeStepper_impl.h b/src/TNL/Solvers/PDE/ExplicitTimeStepper_impl.h
index dff5f2e6c53b79ee348391d702fb4d53e32ef102..a2450e38d2acdd331da85a59ef3ca639c022660d 100644
--- a/src/TNL/Solvers/PDE/ExplicitTimeStepper_impl.h
+++ b/src/TNL/Solvers/PDE/ExplicitTimeStepper_impl.h
@@ -122,7 +122,7 @@ solve( const RealType& time,
DofVectorPointer& dofVector,
MeshDependentDataPointer& meshDependentData )
{
- TNL_ASSERT( this->odeSolver, );
+ TNL_ASSERT_TRUE( this->odeSolver, "ODE solver was not set" );
mainTimer.start();
this->odeSolver->setTau( this->timeStep );
this->odeSolver->setProblem( * this );
diff --git a/src/TNL/Solvers/PDE/ExplicitUpdater.h b/src/TNL/Solvers/PDE/ExplicitUpdater.h
index cc4056b70eeb082154dedbf2e2c6318a8949a77a..264ecef7c0a83a3f6ead920de11a0225fb61cd36 100644
--- a/src/TNL/Solvers/PDE/ExplicitUpdater.h
+++ b/src/TNL/Solvers/PDE/ExplicitUpdater.h
@@ -122,12 +122,12 @@ class ExplicitUpdater
typename MeshFunction::IndexType > >::value != true,
"Error: I am getting Vector instead of MeshFunction or similar object. You might forget to bind DofVector into MeshFunction in you method getExplicitUpdate." );
- TNL_ASSERT( this->userDataPointer->differentialOperator,
- std::cerr << "The differential operator is not correctly set-up. Use method setDifferentialOperator() to do it." << std::endl );
- TNL_ASSERT( this->userDataPointer->boundaryConditions,
- std::cerr << "The boundary conditions are not correctly set-up. Use method setBoundaryCondtions() to do it." << std::endl );
- TNL_ASSERT( this->userDataPointer->rightHandSide,
- std::cerr << "The right-hand side is not correctly set-up. Use method setRightHandSide() to do it." << std::endl );
+ TNL_ASSERT_TRUE( this->userDataPointer->differentialOperator,
+ "The differential operator is not correctly set-up. Use method setDifferentialOperator() to do it." );
+ TNL_ASSERT_TRUE( this->userDataPointer->boundaryConditions,
+ "The boundary conditions are not correctly set-up. Use method setBoundaryCondtions() to do it." );
+ TNL_ASSERT_TRUE( this->userDataPointer->rightHandSide,
+ "The right-hand side is not correctly set-up. Use method setRightHandSide() to do it." );
this->userDataPointer->time = time;
@@ -175,12 +175,23 @@ class ExplicitUpdater
TraverserUserData& userData,
const EntityType& entity )
{
+ /* std::cerr<<"===========================================================" << std::endl;
+ std::cerr<<"fu:" << userData.fu << std::endl;
+ std::cerr<< "diffOp:" << userData.differentialOperator << std::endl;
+ std::cerr<<"===========================================================" << std::endl;
+
+ std::cerr<<std::flush;*/
+
+ // int blabla;
+ // std::cin >> blabla;
+
( *userData.fu )( entity ) =
- ( *userData.differentialOperator )( *userData.u, entity, userData.time );
-
+ ( *userData.differentialOperator )( *userData.u, entity, userData.time );
+
typedef Functions::FunctionAdapter< MeshType, RightHandSide > FunctionAdapter;
( *userData.fu )( entity ) +=
FunctionAdapter::getValue( *userData.rightHandSide, entity, userData.time );
+
}
};
diff --git a/src/TNL/Solvers/PDE/LinearSystemAssembler.h b/src/TNL/Solvers/PDE/LinearSystemAssembler.h
index 6b96f0b66cda28c970a38153e5ec637892746e88..d97dd15ef530ec33833d016d4ed321a83d711df6 100644
--- a/src/TNL/Solvers/PDE/LinearSystemAssembler.h
+++ b/src/TNL/Solvers/PDE/LinearSystemAssembler.h
@@ -114,7 +114,7 @@ class LinearSystemAssembler
"Error: I am getting Vector instead of MeshFunction or similar object. You might forget to bind DofVector into MeshFunction in you method getExplicitUpdate." );
const IndexType maxRowLength = matrixPointer.template getData< Devices::Host >().getMaxRowLength();
- TNL_ASSERT( maxRowLength > 0, );
+ TNL_ASSERT_GT( maxRowLength, 0, "maximum row length must be positive" );
this->userDataPointer->time = time;
this->userDataPointer->tau = tau;
this->userDataPointer->u = &uPointer.template getData< DeviceType >();
diff --git a/src/TNL/Solvers/PDE/SemiImplicitTimeStepper_impl.h b/src/TNL/Solvers/PDE/SemiImplicitTimeStepper_impl.h
index 6e996c2828e8b63200cf031eb53bef5d1728e251..1c324f8f944d43ae73cea4036ae252768112bc32 100644
--- a/src/TNL/Solvers/PDE/SemiImplicitTimeStepper_impl.h
+++ b/src/TNL/Solvers/PDE/SemiImplicitTimeStepper_impl.h
@@ -11,7 +11,8 @@
#pragma once
#include <TNL/Math.h>
-#include "SemiImplicitTimeStepper.h"
+#include <TNL/Solvers/PDE/SemiImplicitTimeStepper.h>
+#include <TNL/Solvers/Linear/Preconditioners/Dummy.h>
namespace TNL {
namespace Solvers {
@@ -69,8 +70,7 @@ init( const MeshPointer& mesh )
std::cerr << "Please check the method 'setupLinearSystem' in your solver." << std::endl;
return false;
}
- if( ! this->rightHandSidePointer->setSize( this->matrix.getData().getRows() ) )
- return false;
+ this->rightHandSidePointer->setSize( this->matrix.getData().getRows() );
this->preIterateTimer.reset();
this->linearSystemAssemblerTimer.reset();
@@ -154,14 +154,15 @@ solve( const RealType& time,
DofVectorPointer& dofVector,
MeshDependentDataPointer& meshDependentData )
{
- TNL_ASSERT( this->problem != 0, );
+ TNL_ASSERT_TRUE( this->problem, "problem was not set" );
RealType t = time;
this->linearSystemSolver->setMatrix( this->matrix );
PreconditionerPointer preconditioner;
Linear::Preconditioners::SolverStarterSolverPreconditionerSetter< LinearSystemSolverType, PreconditionerType >
::run( *(this->linearSystemSolver), preconditioner );
- while( t < stopTime )
+ // ignore very small steps at the end, most likely caused by truncation errors
+ while( stopTime - t > this->timeStep * 1e-6 )
{
RealType currentTau = min( this->timeStep, stopTime - t );
@@ -207,9 +208,11 @@ solve( const RealType& time,
this->preconditionerUpdateTimer.stop();
this->linearSystemSolverTimer.start();
- if( ! this->linearSystemSolver->template solve< DofVectorType, Linear::LinearResidueGetter< MatrixType, DofVectorType > >( *this->rightHandSidePointer, *dofVector ) )
+ if( ! this->linearSystemSolver->solve( *this->rightHandSidePointer, *dofVector ) )
{
std::cerr << std::endl << "The linear system solver did not converge." << std::endl;
+ // save the linear system for debugging
+ this->problem->saveFailedLinearSystem( *this->matrix, *dofVector, *this->rightHandSidePointer );
return false;
}
this->linearSystemSolverTimer.stop();
diff --git a/src/TNL/Solvers/PDE/TimeDependentPDESolver_impl.h b/src/TNL/Solvers/PDE/TimeDependentPDESolver_impl.h
index 2a22201d1e54f2d75b809c5b2e914abb332bd390..6d20972d28c9f06cdf8da05e90567ac50f656e67 100644
--- a/src/TNL/Solvers/PDE/TimeDependentPDESolver_impl.h
+++ b/src/TNL/Solvers/PDE/TimeDependentPDESolver_impl.h
@@ -80,15 +80,8 @@ setup( const Config::ParameterContainer& parameters,
/****
* Set DOFs (degrees of freedom)
*/
- TNL_ASSERT( problem->getDofs( this->meshPointer ) != 0, );
- std::cout << "Allocating dofs ... ";
- if( ! this->dofsPointer->setSize( problem->getDofs( this->meshPointer ) ) )
- {
- std::cerr << std::endl;
- std::cerr << "I am not able to allocate DOFs (degrees of freedom)." << std::endl;
- return false;
- }
- std::cout << " [ OK ]" << std::endl;
+ TNL_ASSERT_GT( problem->getDofs( this->meshPointer ), 0, "number of DOFs must be positive" );
+ this->dofsPointer->setSize( problem->getDofs( this->meshPointer ) );
this->dofsPointer->setValue( 0.0 );
this->problem->bindDofs( this->meshPointer, this->dofsPointer );
@@ -101,11 +94,11 @@ setup( const Config::ParameterContainer& parameters,
/***
* Set-up the initial condition
*/
- std::cout << "Setting up the initial condition ... ";
+ std::cout << "Setting up the initial condition ... ";
typedef typename Problem :: DofVectorType DofVectorType;
if( ! this->problem->setInitialCondition( parameters, meshPointer, this->dofsPointer, this->meshDependentDataPointer ) )
return false;
- std::cout << " [ OK ]" << std::endl;
+ std::cout << " [ OK ]" << std::endl;
/****
* Initialize the time discretisation
@@ -141,8 +134,12 @@ writeProlog( Logger& logger,
logger.writeParameter< double >( "Adaptivity:", "merson-adaptivity", parameters, 1 );
if( solverName == "sor" )
logger.writeParameter< double >( "Omega:", "sor-omega", parameters, 1 );
- if( solverName == "gmres" )
- logger.writeParameter< int >( "Restarting:", "gmres-restarting", parameters, 1 );
+ if( solverName == "gmres" || solverName == "cwygmres" ) {
+ logger.writeParameter< int >( "Restarting min:", "gmres-restarting-min", parameters, 1 );
+ logger.writeParameter< int >( "Restarting max:", "gmres-restarting-max", parameters, 1 );
+ logger.writeParameter< int >( "Restarting step min:", "gmres-restarting-step-min", parameters, 1 );
+ logger.writeParameter< int >( "Restarting step max:", "gmres-restarting-step-max", parameters, 1 );
+ }
logger.writeParameter< double >( "Convergence residue:", "convergence-residue", parameters );
logger.writeParameter< double >( "Divergence residue:", "divergence-residue", parameters );
logger.writeParameter< int >( "Maximal number of iterations:", "max-iterations", parameters );
@@ -319,10 +316,8 @@ bool
TimeDependentPDESolver< Problem, TimeStepper >::
solve()
{
- TNL_ASSERT( timeStepper != 0,
- std::cerr << "No time stepper was set in PDESolver." );
- TNL_ASSERT( problem != 0,
- std::cerr << "No problem was set in PDESolver." );
+ TNL_ASSERT_TRUE( timeStepper, "No time stepper was set in PDESolver." );
+ TNL_ASSERT_TRUE( problem, "No problem was set in PDESolver." );
if( snapshotPeriod == 0 )
{
diff --git a/src/TNL/Solvers/PDE/TimeIndependentPDESolver_impl.h b/src/TNL/Solvers/PDE/TimeIndependentPDESolver_impl.h
index 0f27de6489ffc3ee7a4ec37d6ca1d69c766f913a..957f710300bbc373b184dbd5736028377b68a7ba 100644
--- a/src/TNL/Solvers/PDE/TimeIndependentPDESolver_impl.h
+++ b/src/TNL/Solvers/PDE/TimeIndependentPDESolver_impl.h
@@ -58,15 +58,8 @@ setup( const Config::ParameterContainer& parameters,
/****
* Set DOFs (degrees of freedom)
*/
- TNL_ASSERT( problem->getDofs( this->mesh ) != 0, );
- cout << "Allocating dofs ... ";
- if( ! this->dofs.setSize( problem->getDofs( this->mesh ) ) )
- {
- cerr << endl;
- cerr << "I am not able to allocate DOFs (degrees of freedom)." << endl;
- return false;
- }
- cout << " [ OK ]" << endl;
+ TNL_ASSERT_GT( problem->getDofs( this->mesh ), 0, "number of DOFs must be positive" );
+ this->dofs.setSize( problem->getDofs( this->mesh ) );
this->dofs.setValue( 0.0 );
this->problem->bindDofs( this->mesh, this->dofs );
@@ -85,7 +78,6 @@ setup( const Config::ParameterContainer& parameters,
return false;
cout << " [ OK ]" << endl;
-
return true;
}
@@ -149,8 +141,7 @@ bool
tnlTimeIndependentPDESolver< Problem >::
solve()
{
- TNL_ASSERT( problem != 0,
- cerr << "No problem was set in tnlPDESolver." );
+ TNL_ASSERT_TRUE( problem, "No problem was set in tnlPDESolver." );
this->computeTimer->reset();
this->computeTimer->start();
diff --git a/src/TNL/Solvers/SolverConfig_impl.h b/src/TNL/Solvers/SolverConfig_impl.h
index a191add046b7f1b4e2b317a26e4abd23ee96606d..e66397310414edbecc68d7a1090e2186b952f851 100644
--- a/src/TNL/Solvers/SolverConfig_impl.h
+++ b/src/TNL/Solvers/SolverConfig_impl.h
@@ -26,6 +26,10 @@ bool SolverConfig< ConfigTag, ProblemConfig >::configSetup( Config::ConfigDescri
typedef DummyProblem< double, Devices::Host, int > DummyProblemType;
config.addDelimiter( " === General parameters ==== " );
+ config.addEntry< bool >( "catch-exceptions",
+ "Catch C++ exceptions. Disabling it allows the program to drop into the debugger "
+ "and track the origin of the exception.",
+ true );
/****
* Setup real type
*/
@@ -51,6 +55,12 @@ bool SolverConfig< ConfigTag, ProblemConfig >::configSetup( Config::ConfigDescri
if( ConfigTagDevice< ConfigTag, Devices::Cuda >::enabled )
config.addEntryEnum( "cuda" );
#endif
+
+#ifdef HAVE_MIC
+ if( ConfigTagDevice< ConfigTag, Devices::MIC >::enabled )
+ config.addEntryEnum( "mic" );
+#endif
+
/****
* Setup index type.
@@ -107,6 +117,8 @@ bool SolverConfig< ConfigTag, ProblemConfig >::configSetup( Config::ConfigDescri
config.addEntryEnum( "cg" );
if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitBICGStabSolverTag >::enabled )
config.addEntryEnum( "bicgstab" );
+ if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitBICGStabLSolverTag >::enabled )
+ config.addEntryEnum( "bicgstabl" );
if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitCWYGMRESSolverTag >::enabled )
config.addEntryEnum( "cwygmres" );
if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitGMRESSolverTag >::enabled )
@@ -123,6 +135,10 @@ bool SolverConfig< ConfigTag, ProblemConfig >::configSetup( Config::ConfigDescri
config.addEntry< String >( "preconditioner", "The preconditioner for the discrete solver:", "none" );
config.addEntryEnum( "none" );
config.addEntryEnum( "diagonal" );
+// TODO: implement parallel ILU or device-dependent build config tags for preconditioners
+#ifndef HAVE_CUDA
+ config.addEntryEnum( "ilu0" );
+#endif
if( ConfigTagTimeDiscretisation< ConfigTag, ExplicitTimeDiscretisationTag >::enabled ||
ConfigTagTimeDiscretisation< ConfigTag, SemiImplicitTimeDiscretisationTag >::enabled )
{
@@ -147,10 +163,15 @@ bool SolverConfig< ConfigTag, ProblemConfig >::configSetup( Config::ConfigDescri
Linear::CG< MatrixType >::configSetup( config );
if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitBICGStabSolverTag >::enabled )
Linear::BICGStab< MatrixType >::configSetup( config );
+ if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitBICGStabLSolverTag >::enabled )
+ Linear::BICGStabL< MatrixType >::configSetup( config );
+
+ // GMRES and CWYGMRES have the same options
if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitCWYGMRESSolverTag >::enabled )
Linear::CWYGMRES< MatrixType >::configSetup( config );
- if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitGMRESSolverTag >::enabled )
+ else if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitGMRESSolverTag >::enabled )
Linear::GMRES< MatrixType >::configSetup( config );
+
if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitTFQMRSolverTag >::enabled )
Linear::TFQMR< MatrixType >::configSetup( config );
if( ConfigTagSemiImplicitSolver< ConfigTag, SemiImplicitSORSolverTag >::enabled )
@@ -167,4 +188,3 @@ bool SolverConfig< ConfigTag, ProblemConfig >::configSetup( Config::ConfigDescri
} // namespace Solvers
} // namespace TNL
-
diff --git a/src/TNL/Solvers/SolverInitiator_impl.h b/src/TNL/Solvers/SolverInitiator_impl.h
index c826563a3e2717c7119ad6e464b5521e228d04a2..1b5ddc1a3bd15ccd16cb4bcfa7bc6940f7342134 100644
--- a/src/TNL/Solvers/SolverInitiator_impl.h
+++ b/src/TNL/Solvers/SolverInitiator_impl.h
@@ -17,6 +17,7 @@
#include <TNL/Solvers/Linear/GMRES.h>
#include <TNL/Devices/Host.h>
#include <TNL/Devices/Cuda.h>
+#include <TNL/Devices/MIC.h>
namespace TNL {
namespace Solvers {
@@ -75,6 +76,8 @@ class SolverInitiatorRealResolver< ProblemSetter, Real, ConfigTag, true >
return SolverInitiatorDeviceResolver< ProblemSetter, Real, Devices::Host, ConfigTag >::run( parameters );
if( device == "cuda" )
return SolverInitiatorDeviceResolver< ProblemSetter, Real, Devices::Cuda, ConfigTag >::run( parameters );
+ if(device == "mic")
+ return SolverInitiatorDeviceResolver< ProblemSetter, Real, Devices::MIC, ConfigTag >::run( parameters );
std::cerr << "The device '" << device << "' is not defined. " << std::endl;
return false;
}
diff --git a/src/TNL/Solvers/SolverMonitor.h b/src/TNL/Solvers/SolverMonitor.h
index 46911415f469aa16f52ac683607e598e2736cd2e..73d78ea4cd458f4e6981d56bf5357d5df6eac3da 100644
--- a/src/TNL/Solvers/SolverMonitor.h
+++ b/src/TNL/Solvers/SolverMonitor.h
@@ -20,16 +20,14 @@ namespace Solvers {
class SolverMonitor
{
- public:
-
+public:
SolverMonitor()
- : timeout_milliseconds(500),
- stopped(true),
- timer(nullptr)
- {};
+ : timeout_milliseconds( 500 ),
+ started( false ),
+ stopped( false ),
+ timer( nullptr )
+ {}
- ~SolverMonitor() {};
-
virtual void refresh( bool force = false ) = 0;
void setRefreshRate( const int& refreshRate )
@@ -44,7 +42,11 @@ class SolverMonitor
void runMainLoop()
{
- stopped = false;
+ // We need to use both 'started' and 'stopped' to avoid a deadlock
+ // when the loop thread runs this method delayed after the
+ // SolverMonitorThread's destructor has already called stopMainLoop()
+ // from the main thread.
+ started = true;
const int timeout_base = 100;
const std::chrono::milliseconds timeout( timeout_base );
@@ -62,6 +64,10 @@ class SolverMonitor
std::this_thread::sleep_for( timeout );
}
}
+
+ // reset to initial state
+ started = false;
+ stopped = false;
}
void stopMainLoop()
@@ -69,8 +75,12 @@ class SolverMonitor
stopped = true;
}
- protected:
+ bool isStopped() const
+ {
+ return stopped;
+ }
+protected:
double getElapsedTime()
{
if( ! timer )
@@ -80,6 +90,7 @@ class SolverMonitor
std::atomic_int timeout_milliseconds;
+ std::atomic_bool started;
std::atomic_bool stopped;
Timer* timer;
@@ -110,5 +121,4 @@ class SolverMonitorThread
};
} // namespace Solvers
-} // namespace TNL
-
+} // namespace TNL
\ No newline at end of file
diff --git a/src/TNL/Solvers/SolverStarter_impl.h b/src/TNL/Solvers/SolverStarter_impl.h
index 516dab97d42bf11b0aa4185bc1acf01db2ab1542..a8fae47d95980bf6f1414c6adeb38c1413a82fa0 100644
--- a/src/TNL/Solvers/SolverStarter_impl.h
+++ b/src/TNL/Solvers/SolverStarter_impl.h
@@ -14,17 +14,21 @@
#include <TNL/Logger.h>
#include <TNL/String.h>
#include <TNL/Devices/Cuda.h>
+#include <TNL/Solvers/SolverStarter.h>
+#include <TNL/Solvers/BuildConfigTags.h>
#include <TNL/Solvers/ODE/Merson.h>
#include <TNL/Solvers/ODE/Euler.h>
#include <TNL/Solvers/Linear/SOR.h>
#include <TNL/Solvers/Linear/CG.h>
#include <TNL/Solvers/Linear/BICGStab.h>
+#include <TNL/Solvers/Linear/BICGStabL.h>
#include <TNL/Solvers/Linear/GMRES.h>
#include <TNL/Solvers/Linear/CWYGMRES.h>
#include <TNL/Solvers/Linear/TFQMR.h>
#include <TNL/Solvers/Linear/UmfpackWrapper.h>
#include <TNL/Solvers/Linear/Preconditioners/Dummy.h>
#include <TNL/Solvers/Linear/Preconditioners/Diagonal.h>
+#include <TNL/Solvers/Linear/Preconditioners/ILU0.h>
#include <TNL/Solvers/PDE/ExplicitTimeStepper.h>
#include <TNL/Solvers/PDE/SemiImplicitTimeStepper.h>
#include <TNL/Solvers/PDE/TimeDependentPDESolver.h>
@@ -179,23 +183,25 @@ class SolverStarterTimeDiscretisationSetter< Problem, SemiImplicitTimeDiscretisa
if( discreteSolver != "sor" &&
discreteSolver != "cg" &&
discreteSolver != "bicgstab" &&
+ discreteSolver != "bicgstabl" &&
discreteSolver != "gmres" &&
discreteSolver != "cwygmres" &&
discreteSolver != "tfqmr" )
{
- std::cerr << "Unknown semi-implicit discrete solver " << discreteSolver << ". It can be only: sor, cg, bicgstab, gmres, cwygmres or tfqmr." << std::endl;
+ std::cerr << "Unknown semi-implicit discrete solver " << discreteSolver << ". It can be only: sor, cg, bicgstab, bicgstabl, gmres, cwygmres or tfqmr." << std::endl;
return false;
}
#else
if( discreteSolver != "sor" &&
discreteSolver != "cg" &&
discreteSolver != "bicgstab" &&
+ discreteSolver != "bicgstabl" &&
discreteSolver != "gmres" &&
discreteSolver != "cwygmres" &&
discreteSolver != "tfqmr" &&
discreteSolver != "umfpack" )
{
- std::cerr << "Unknown semi-implicit discrete solver " << discreteSolver << ". It can be only: sor, cg, bicgstab, gmres, cwygmres, tfqmr or umfpack." << std::endl;
+ std::cerr << "Unknown semi-implicit discrete solver " << discreteSolver << ". It can be only: sor, cg, bicgstab, bicgstabl, gmres, cwygmres, tfqmr or umfpack." << std::endl;
return false;
}
#endif
@@ -206,6 +212,8 @@ class SolverStarterTimeDiscretisationSetter< Problem, SemiImplicitTimeDiscretisa
return SolverStarterPreconditionerSetter< Problem, SemiImplicitCGSolverTag, ConfigTag >::run( problem, parameters );
if( discreteSolver == "bicgstab" )
return SolverStarterPreconditionerSetter< Problem, SemiImplicitBICGStabSolverTag, ConfigTag >::run( problem, parameters );
+ if( discreteSolver == "bicgstabl" )
+ return SolverStarterPreconditionerSetter< Problem, SemiImplicitBICGStabLSolverTag, ConfigTag >::run( problem, parameters );
if( discreteSolver == "gmres" )
return SolverStarterPreconditionerSetter< Problem, SemiImplicitGMRESSolverTag, ConfigTag >::run( problem, parameters );
if( discreteSolver == "cwygmres" )
@@ -228,7 +236,7 @@ class SolverStarterTimeDiscretisationSetter< Problem, ImplicitTimeDiscretisation
static bool run( Problem& problem,
const Config::ParameterContainer& parameters )
{
- const String& discreteSolver = parameters. getParameter< String>( "discrete-solver" );
+// const String& discreteSolver = parameters. getParameter< String>( "discrete-solver" );
return false;
}
};
@@ -286,8 +294,10 @@ class SolverStarterPreconditionerSetter
return SolverStarterSemiImplicitSolverSetter< Problem, SemiImplicitSolverTag, Linear::Preconditioners::Dummy, ConfigTag >::run( problem, parameters );
if( preconditioner == "diagonal" )
return SolverStarterSemiImplicitSolverSetter< Problem, SemiImplicitSolverTag, Linear::Preconditioners::Diagonal, ConfigTag >::run( problem, parameters );
+ if( preconditioner == "ilu0" )
+ return SolverStarterSemiImplicitSolverSetter< Problem, SemiImplicitSolverTag, Linear::Preconditioners::ILU0, ConfigTag >::run( problem, parameters );
- std::cerr << "Unknown preconditioner " << preconditioner << ". It can be only: none, diagonal." << std::endl;
+ std::cerr << "Unknown preconditioner " << preconditioner << ". It can be only: none, diagonal, ilu0." << std::endl;
return false;
}
};
@@ -356,42 +366,58 @@ bool SolverStarter< ConfigTag > :: runPDESolver( Problem& problem,
timeStepper.setSolver( discreteSolver );
/****
- * Set-up the PDE solver
+ * Open the log file
*/
- PDE::TimeDependentPDESolver< Problem, TimeStepper > solver;
- solver.setProblem( problem );
- solver.setTimeStepper( timeStepper );
- if( ! solver.setup( parameters ) )
+ const String logFileName = parameters.getParameter< String >( "log-file" );
+ std::ofstream logFile( logFileName.getString() );
+ if( ! logFile ) {
+ std::cerr << "Unable to open the log file " << logFileName << "." << std::endl;
return false;
+ }
/****
- * Write a prolog
+ * Set-up the PDE solver
*/
- int verbose = parameters.getParameter< int >( "verbose" );
- parameters. getParameter< int >( "log-width", logWidth );
- if( verbose )
- {
- Logger logger( logWidth,std::cout );
- solver.writeProlog( logger, parameters );
- }
- String logFileName;
- bool haveLogFile = parameters.getParameter< String >( "log-file", logFileName );
- if( haveLogFile )
- {
- std::fstream logFile;
- logFile.open( logFileName.getString(), std::ios::out );
- if( ! logFile )
- {
- std::cerr << "Unable to open the log file " << logFileName << "." << std::endl;
+ PDE::TimeDependentPDESolver< Problem, TimeStepper > solver;
+ // catching exceptions ala gtest:
+ // https://github.com/google/googletest/blob/59c795ce08be0c8b225bc894f8da6c7954ea5c14/googletest/src/gtest.cc#L2409-L2431
+ const int catch_exceptions = parameters.getParameter< bool >( "catch-exceptions" );
+ if( catch_exceptions ) {
+ try {
+ solver.setProblem( problem );
+ solver.setTimeStepper( timeStepper );
+ if( ! solver.setup( parameters ) )
+ return false;
+ }
+ catch ( const std::exception& e ) {
+ std::cerr << "Setting up the solver failed due to a C++ exception with description: " << e.what() << std::endl;
+ logFile << "Setting up The solver failed due to a C++ exception with description: " << e.what() << std::endl;
return false;
}
- else
- {
- Logger logger( logWidth, logFile );
- solver.writeProlog( logger, parameters );
- logFile.close();
+ catch (...) {
+ std::cerr << "Setting up the solver failed due to an unknown C++ exception." << std::endl;
+ logFile << "Setting up The solver failed due to an unknown C++ exception." << std::endl;
+ throw;
}
}
+ else {
+ solver.setProblem( problem );
+ solver.setTimeStepper( timeStepper );
+ if( ! solver.setup( parameters ) )
+ return false;
+ }
+
+ /****
+ * Write a prolog
+ */
+ const int verbose = parameters.getParameter< int >( "verbose" );
+ parameters.getParameter< int >( "log-width", logWidth );
+ if( verbose ) {
+ Logger logger( logWidth, std::cout );
+ solver.writeProlog( logger, parameters );
+ }
+ Logger logger( logWidth, logFile );
+ solver.writeProlog( logger, parameters );
/****
* Set-up solver monitor and launch the main loop.
@@ -418,25 +444,34 @@ bool SolverStarter< ConfigTag > :: runPDESolver( Problem& problem,
/****
* Start the solver
*/
- bool returnCode = solver.solve();
- solverMonitorPointer->stopMainLoop();
- if( ! returnCode )
- {
- if( verbose )
- std::cerr << std::endl << "The solver did not converge. " << std::endl;
- std::fstream logFile;
- logFile.open( logFileName.getString(), std::ios::out | std::ios::app );
- if( ! logFile )
- {
- std::cerr << "Unable to open the log file " << logFileName << "." << std::endl;
+ bool returnCode = true;
+ // catching exceptions ala gtest:
+ // https://github.com/google/googletest/blob/59c795ce08be0c8b225bc894f8da6c7954ea5c14/googletest/src/gtest.cc#L2409-L2431
+ if( catch_exceptions ) {
+ try {
+ returnCode = solver.solve();
+ }
+ catch ( const std::exception& e ) {
+ std::cerr << "The solver failed due to a C++ exception with description: " << e.what() << std::endl;
+ logFile << "The solver failed due to a C++ exception with description: " << e.what() << std::endl;
return false;
}
- else
- {
- logFile << "The solver did not converge. " << std::endl;
- logFile.close();
+ catch (...) {
+ std::cerr << "The solver failed due to an unknown C++ exception." << std::endl;
+ logFile << "The solver failed due to an unknown C++ exception." << std::endl;
+ throw;
}
}
+ else {
+ returnCode = solver.solve();
+ }
+
+ solverMonitorPointer->stopMainLoop();
+ if( ! returnCode ) {
+ if( verbose )
+ std::cerr << std::endl << "The solver did not converge. " << std::endl;
+ logFile << "The solver did not converge. " << std::endl;
+ }
/****
* Stop timers
@@ -448,22 +483,10 @@ bool SolverStarter< ConfigTag > :: runPDESolver( Problem& problem,
* Write an epilog
*/
if( verbose )
- writeEpilog(std::cout, solver );
- if( haveLogFile )
- {
- std::fstream logFile;
- logFile.open( logFileName.getString(), std::ios::out | std::ios::app );
- if( ! logFile )
- {
- std::cerr << "Unable to open the log file " << logFileName << "." << std::endl;
- return false;
- }
- else
- {
- writeEpilog( logFile, solver );
- logFile.close();
- }
- }
+ writeEpilog( std::cout, solver );
+ writeEpilog( logFile, solver );
+ logFile.close();
+
return returnCode;
}
diff --git a/src/TNL/StaticFor.h b/src/TNL/StaticFor.h
index 322c509852749fca2b063e54707ae776c7096b8a..bc63c43848b07251bf92b5db12b2a35f351f9f78 100644
--- a/src/TNL/StaticFor.h
+++ b/src/TNL/StaticFor.h
@@ -10,6 +10,8 @@
#pragma once
+#undef __INTEL_COMPILER
+
namespace TNL {
template< typename IndexType, IndexType val >
@@ -123,7 +125,7 @@ class StaticFor
__cuda_callable__
static void exec()
{
-#ifndef HAVE_ICPC
+#ifndef __INTEL_COMPILER
StaticForExecutor< IndexType,
StaticForIndexTag< IndexType, begin >,
StaticForIndexTag< IndexType, end - begin >,
@@ -137,7 +139,7 @@ class StaticFor
__cuda_callable__
static void exec( T &p )
{
-#ifndef HAVE_ICPC
+#ifndef __INTEL_COMPILER
StaticForExecutor< IndexType,
StaticForIndexTag< IndexType, begin >,
StaticForIndexTag< IndexType, end - begin >,
@@ -152,7 +154,7 @@ class StaticFor
__cuda_callable__
static void exec( T0& p0, T1& p1 )
{
-#ifndef HAVE_ICPC
+#ifndef __INTEL_COMPILER
StaticForExecutor< IndexType,
StaticForIndexTag< IndexType, begin >,
StaticForIndexTag< IndexType, end - begin >,
@@ -168,7 +170,7 @@ class StaticFor
__cuda_callable__
static void exec( T0& p0, T1& p1, T2& p2 )
{
-#ifndef HAVE_ICPC
+#ifndef __INTEL_COMPILER
StaticForExecutor< IndexType,
StaticForIndexTag< IndexType, begin >,
StaticForIndexTag< IndexType, end - begin >,
@@ -185,7 +187,7 @@ class StaticFor
__cuda_callable__
static void exec( T0& p0, T1& p1, T2& p2, T3& p3 )
{
-#ifndef HAVE_ICPC
+#ifndef __INTEL_COMPILER
StaticForExecutor< IndexType,
StaticForIndexTag< IndexType, begin >,
StaticForIndexTag< IndexType, end - begin >,
diff --git a/src/TNL/String.cpp b/src/TNL/String.cpp
index a3740961d26523d338679dc574b1163895960b80..af8b46da8af1a142aca151027f8149cc266c426a 100644
--- a/src/TNL/String.cpp
+++ b/src/TNL/String.cpp
@@ -10,7 +10,6 @@
#include <cstring>
#include <string.h>
-#include <assert.h>
#include <TNL/String.h>
#include <TNL/Assert.h>
#include <TNL/Containers/List.h>
@@ -24,123 +23,123 @@ namespace TNL {
const unsigned int STRING_PAGE = 256;
-String :: String()
+String::String()
+ : string( nullptr ), length( 0 )
{
- string = new char[ STRING_PAGE ];
- string[ 0 ] = 0;
- length = STRING_PAGE;
+ setString( nullptr );
}
-String :: String( const char* c, int prefix_cut_off, int sufix_cut_off )
- : string( 0 ), length( 0 )
+String::String( const char* c, int prefix_cut_off, int sufix_cut_off )
+ : string( nullptr ), length( 0 )
{
setString( c, prefix_cut_off, sufix_cut_off );
}
-String :: String( const String& str )
-: string( 0 ), length( 0 )
+String::String( const String& str )
+ : string( nullptr ), length( 0 )
{
- setString( str. getString() );
+ setString( str.getString() );
}
-String :: String( unsigned number )
-: string( 0 ), length( 0 )
+String String::getType()
{
- this->setString( convertToString( number ).getString() );
-}
-
-String :: String( int number )
-: string( 0 ), length( 0 )
-{
- this->setString( convertToString( number ).getString() );
-}
-
-String :: String( unsigned long int number )
-: string( 0 ), length( 0 )
-{
- this->setString( convertToString( number ).getString() );
+ return String( "String" );
}
-String :: String( long int number )
-: string( 0 ), length( 0 )
+String::~String()
{
- this->setString( convertToString( number ).getString() );
+ if( string ) delete[] string;
}
-String :: String( float number )
-: string( 0 ), length( 0 )
+int String::getLength() const
{
- this->setString( convertToString( number ).getString() );
+ return getSize();
}
-String :: String( double number )
-: string( 0 ), length( 0 )
+int String::getSize() const
{
- this->setString( convertToString( number ).getString() );
+ return strlen( string );
}
-String String :: getType()
+int String::getAllocatedSize() const
{
- return String( "String" );
+ return length;
}
-String :: ~String()
+void String::setSize( int size )
{
- if( string ) delete[] string;
+ TNL_ASSERT_GE( size, 0, "string size must be non-negative" );
+ const int _length = STRING_PAGE * ( size / STRING_PAGE + 1 );
+ TNL_ASSERT_GE( _length, 0, "_length size must be non-negative" );
+ if( length != _length ) {
+ if( string ) {
+ delete[] string;
+ string = nullptr;
+ }
+ string = new char[ _length ];
+ length = _length;
+ }
}
-void String :: setString( const char* c, int prefix_cut_off, int sufix_cut_off )
+void String::setString( const char* c, int prefix_cut_off, int sufix_cut_off )
{
- if( ! c )
- {
+ if( ! c ) {
if( ! string )
- {
- string = new char[ STRING_PAGE ];
- length = STRING_PAGE;
- }
+ setSize( 1 );
string[ 0 ] = 0;
return;
}
- int c_len = ( int ) strlen( c );
- int _length = max( 0, c_len - prefix_cut_off - sufix_cut_off );
+ const int c_len = ( int ) strlen( c );
+ const int _length = max( 0, c_len - prefix_cut_off - sufix_cut_off );
if( length < _length || length == 0 )
- {
- if( string ) delete[] string;
- length = STRING_PAGE * ( _length / STRING_PAGE + 1 );
- string = new char[ length ];
- }
+ setSize( _length );
TNL_ASSERT( string, );
- memcpy( string, c + min( c_len, prefix_cut_off ), sizeof( char ) * ( _length ) );
+ memcpy( string, c + min( c_len, prefix_cut_off ), _length * sizeof( char ) );
string[ _length ] = 0;
}
-const char& String :: operator[]( int i ) const
+const char* String::getString() const
+{
+ return string;
+}
+
+char* String::getString()
+{
+ return string;
+}
+
+
+const char& String::operator[]( int i ) const
{
TNL_ASSERT( i >= 0 && i < length,
- std::cerr << "Accessing char outside the string." );
+ std::cerr << "Accessing char outside the string." );
return string[ i ];
}
-char& String :: operator[]( int i )
+char& String::operator[]( int i )
{
TNL_ASSERT( i >= 0 && i < length,
- std::cerr << "Accessing char outside the string." );
+ std::cerr << "Accessing char outside the string." );
return string[ i ];
}
-String& String :: operator = ( const String& str )
+
+/****
+ * Operators for C strings
+ */
+String& String::operator=( const char* str )
{
- setString( str. getString() );
- return * this;
+ setString( str );
+ return *this;
}
-String& String :: operator += ( const char* str )
+String& String::operator+=( const char* str )
{
if( str )
{
- int len1 = strlen( string );
- int len2 = strlen( str );
+ const int len1 = strlen( string );
+ const int len2 = strlen( str );
if( len1 + len2 < length )
memcpy( string + len1, str, sizeof( char ) * ( len2 + 1 ) );
else
@@ -152,12 +151,70 @@ String& String :: operator += ( const char* str )
memcpy( string + len1, str, sizeof( char ) * ( len2 + 1 ) );
}
}
- return * this;
+ return *this;
+}
+
+String String::operator+( const char* str ) const
+{
+ return String( *this ) += str;
+}
+
+bool String::operator==( const char* str ) const
+{
+ TNL_ASSERT( string && str, );
+ return strcmp( string, str ) == 0;
+}
+
+bool String::operator!=( const char* str ) const
+{
+ return ! operator==( str );
+}
+
+
+/****
+ * Operators for Strings
+ */
+String& String::operator=( const String& str )
+{
+ setString( str.getString() );
+ return *this;
+}
+
+String& String::operator+=( const String& str )
+{
+ return operator+=( str.getString() );
+}
+
+String String::operator+( const String& str ) const
+{
+ return String( *this ) += str;
+}
+
+bool String::operator==( const String& str ) const
+{
+ TNL_ASSERT( string && str.string, );
+ return strcmp( string, str.string ) == 0;
}
-String& String :: operator += ( const char str )
+bool String::operator!=( const String& str ) const
{
- int len1 = strlen( string );
+ return ! operator==( str );
+}
+
+
+/****
+ * Operators for single characters
+ */
+String& String::operator=( char str )
+{
+ string[ 0 ] = str;
+ string[ 1 ] = 0;
+ return *this;
+}
+
+String& String::operator+=( const char str )
+{
+ const int len1 = strlen( string );
if( len1 + 1 < length )
{
string[ len1 ] = str;
@@ -173,106 +230,92 @@ String& String :: operator += ( const char str )
string[ len1 + 1 ] = 0;
}
- return * this;
-}
-
-String& String :: operator += ( const String& str )
-{
- return operator += ( str. getString() );
+ return *this;
}
-String String :: operator + ( const String& str ) const
+String String::operator+( char str ) const
{
return String( *this ) += str;
}
-String String :: operator + ( const char* str ) const
+bool String::operator==( char str ) const
{
- return String( *this ) += str;
+ return *this == String( str );
}
-bool String :: operator == ( const String& str ) const
+bool String::operator!=( char str ) const
{
- assert( string && str. string );
- if( str. length != length )
- return false;
- if( strcmp( string, str. string ) == 0 )
- return true;
- return false;
+ return ! operator==( str );
}
-bool String :: operator != ( const String& str ) const
-{
- return ! operator == ( str );
-}
-bool String :: operator == ( const char* str ) const
-{
- //cout << ( void* ) string << " " << ( void* ) str << std::endl;
- assert( string && str );
- if( strcmp( string, str ) == 0 ) return true;
- return false;
-}
-
-String :: operator bool () const
+String::operator bool () const
{
if( string[ 0 ] ) return true;
return false;
}
-bool String :: operator != ( const char* str ) const
+bool String::operator!() const
{
- return ! operator == ( str );
+ return ! operator bool();
}
-int String :: getLength() const
+String String::replace( const String& pattern,
+ const String& replaceWith,
+ int count ) const
{
- return strlen( string );
-}
-
-void
-String::
-replace( const String& pattern,
- const String& replaceWith )
-{
- int occurences( 0 );
- int patternLength = pattern.getLength();
const int length = this->getLength();
- int patternPointer( 0 );
+ const int patternLength = pattern.getLength();
+ const int replaceWithLength = replaceWith.getLength();
+
+ int patternPointer = 0;
+ int occurrences = 0;
for( int i = 0; i < length; i++ )
{
if( this->string[ i ] == pattern[ patternPointer ] )
patternPointer++;
if( patternPointer == patternLength )
{
- occurences++;
+ occurrences++;
patternPointer = 0;
}
}
- const int replaceWithLength = replaceWith.getLength();
- int newStringLength = length + occurences * ( replaceWithLength - patternLength );
- char* newString = new char[ newStringLength ];
- int newStringPointer( 0 );
- int lastPatternStart( 0 );
- for( int i = 0; i < length; i++ )
- {
+ if( count > 0 && occurrences > count )
+ occurrences = count;
+
+ String newString;
+ const int newStringLength = length + occurrences * ( replaceWithLength - patternLength );
+ newString.setSize( newStringLength );
+
+ int newStringHead = 0;
+ patternPointer = 0;
+ occurrences = 0;
+ for( int i = 0; i < length; i++ ) {
+ // copy current character
+ newString[ newStringHead++ ] = this->string[ i ];
+
+ // check if pattern matches
if( this->string[ i ] == pattern[ patternPointer ] )
- {
- if( patternPointer == 0 )
- lastPatternStart = newStringPointer;
patternPointer++;
- }
- newString[ newStringPointer++ ] = this->string[ i ];
- if( patternPointer == patternLength )
- {
- newStringPointer = lastPatternStart;
- for( int j = 0; j < replaceWithLength; j++ )
- newString[ newStringPointer++ ] = replaceWith[ j ];
+ else
+ patternPointer = 0;
+
+ // handle full match
+ if( patternPointer == patternLength ) {
+ // skip unwanted replacements
+ if( count == 0 || occurrences < count ) {
+ newStringHead -= patternLength;
+ for( int j = 0; j < replaceWithLength; j++ )
+ newString[ newStringHead++ ] = replaceWith[ j ];
+ }
+ occurrences++;
patternPointer = 0;
}
}
- delete[] this->string;
- this->string = newString;
+
+ newString[ newStringHead ] = 0;
+
+ return newString;
}
String
@@ -292,115 +335,48 @@ String::strip( char strip ) const
return "";
}
-
-const char* String :: getString() const
+int String::split( Containers::List< String >& list, const char separator ) const
{
- return string;
-}
-
-char* String :: getString()
-{
- return string;
-}
-
-
-bool String :: save( std::ostream& file ) const
-{
- assert( string );
-
- int len = strlen( string );
- file. write( ( char* ) &len, sizeof( int ) );
- file. write( string, len );
- if( file. bad() ) return false;
- return true;
-}
-
-bool String :: load( std::istream& file )
-{
- int _length;
- file. read( ( char* ) &_length, sizeof( int ) );
- if( file. bad() ) return false;
- if( ! _length )
- {
- string[ 0 ] = 0;
- length = 0;
- return true;
- }
- if( string && length < _length )
- {
- delete[] string;
- string = NULL;
- }
- if( ! string )
+ list.reset();
+ String copy( *this );
+ int len = copy.getLength();
+ for( int i = 0; i < len; i ++ )
+ if( copy[ i ] == separator )
+ copy[ i ] = 0;
+ for( int i = 0; i < len; i ++ )
{
- //dbgCout( "Reallocating string..." );
- length = STRING_PAGE * ( _length / STRING_PAGE + 1 );
- string = new char[ length ];
+ if( copy[ i ] == 0 ) continue;
+ String new_string;
+ new_string.setString( ©.getString()[ i ] );
+ i += new_string.getLength();
+ list.Append( new_string );
}
-
- file. read( string, _length );
- if( file. bad() ) return false;
- string[ _length ] = 0;
- return true;
+ return list.getSize();
}
-bool String :: save( File& file ) const
+
+bool String::save( File& file ) const
{
TNL_ASSERT( string,
std::cerr << "string = " << string );
int len = strlen( string );
-#ifdef HAVE_NOT_CXX11
- if( ! file. write< int, Devices::Host >( &len ) )
-#else
- if( ! file. write( &len ) )
-#endif
+ if( ! file.write( &len ) )
return false;
-#ifdef HAVE_NOT_CXX11
- if( ! file. write< char, Devices::Host, int >( string, len ) )
-#else
- if( ! file. write( string, len ) )
-#endif
+ if( ! file.write( string, len ) )
return false;
return true;
}
-bool String :: load( File& file )
+bool String::load( File& file )
{
int _length;
-#ifdef HAVE_NOT_CXX11
- if( ! file. read< int, Devices::Host >( &_length ) )
-#else
- if( ! file. read( &_length ) )
-#endif
- {
+ if( ! file.read( &_length ) ) {
std::cerr << "I was not able to read String length." << std::endl;
return false;
}
- if( ! _length )
- {
- string[ 0 ] = 0;
- length = 0;
- return true;
- }
- if( string && length < _length )
- {
- delete[] string;
- string = NULL;
- }
- if( ! string )
- {
- //dbgCout( "Reallocating string..." );
- length = STRING_PAGE * ( _length / STRING_PAGE + 1 );
- string = new char[ length ];
- }
-
-#ifdef HAVE_NOT_CXX11
- if( ! file. read< char, Devices::Host, int >( string, _length ) )
-#else
- if( ! file. read( string, _length ) )
-#endif
- {
+ setSize( _length );
+ if( _length && ! file.read( string, _length ) ) {
std::cerr << "I was not able to read a String with a length " << length << "." << std::endl;
return false;
}
@@ -414,7 +390,7 @@ bool String :: load( File& file )
dbgFunctionName( "mString", "MPIBcast" );
int iproc;
MPI_Comm_rank( MPI_COMM_WORLD, &iproc );
- assert( string );
+ TNL_ASSERT( string, );
int len = strlen( string );
MPI_Bcast( &len, 1, MPI_INT, root, comm );
dbgExpr( iproc );
@@ -437,40 +413,26 @@ bool String :: load( File& file )
*/
bool String :: getLine( std::istream& stream )
{
- std :: string str;
+ std::string str;
getline( stream, str );
- this->setString( str. data() );
+ this->setString( str.c_str() );
if( ! ( *this ) ) return false;
return true;
}
-int String :: parse( Containers::List< String >& list, const char separator ) const
+String operator+( char string1, const String& string2 )
{
- list.reset();
- String copy( *this );
- int len = copy. getLength();
- for( int i = 0; i < len; i ++ )
- if( copy[ i ] == separator )
- copy[ i ] = 0;
- for( int i = 0; i < len; i ++ )
- {
- if( copy[ i ] == 0 ) continue;
- String new_string;
- new_string. setString( ©. getString()[ i ] );
- i += new_string. getLength();
- list. Append( new_string );
- }
- return list. getSize();
+ return String( string1 ) + string2;
}
-String operator + ( const char* string1, const String& string2 )
+String operator+( const char* string1, const String& string2 )
{
return String( string1 ) + string2;
}
-std::ostream& operator << ( std::ostream& stream, const String& str )
+std::ostream& operator<<( std::ostream& stream, const String& str )
{
- stream << str. getString();
+ stream << str.getString();
return stream;
}
diff --git a/src/TNL/String.h b/src/TNL/String.h
index 88a52704e576e5d1032f13633ebf79c4267fb845..66c8068968f0da79ddb2ecedd418b4445dbc500b 100644
--- a/src/TNL/String.h
+++ b/src/TNL/String.h
@@ -10,7 +10,6 @@
#pragma once
-#include <stdio.h>
#include <iostream>
#include <sstream>
#include <TNL/mpi-supp.h>
@@ -23,6 +22,9 @@ namespace Containers {
template< class T > class List;
}
+template< typename T >
+String convertToString( const T& value );
+
//! Class for managing strings
class String
{
@@ -32,8 +34,7 @@ class String
//! Length of the allocated piece of memory
int length;
- public:
-
+public:
//! Basic constructor
String();
@@ -42,37 +43,42 @@ class String
@param sufix_cut_off says the same about sufix.
*/
String( const char* c,
- int prefix_cut_off = 0,
- int sufix_cut_off = 0 );
+ int prefix_cut_off = 0,
+ int sufix_cut_off = 0 );
static String getType();
//! Copy constructor
String( const String& str );
- //! Convert number to a string
- String( unsigned number );
-
- String( int number );
-
- String( unsigned long int number );
+ //! Convert anything to a string
+ template< typename T >
+ String( T value )
+ : string( nullptr ), length( 0 )
+ {
+ setString( convertToString( value ).getString() );
+ }
- String( long int number );
+ //! Destructor
+ ~String();
- String( float number );
+ //! Return length of the string
+ int getLength() const;
+ int getSize() const;
- String( double number );
+ //! Return currently allocated size
+ int getAllocatedSize() const;
- //! Destructor
- ~String();
+ //! Reserve space for given number of characters
+ void setSize( int size );
//! Set string from given char pointer
/*! @param prefix_cut_off says length of the prefix that is going to be omitted and
@param sufix_cut_off says the same about sufix.
*/
void setString( const char* c,
- int prefix_cut_off = 0,
- int sufix_cut_off = 0 );
+ int prefix_cut_off = 0,
+ int sufix_cut_off = 0 );
//! Return pointer to data
const char* getString() const;
@@ -87,58 +93,46 @@ class String
char& operator[]( int i );
/****
- * TODO: the operators do not work properly
- * for example String + const char*
+ * Operators for C strings
*/
-
- //! Operator =
- String& operator = ( const String& str );
-
- //! Operator +=
- String& operator += ( const char* str );
-
- //! Operator +=
- String& operator += ( const char str );
-
- //! Operator +=
- String& operator += ( const String& str );
+ String& operator=( const char* str );
+ String& operator+=( const char* str );
+ String operator+( const char* str ) const;
+ bool operator==( const char* str ) const;
+ bool operator!=( const char* str ) const;
- //! Operator +
- String operator + ( const String& str ) const;
-
- //! Operator +
- String operator + ( const char* str ) const;
+ /****
+ * Operators for Strings
+ */
+ String& operator=( const String& str );
+ String& operator+=( const String& str );
+ String operator+( const String& str ) const;
+ bool operator==( const String& str ) const;
+ bool operator!=( const String& str ) const;
- //! Comparison operator
- bool operator == ( const String& str ) const;
+ /****
+ * Operators for single characters
+ */
+ String& operator=( char str );
+ String& operator+=( char str );
+ String operator+( char str ) const;
+ bool operator==( char str ) const;
+ bool operator!=( char str ) const;
- //! Comparison operator
- bool operator != ( const String& str ) const;
+ //! Cast to bool operator
+ operator bool() const;
- //! Comparison operator
- bool operator == ( const char* ) const;
+ //! Cast to bool with negation operator
+ bool operator!() const;
- //! Comparison operator
- bool operator != ( const char* ) const;
-
- //! Retyping operator
- operator bool () const;
-
- //! Return length of the string
- int getLength() const;
-
- void replace( const String& pattern,
- const String& replaceWith );
+ String replace( const String& pattern,
+ const String& replaceWith,
+ int count = 0 ) const;
String strip( char strip = ' ' ) const;
- // TODO: remove
- //! Write to a binary file
- bool save( std::ostream& file ) const;
-
- // TODO: remove
- //! Read from binary file
- bool load( std::istream& file );
+ //! Split the string into list of strings w.r.t. given separator.
+ int split( Containers::List< String >& list, const char separator = ' ' ) const;
//! Write to a binary file
bool save( File& file ) const;
@@ -152,15 +146,14 @@ class String
//! Read one line from given stream.
bool getLine( std::istream& stream );
- //! Parse the string into list of strings w.r.t. given separator.
- int parse( Containers::List< String >& list, const char separator = ' ' ) const;
-
- friend std::ostream& operator << ( std::ostream& stream, const String& str );
+ friend std::ostream& operator<<( std::ostream& stream, const String& str );
};
-String operator + ( const char* string1, const String& string2 );
+String operator+( char string1, const String& string2 );
+
+String operator+( const char* string1, const String& string2 );
-std::ostream& operator << ( std::ostream& stream, const String& str );
+std::ostream& operator<<( std::ostream& stream, const String& str );
template< typename T >
String convertToString( const T& value )
diff --git a/src/TNL/UniquePointer.h b/src/TNL/UniquePointer.h
index f8e85e04756e218463d7d6f21c85792238f0cb79..606f86cc4326c57cd73f7efa525dbcc4b344306b 100644
--- a/src/TNL/UniquePointer.h
+++ b/src/TNL/UniquePointer.h
@@ -18,6 +18,8 @@
#include <cstring>
+#include "Devices/MIC.h"
+
namespace TNL {
@@ -61,11 +63,18 @@ class UniquePointer< Object, Devices::Host > : public SmartPointer
return *( this->pointer );
}
- operator bool()
+ __cuda_callable__
+ operator bool() const
{
return this->pointer;
}
+ __cuda_callable__
+ bool operator!() const
+ {
+ return ! this->pointer;
+ }
+
template< typename Device = Devices::Host >
const Object& getData() const
{
@@ -148,11 +157,18 @@ class UniquePointer< Object, Devices::Cuda > : public SmartPointer
return this->pd->data;
}
- operator bool()
+ __cuda_callable__
+ operator bool() const
{
return this->pd;
}
+ __cuda_callable__
+ bool operator!() const
+ {
+ return ! this->pd;
+ }
+
template< typename Device = Devices::Host >
const Object& getData() const
{
@@ -203,7 +219,7 @@ class UniquePointer< Object, Devices::Cuda > : public SmartPointer
if( this->modified() )
{
cudaMemcpy( (void*) this->cuda_pointer, (void*) &this->pd->data, sizeof( Object ), cudaMemcpyHostToDevice );
- if( ! checkCudaDevice )
+ if( ! TNL_CHECK_CUDA_DEVICE )
return false;
this->set_last_sync_state();
return true;
@@ -239,12 +255,8 @@ class UniquePointer< Object, Devices::Cuda > : public SmartPointer
bool allocate( Args... args )
{
this->pd = new PointerData( args... );
- if( ! this->pd )
- return false;
// pass to device
this->cuda_pointer = Devices::Cuda::passToDevice( this->pd->data );
- if( ! this->cuda_pointer )
- return false;
// set last-sync state
this->set_last_sync_state();
Devices::Cuda::insertSmartPointer( this );
@@ -282,5 +294,194 @@ class UniquePointer< Object, Devices::Cuda > : public SmartPointer
Object* cuda_pointer;
};
+#ifdef HAVE_MIC
+template< typename Object >
+class UniquePointer< Object, Devices::MIC > : public SmartPointer
+{
+ public:
+
+ typedef Object ObjectType;
+ typedef Devices::MIC DeviceType;
+ typedef UniquePointer< Object, Devices::MIC > ThisType;
+
+ template< typename... Args >
+ explicit UniquePointer( const Args... args )
+ : pd( nullptr ),
+ cuda_pointer( nullptr )
+ {
+ this->allocate( args... );
+ }
+
+ const Object* operator->() const
+ {
+ return &this->pd->data;
+ }
+
+ Object* operator->()
+ {
+ this->pd->maybe_modified = true;
+ return &this->pd->data;
+ }
+
+ const Object& operator *() const
+ {
+ return this->pd->data;
+ }
+
+ Object& operator *()
+ {
+ this->pd->maybe_modified = true;
+ return this->pd->data;
+ }
+
+ operator bool()
+ {
+ return this->pd;
+ }
+
+ template< typename Device = Devices::Host >
+ const Object& getData() const
+ {
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::MIC >::value, "Only Devices::Host or Devices::MIC devices are accepted here." );
+ TNL_ASSERT( this->pd, );
+ TNL_ASSERT( this->mic_pointer, );
+ if( std::is_same< Device, Devices::Host >::value )
+ return this->pd->data;
+ if( std::is_same< Device, Devices::MIC >::value )
+ return *( this->mic_pointer );
+ }
+
+ template< typename Device = Devices::Host >
+ Object& modifyData()
+ {
+ static_assert( std::is_same< Device, Devices::Host >::value || std::is_same< Device, Devices::MIC >::value, "Only Devices::Host or Devices::MIC devices are accepted here." );
+ TNL_ASSERT( this->pd, );
+ TNL_ASSERT( this->mic_pointer, );
+ if( std::is_same< Device, Devices::Host >::value )
+ {
+ this->pd->maybe_modified = true;
+ return this->pd->data;
+ }
+ if( std::is_same< Device, Devices::MIC >::value )
+ return *( this->mic_pointer );
+ }
+
+ const ThisType& operator=( ThisType& ptr )
+ {
+ this->free();
+ this->pd = ptr.pd;
+ this->mic_pointer = ptr.mic_pointer;
+ ptr.pd = nullptr;
+ ptr.mic_pointer = nullptr;
+ return *this;
+ }
+
+ const ThisType& operator=( ThisType&& ptr )
+ {
+ return this->operator=( ptr );
+ }
+
+ bool synchronize()
+ {
+ if( ! this->pd )
+ return true;
+ if( this->modified() )
+ {
+ Devices::MIC::CopyToMIC(this->mic_pointer,(void*) &this->pd->data,sizeof(Object));
+ this->set_last_sync_state();
+ return true;
+ }
+ return true;//??
+ }
+
+ ~UniquePointer()
+ {
+ this->free();
+ Devices::MIC::removeSmartPointer( this );
+ }
+
+ protected:
+
+ struct PointerData
+ {
+ Object data;
+ char data_image[ sizeof(Object) ];
+ bool maybe_modified;
+
+ template< typename... Args >
+ explicit PointerData( Args... args )
+ : data( args... ),
+ maybe_modified( false )
+ {}
+ };
+
+ template< typename... Args >
+ bool allocate( Args... args )
+ {
+ this->pd = new PointerData( args... );
+ if( ! this->pd )
+ return false;
+ // pass to device
+ this->mic_pointer=(Object*)Devices::MIC::AllocMIC(sizeof(Object));
+ if( ! this->mic_pointer )
+ return false;
+ Devices::MIC::CopyToMIC((void*)mic_pointer,(void*)&this->pd->data,sizeof(Object));
+ // set last-sync state
+ this->set_last_sync_state();
+ Devices::MIC::insertSmartPointer( this );
+ return true;
+ }
+
+ void set_last_sync_state()
+ {
+ TNL_ASSERT( this->pd, );
+ std::memcpy( (void*) &this->pd->data_image, (void*) &this->pd->data, sizeof( ObjectType ) );
+ this->pd->maybe_modified = false;
+ }
+
+ bool modified()
+ {
+ TNL_ASSERT( this->pd, );
+ // optimization: skip bitwise comparison if we're sure that the data is the same
+ if( ! this->pd->maybe_modified )
+ return false;
+ return std::memcmp( (void*) &this->pd->data_image, (void*) &this->pd->data, sizeof( ObjectType ) ) != 0;
+ }
+
+ void free()
+ {
+ if( this->pd )
+ delete this->pd;
+ if( this->mic_pointer )
+ Devices::MIC::FreeMIC(mic_pointer);
+ }
+
+ PointerData* pd;
+
+ // cuda_pointer can't be part of PointerData structure, since we would be
+ // unable to dereference this-pd on the device
+ Object* mic_pointer;
+};
+#endif
+
+#if (!defined(NDEBUG)) && (!defined(HAVE_MIC))
+namespace Assert {
+
+template< typename Object, typename Device >
+struct Formatter< UniquePointer< Object, Device > >
+{
+ static std::string
+ printToString( const UniquePointer< Object, Device >& value )
+ {
+ ::std::stringstream ss;
+ ss << "(UniquePointer< " << Object::getType() << ", " << Device::getDeviceType()
+ << " > object at " << &value << ")";
+ return ss.str();
+ }
+};
+
+} // namespace Assert
+#endif
+
} // namespace TNL
diff --git a/src/TNL/legacy/CMakeLists.txt b/src/TNL/legacy/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/src/TNL/legacy/benchmarks/matrix-solvers-benchmark.h b/src/TNL/legacy/benchmarks/matrix-solvers-benchmark.h
index 33153a50ff0e192db596897a7b701ace99171f68..1bb3dfd96c4d807af244c5f3424990b9278f1824 100644
--- a/src/TNL/legacy/benchmarks/matrix-solvers-benchmark.h
+++ b/src/TNL/legacy/benchmarks/matrix-solvers-benchmark.h
@@ -15,6 +15,7 @@
#include <TNL/File.h>
#include <TNL/Object.h>
#include <TNL/Devices/Cuda.h>
+#include <TNL/Exceptions/CudaSupportMissing.h>
#include <TNL/Config/ConfigDescription.h>
#include <TNL/Config/ParameterContainer.h>
#include <TNL/Matrices/CSR.h>
@@ -78,11 +79,7 @@ bool benchmarkSolver( const Config::ParameterContainer& parameters,
solver. setSolverMonitor( solverMonitor );
solver. setRefreshRate( 10 );
solverMonitor. resetTimers();
-#ifdef HAVE_NOT_CXX11
- solver. template solve< Vector, LinearResidueGetter< Matrix, Vector > >( b, x );
-#else
solver. solve( b, x );
-#endif
bool solverConverged( solver. getResidue() < maxResidue );
const String& logFileName = parameters. getParameter< String >( "log-file" );
@@ -313,8 +310,7 @@ bool benchmarkMatrix( const Config::ParameterContainer& parameters )
return false;
x = cudaX;*/
#else
- CudaSupportMissingMessage;;
- return false;
+ throw Exceptions::CudaSupportMissing();
#endif
}
diff --git a/src/TNL/legacy/benchmarks/tnlSpmvBenchmarkAdaptiveRgCSRMatrix.h b/src/TNL/legacy/benchmarks/tnlSpmvBenchmarkAdaptiveRgCSRMatrix.h
index 846a0c8cc7590f9d332073faf3c0011a5fdb2840..35a5b388c9b317d3fbf4292337814b76d6d7122b 100644
--- a/src/TNL/legacy/benchmarks/tnlSpmvBenchmarkAdaptiveRgCSRMatrix.h
+++ b/src/TNL/legacy/benchmarks/tnlSpmvBenchmarkAdaptiveRgCSRMatrix.h
@@ -12,7 +12,9 @@
#define TNLSPMVBENCHMARKADAPTIVERGCSRMATRIX_H_
#include "tnlSpmvBenchmark.h"
+
#include <TNL/Assert.h>
+#include <TNL/Exceptions/CudaSupportMissing.h>
template< typename Real, typename Device, typename Index>
class tnlSpmvBenchmarkAdaptiveRgCSR : public tnlSpmvBenchmark< Real, Device, Index, tnlAdaptiveRgCSR >
@@ -131,7 +133,7 @@ void tnlSpmvBenchmarkAdaptiveRgCSR< Real, Device, Index > :: writeProgress() con
std::cout << right << std::setw( this->benchmarkStatusColumnWidth ) << " FAILED";
#ifndef HAVE_CUDA
if( Device :: getDevice() == Devices::CudaDevice )
- CudaSupportMissingMessage;;
+ throw Exceptions::CudaSupportMissing();
#endif
std::cout << std::endl;
}
diff --git a/src/TNL/legacy/benchmarks/tnlSpmvBenchmarkRgCSRMatrix.h b/src/TNL/legacy/benchmarks/tnlSpmvBenchmarkRgCSRMatrix.h
index b0be71e993fde254731aef550d5916401eec073c..6327ac95d659ebc6592d458f27288a11cf34d141 100644
--- a/src/TNL/legacy/benchmarks/tnlSpmvBenchmarkRgCSRMatrix.h
+++ b/src/TNL/legacy/benchmarks/tnlSpmvBenchmarkRgCSRMatrix.h
@@ -13,6 +13,8 @@
#include "tnlSpmvBenchmark.h"
+#include <TNL/Exceptions/CudaSupportMissing.h>
+
template< typename Real, typename Device, typename Index>
class tnlSpmvBenchmarkRgCSR : public tnlSpmvBenchmark< Real, Device, Index, tnlRgCSR >
{
@@ -134,7 +136,7 @@ void tnlSpmvBenchmarkRgCSR< Real, Device, Index > :: writeProgress() const
std::cout << right << std::setw( this->benchmarkStatusColumnWidth ) << " FAILED - maxError is " << this->maxError << ". ";
#ifndef HAVE_CUDA
if( Device :: getDevice() == Devices::CudaDevice )
- CudaSupportMissingMessage;;
+ throw Exceptions::CudaSupportMissing();
#endif
std::cout << std::endl;
}
diff --git a/src/TNL/param-types.h b/src/TNL/param-types.h
index 790cf627abc0c58094e3113b664019130d53e825..8a13e09565dd3ede7c2414c9504b0e12ab142a5d 100644
--- a/src/TNL/param-types.h
+++ b/src/TNL/param-types.h
@@ -16,7 +16,7 @@
namespace TNL {
template< typename T >
-String getType() { return T :: getType(); };
+String getType() { return T::getType(); };
template<> inline String getType< bool >() { return String( "bool" ); };
template<> inline String getType< short int >() { return String( "short int" ); };
diff --git a/src/Tools/CMakeLists.txt b/src/Tools/CMakeLists.txt
old mode 100755
new mode 100644
index 3934359e9ba8448dc0fdf69bddac7a3101eb70ab..7069abf67277eda2f4d77835c868d166482715b3
--- a/src/Tools/CMakeLists.txt
+++ b/src/Tools/CMakeLists.txt
@@ -56,12 +56,7 @@ ADD_EXECUTABLE(tnl-curve2gnuplot${debugExt} ${tnlcurve2gnuplotsources})
target_link_libraries (tnl-curve2gnuplot${debugExt} tnl${debugExt}-${tnlVersion} )
IF( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE( tnl-cuda-arch${debugExt} tnl-cuda-arch.cu
- OPTIONS ${CUDA_ADD_EXECUTABLE_OPTIONS} )
- SET_TARGET_PROPERTIES( tnl-cuda-arch${debugExt} PROPERTIES CUDA_COMPILE_FLAGS "${CXX_OPTIMIZE_FLAGS}" )
-ENDIF()
-
-IF( BUILD_CUDA )
+ CUDA_ADD_EXECUTABLE( tnl-cuda-arch${debugExt} tnl-cuda-arch.cu )
INSTALL( TARGETS tnl-cuda-arch${debugExt}
RUNTIME DESTINATION bin
PERMISSIONS OWNER_READ OWNER_WRITE OWNER_EXECUTE GROUP_READ GROUP_EXECUTE WORLD_READ WORLD_EXECUTE )
diff --git a/src/Tools/tnl-cuda-arch.cu b/src/Tools/tnl-cuda-arch.cu
index 7f1fb17496601adc5ffc84467a9c5db25feae244..7d880a19a32b7f41bf0fd45d7949f73f8c5933e9 100644
--- a/src/Tools/tnl-cuda-arch.cu
+++ b/src/Tools/tnl-cuda-arch.cu
@@ -1,6 +1,8 @@
#include <stdio.h>
+#include <string.h>
-int main() {
+int main( int argc, char** argv )
+{
int num_devices = 0;
cudaError_t error_id = cudaGetDeviceCount( &num_devices );
@@ -21,8 +23,15 @@ int main() {
if( i > 0 )
printf(" ");
- printf( "-gencode arch=compute_%d%d,code=sm_%d%d",
- prop.major, compute_minor, prop.major, prop.minor );
+
+ if( argc == 2 && strcmp( argv[1], "--clang" ) == 0 ) {
+ printf( "--cuda-gpu-arch=sm_%d%d",
+ prop.major, prop.minor );
+ }
+ else {
+ printf( "-gencode arch=compute_%d%d,code=sm_%d%d",
+ prop.major, compute_minor, prop.major, prop.minor );
+ }
}
printf("\n");
}
diff --git a/src/Tools/tnl-diff.cpp b/src/Tools/tnl-diff.cpp
index 76c2e792bc5e7b2a8f09304ac038334506c7e586..013f7bf21a2218f940a037ca94e39709a14b0505 100644
--- a/src/Tools/tnl-diff.cpp
+++ b/src/Tools/tnl-diff.cpp
@@ -60,7 +60,7 @@ int main( int argc, char* argv[] )
if( ! parseObjectType( meshType, parsedMeshType ) )
{
std::cerr << "Unable to parse the mesh type " << meshType << "." << std::endl;
- return false;
+ return EXIT_FAILURE;
}
if( parsedMeshType[ 0 ] == "Meshes::Grid" ||
parsedMeshType[ 0 ] == "tnlGrid" ) // TODO: remove deprecated type name
diff --git a/src/Tools/tnl-init.cpp b/src/Tools/tnl-init.cpp
index 90870bad68daa3761c20c44b2d04e24ad22c5677..f42b9ad3c9aeb02690ef4f7def74acf6c2ce3087 100644
--- a/src/Tools/tnl-init.cpp
+++ b/src/Tools/tnl-init.cpp
@@ -65,7 +65,7 @@ int main( int argc, char* argv[] )
if( ! parseObjectType( meshType, parsedMeshType ) )
{
std::cerr << "Unable to parse the mesh type " << meshType << "." << std::endl;
- return false;
+ return EXIT_FAILURE;
}
if( ! resolveMeshType( parsedMeshType, parameters ) )
return EXIT_FAILURE;
diff --git a/src/Tools/tnl-init.h b/src/Tools/tnl-init.h
index 182f51b28536b06ac86d5e76c3d997e01367c83a..355c553f02695d28f45c3994ef80891ead05b4cd 100644
--- a/src/Tools/tnl-init.h
+++ b/src/Tools/tnl-init.h
@@ -34,14 +34,14 @@ bool renderFunction( const Config::ParameterContainer& parameters )
if( ! meshPointer->load( meshFile ) )
return false;
- typedef Functions::TestFunction< MeshType::meshDimension, RealType > FunctionType;
+ typedef Functions::TestFunction< MeshType::getMeshDimension(), RealType > FunctionType;
typedef SharedPointer< FunctionType, typename MeshType::DeviceType > FunctionPointer;
FunctionPointer function;
std::cout << "Setting up the function ... " << std::endl;
if( ! function->setup( parameters, "" ) )
return false;
std::cout << "done." << std::endl;
- typedef Functions::MeshFunction< MeshType, MeshType::meshDimension > MeshFunctionType;
+ typedef Functions::MeshFunction< MeshType, MeshType::getMeshDimension() > MeshFunctionType;
typedef SharedPointer< MeshFunctionType, typename MeshType::DeviceType > MeshFunctionPointer;
MeshFunctionPointer meshFunction( meshPointer );
//if( ! discreteFunction.setSize( mesh.template getEntitiesCount< typename MeshType::Cell >() ) )
diff --git a/src/Tools/tnl-mesh-convert.cpp b/src/Tools/tnl-mesh-convert.cpp
index 84cfd7fbaa57c10679127a365383494ac3f8dd29..d7de402bc72bc3cc84bd03452837cb7bff40833d 100644
--- a/src/Tools/tnl-mesh-convert.cpp
+++ b/src/Tools/tnl-mesh-convert.cpp
@@ -8,7 +8,7 @@
/* See Copyright Notice in tnl/Copyright */
-#ifndef HAVE_ICPC
+#ifndef __INTEL_COMPILER
#include "tnl-mesh-convert.h"
#endif
#include <TNL/Config/ParameterContainer.h>
@@ -35,7 +35,7 @@ int main( int argc, char* argv[] )
conf_desc.printUsage( argv[ 0 ] );
return EXIT_FAILURE;
}
-#ifndef HAVE_ICPC
+#ifndef __INTEL_COMPILER
if( ! convertMesh( parameters ) )
return EXIT_FAILURE;
#endif
diff --git a/src/Tools/tnl-quickstart/Makefile.in b/src/Tools/tnl-quickstart/Makefile.in
index 48f3908b58d9888aa6cda6959ba65c96745793c7..c2954e00e8f855f14bd52eda44fc77f95c8a9416 100644
--- a/src/Tools/tnl-quickstart/Makefile.in
+++ b/src/Tools/tnl-quickstart/Makefile.in
@@ -1,6 +1,3 @@
-# Uncomment the following line to enable CUDA
-#WITH_CUDA = yes
-
TARGET = {problemBaseName}
INSTALL_DIR = ${{HOME}}/local
@@ -16,11 +13,12 @@ endif
SOURCES = {problemBaseName}.cpp
HEADERS = {problemBaseName}.h
-OBJECTS = {problemBaseName}.o
DIST = $(SOURCES) $(CUDA_SOURCES) $(HEADERS) Makefile
ifdef WITH_CUDA
OBJECTS = {problemBaseName}-cuda.o
+else
+ OBJECTS = {problemBaseName}.o
endif
all: $(TARGET)
diff --git a/src/Tools/tnl-quickstart/explicit-laplace-grid-1d_impl.h.in b/src/Tools/tnl-quickstart/explicit-laplace-grid-1d_impl.h.in
index d952cb637c3ca0afdedd577a7b6341d65c8e211e..1e1eaa44a175c7688f1b6864e9f7e5ffaea1fa5e 100644
--- a/src/Tools/tnl-quickstart/explicit-laplace-grid-1d_impl.h.in
+++ b/src/Tools/tnl-quickstart/explicit-laplace-grid-1d_impl.h.in
@@ -1,5 +1,5 @@
const RealType& hxSquareInverse = entity.getMesh().template getSpaceStepsProducts< -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
return ( u[ west ] - 2.0 * u[ center ] + u[ east ] ) * hxSquareInverse;
\ No newline at end of file
diff --git a/src/Tools/tnl-quickstart/explicit-laplace-grid-2d_impl.h.in b/src/Tools/tnl-quickstart/explicit-laplace-grid-2d_impl.h.in
index ad817b3b31fc422759a76d4093336e67dee6a6dc..8c3b7d796ce12d8bb96fda55e47d55568bd8d5bd 100644
--- a/src/Tools/tnl-quickstart/explicit-laplace-grid-2d_impl.h.in
+++ b/src/Tools/tnl-quickstart/explicit-laplace-grid-2d_impl.h.in
@@ -1,9 +1,9 @@
const RealType& hxSquareInverse = entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const RealType& hySquareInverse = entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
return ( u[ west ] - 2.0 * u[ center ] + u[ east ] ) * hxSquareInverse +
( u[ south ] - 2.0 * u[ center ] + u[ north ] ) * hySquareInverse;
\ No newline at end of file
diff --git a/src/Tools/tnl-quickstart/explicit-laplace-grid-3d_impl.h.in b/src/Tools/tnl-quickstart/explicit-laplace-grid-3d_impl.h.in
index 7b2f234c2037ee8de516d2d04a0877e307e64e12..aa6ff5f3f7fcc731c57667cc3e12c52ca0087218 100644
--- a/src/Tools/tnl-quickstart/explicit-laplace-grid-3d_impl.h.in
+++ b/src/Tools/tnl-quickstart/explicit-laplace-grid-3d_impl.h.in
@@ -2,12 +2,12 @@
const RealType& hySquareInverse = entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const RealType& hzSquareInverse = entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
return ( u[ west ] - 2.0 * u[ center ] + u[ east ] ) * hxSquareInverse +
( u[ south ] - 2.0 * u[ center ] + u[ north ] ) * hySquareInverse +
( u[ up ] - 2.0 * u[ center ] + u[ down ] ) * hzSquareInverse;
\ No newline at end of file
diff --git a/src/Tools/tnl-quickstart/implicit-laplace-grid-1d_impl.h.in b/src/Tools/tnl-quickstart/implicit-laplace-grid-1d_impl.h.in
index de086c72e305d76b2130e0a2e09da73f68c40fd1..e5a139e22c202672510f59b607c5d099d2772c4e 100644
--- a/src/Tools/tnl-quickstart/implicit-laplace-grid-1d_impl.h.in
+++ b/src/Tools/tnl-quickstart/implicit-laplace-grid-1d_impl.h.in
@@ -1,7 +1,7 @@
const RealType& lambdaX = tau * entity.getMesh().template getSpaceStepsProducts< -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
matrixRow.setElement( 0, west, - lambdaX );
matrixRow.setElement( 1, center, 2.0 * lambdaX );
matrixRow.setElement( 2, east, - lambdaX );
\ No newline at end of file
diff --git a/src/Tools/tnl-quickstart/implicit-laplace-grid-2d_impl.h.in b/src/Tools/tnl-quickstart/implicit-laplace-grid-2d_impl.h.in
index 048949469ddff7d49bbd7ca82e74ff322fa789cd..808d829a88028d9e75c6a6bbaa1b092d3ce4d361 100644
--- a/src/Tools/tnl-quickstart/implicit-laplace-grid-2d_impl.h.in
+++ b/src/Tools/tnl-quickstart/implicit-laplace-grid-2d_impl.h.in
@@ -1,10 +1,10 @@
const RealType& lambdaX = tau * entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const RealType& lambdaY = tau * entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
matrixRow.setElement( 0, south, -lambdaY );
matrixRow.setElement( 1, west, -lambdaX );
matrixRow.setElement( 2, center, 2.0 * ( lambdaX + lambdaY ) );
diff --git a/src/Tools/tnl-quickstart/implicit-laplace-grid-3d_impl.h.in b/src/Tools/tnl-quickstart/implicit-laplace-grid-3d_impl.h.in
index 2343334c4044143815dc5ebfad19fc2c26e767c3..7bbec43866834f69db96e250aaaec9af8c717586 100644
--- a/src/Tools/tnl-quickstart/implicit-laplace-grid-3d_impl.h.in
+++ b/src/Tools/tnl-quickstart/implicit-laplace-grid-3d_impl.h.in
@@ -2,12 +2,12 @@
const RealType& lambdaY = tau * entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const RealType& lambdaZ = tau * entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
matrixRow.setElement( 0, down, -lambdaZ );
matrixRow.setElement( 1, south, -lambdaY );
matrixRow.setElement( 2, west, -lambdaX );
diff --git a/src/Tools/tnl-quickstart/operator-grid-specialization_impl.h.in b/src/Tools/tnl-quickstart/operator-grid-specialization_impl.h.in
index 1714c59b7395e62486d53775e424d04fa8f7a7eb..3e7d0d670f5de48659c146bbf7b1fb62287e4a38 100644
--- a/src/Tools/tnl-quickstart/operator-grid-specialization_impl.h.in
+++ b/src/Tools/tnl-quickstart/operator-grid-specialization_impl.h.in
@@ -36,7 +36,7 @@ operator()( const MeshFunction& u,
*/
static_assert( MeshEntity::entityDimension == {meshDimension}, "Wrong mesh entity dimensions." );
static_assert( MeshFunction::getEntitiesDimension() == {meshDimension}, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< {meshDimension} >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< {meshDimension} >& neighborEntities = entity.getNeighborEntities();
{explicitScheme}
}}
@@ -92,7 +92,7 @@ setMatrixElements( const PreimageFunction& u,
* The following example is the Laplace operator approximated
* by the Finite difference method.
*/
- const typename MeshEntity::template NeighbourEntities< {meshDimension} >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< {meshDimension} >& neighborEntities = entity.getNeighborEntities();
const IndexType& index = entity.getIndex();
typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
{semiimplicitScheme}
diff --git a/src/Tools/tnl-quickstart/tnl-quickstart.py b/src/Tools/tnl-quickstart/tnl-quickstart.py
index 5992503861af6ce1faa274a0796237847e2ff53e..f456952696ad52d06fb7e3806e885dbc62d33be1 100644
--- a/src/Tools/tnl-quickstart/tnl-quickstart.py
+++ b/src/Tools/tnl-quickstart/tnl-quickstart.py
@@ -18,7 +18,7 @@ print( "----------------------------------")
definitions = {}
-definitions['problemName'] = input( "Problam name:" )
+definitions['problemName'] = input( "Problem name:" )
definitions['problemBaseName'] = input( "Problem class base name (base name acceptable in C++ code):" )
definitions['operatorName'] = input( "Operator name:")
diff --git a/src/Tools/tnl-view.cpp b/src/Tools/tnl-view.cpp
index 3f4e19304eda56866e0350e888c728aa8457b87f..4a5865ac610e5d0713899f9a72678ab878f13d76 100644
--- a/src/Tools/tnl-view.cpp
+++ b/src/Tools/tnl-view.cpp
@@ -75,7 +75,7 @@ int main( int argc, char* argv[] )
if( ! parseObjectType( meshType, parsedMeshType ) )
{
std::cerr << "Unable to parse the mesh type " << meshType << "." << std::endl;
- return false;
+ return EXIT_FAILURE;
}
if( parsedMeshType[ 0 ] == "Meshes::Grid" ||
parsedMeshType[ 0 ] == "tnlGrid" ) // TODO: remove deprecated type name
diff --git a/src/Tools/tnl-view.h b/src/Tools/tnl-view.h
index f4e6542ecbacc9e084ac145df56dbe4938a3bcab..c4d25f1d5520d2886be390bc41763c1f338e45aa 100644
--- a/src/Tools/tnl-view.h
+++ b/src/Tools/tnl-view.h
@@ -282,12 +282,12 @@ bool convertObject( const MeshPointer& meshPointer,
parsedObjectType[ 0 ] == "tnlVector" ) //
{
using MeshType = typename MeshPointer::ObjectType;
- // FIXME: why is MeshType::IndexType not the same as Index?
+ // FIXME: why is MeshType::GlobalIndexType not the same as Index?
// Containers::Vector< Element, Devices::Host, Index > vector;
- Containers::Vector< Element, Devices::Host, typename MeshType::IndexType > vector;
+ Containers::Vector< Element, Devices::Host, typename MeshType::GlobalIndexType > vector;
if( ! vector.load( inputFileName ) )
return false;
- Functions::MeshFunction< MeshType, MeshType::meshDimension, Element > mf;
+ Functions::MeshFunction< MeshType, MeshType::getMeshDimension(), Element > mf;
mf.bind( meshPointer, vector );
if( ! mf.write( outputFileName, outputFormat ) )
return false;
diff --git a/src/Tools/tnlcurve2gnuplot.cpp b/src/Tools/tnlcurve2gnuplot.cpp
index 16f7f372f37ea7881506aac6d7894036676a0256..5a1e297d60973b8a1a6f8f7f053732cdcce9fb0c 100644
--- a/src/Tools/tnlcurve2gnuplot.cpp
+++ b/src/Tools/tnlcurve2gnuplot.cpp
@@ -58,7 +58,7 @@ int main( int argc, char* argv[] )
std::cout << "Processing file " << input_file << " ... " << std::flush;
File file;
- if( ! file. open( input_files[ i ], tnlReadMode ) )
+ if( ! file. open( input_files[ i ], IOMode::read ) )
{
std::cout << " unable to open file " << input_files[ i ] << std::endl;
continue;
diff --git a/src/UnitTests/CMakeLists.txt b/src/UnitTests/CMakeLists.txt
old mode 100755
new mode 100644
index ac80ff9471a888c101a4d4f699eb04c1734e3427..f702830a06cc20b76171ec10fffbe9a89711cce4
--- a/src/UnitTests/CMakeLists.txt
+++ b/src/UnitTests/CMakeLists.txt
@@ -1,39 +1,43 @@
-#ADD_SUBDIRECTORY( Containers )
+if( WITH_TESTS STREQUAL "yes" )
-ADD_EXECUTABLE( UniquePointerTest${mpiExt}${debugExt} ${headers} UniquePointerTest.cpp )
-SET_TARGET_PROPERTIES( UniquePointerTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
-TARGET_LINK_LIBRARIES( UniquePointerTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
+ADD_SUBDIRECTORY( Containers )
+ADD_SUBDIRECTORY( Matrices )
+
+ADD_EXECUTABLE( UniquePointerTest${mpiExt}${debugExt} UniquePointerTest.cpp )
+TARGET_COMPILE_OPTIONS( UniquePointerTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+TARGET_LINK_LIBRARIES( UniquePointerTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
IF( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE( FileTest${mpiExt}${debugExt} FileTest.h FileTest.cu )
- SET_TARGET_PROPERTIES( FileTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( FileTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
+ CUDA_ADD_EXECUTABLE( FileTest${mpiExt}${debugExt} FileTest.cu OPTIONS ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( FileTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
ELSE( BUILD_CUDA )
- ADD_EXECUTABLE( FileTest${mpiExt}${debugExt} FileTest.h FileTest.cpp )
- SET_TARGET_PROPERTIES( FileTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( FileTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
+ ADD_EXECUTABLE( FileTest${mpiExt}${debugExt} FileTest.cpp )
+ TARGET_COMPILE_OPTIONS( FileTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( FileTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
ENDIF( BUILD_CUDA )
-ADD_EXECUTABLE( StringTest${mpiExt}${debugExt} ${headers} StringTest.cpp )
-SET_TARGET_PROPERTIES( StringTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
-TARGET_LINK_LIBRARIES( StringTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
-
-ADD_EXECUTABLE( ListTest${mpiExt}${debugExt} ${headers} ListTest.cpp )
-SET_TARGET_PROPERTIES( ListTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
-TARGET_LINK_LIBRARIES( ListTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
+ADD_EXECUTABLE( StringTest${mpiExt}${debugExt} StringTest.cpp )
+TARGET_COMPILE_OPTIONS( StringTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+TARGET_LINK_LIBRARIES( StringTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
-ADD_EXECUTABLE( ObjectTest${mpiExt}${debugExt} ${headers} ObjectTest.cpp )
-SET_TARGET_PROPERTIES( ObjectTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
-TARGET_LINK_LIBRARIES( ObjectTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
+ADD_EXECUTABLE( ObjectTest${mpiExt}${debugExt} ObjectTest.cpp )
+TARGET_COMPILE_OPTIONS( ObjectTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+TARGET_LINK_LIBRARIES( ObjectTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
ADD_TEST( FileTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/FileTest${mpiExt}${debugExt} )
ADD_TEST( StringTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/StringTest${mpiExt}${debugExt} )
-ADD_TEST( ListTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/ListTest${mpiExt}${debugExt} )
-ADD_TEST( ObjectTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/ObjectTest${mpiExt}${debugExt} )
\ No newline at end of file
+ADD_TEST( ObjectTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/ObjectTest${mpiExt}${debugExt} )
+ADD_TEST( UniquePointerTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/UniquePointerTest${mpiExt}${debugExt} )
+
+endif( WITH_TESTS STREQUAL "yes" )
diff --git a/src/UnitTests/Containers/ArrayOperationsTest.cpp b/src/UnitTests/Containers/ArrayOperationsTest.cpp
index 92c24428b74513d8bda61db2fe5095a7d4e4ff8a..c499a61b2cb1b50eebf9fc7fedacf56b9c7cb68a 100644
--- a/src/UnitTests/Containers/ArrayOperationsTest.cpp
+++ b/src/UnitTests/Containers/ArrayOperationsTest.cpp
@@ -9,5 +9,3 @@
/* See Copyright Notice in tnl/Copyright */
#include "ArrayOperationsTest.h"
-
-
diff --git a/src/UnitTests/Containers/ArrayOperationsTest.h b/src/UnitTests/Containers/ArrayOperationsTest.h
index d11489d4d561abaf77acadcbe31870cb5ce63a6f..be75e04fe737e6cb4c4ac032af3656c48623211e 100644
--- a/src/UnitTests/Containers/ArrayOperationsTest.h
+++ b/src/UnitTests/Containers/ArrayOperationsTest.h
@@ -1,5 +1,5 @@
/***************************************************************************
- ArrayOperationsTester.h - description
+ ArrayOperationsTest.h - description
-------------------
begin : Jul 15, 2013
copyright : (C) 2013 by Tomas Oberhuber
@@ -10,304 +10,310 @@
#pragma once
+#ifdef HAVE_GTEST
#include <TNL/Containers/Algorithms/ArrayOperations.h>
#include <TNL/Devices/Cuda.h>
-#ifdef HAVE_GTEST
#include "gtest/gtest.h"
-#endif
using namespace TNL;
using namespace TNL::Containers;
using namespace TNL::Containers::Algorithms;
-int getTestSize()
+constexpr int ARRAY_TEST_SIZE = 5000;
+
+// test fixture for typed tests
+template< typename Element >
+class ArrayOperationsTest : public ::testing::Test
{
- return 1 << 16;
- //const int cudaGridSize = 256;
- //return 1.5 * cudaGridSize * maxCudaBlockSize;
- //return 1 << 22;
+protected:
+ using ElementType = Element;
};
-typedef int Element;
+// types for which ArrayTest is instantiated
+using ElementTypes = ::testing::Types< short int, int, long, float, double >;
-#ifdef HAVE_GTEST
+TYPED_TEST_CASE( ArrayOperationsTest, ElementTypes );
-TEST( ArrayOperationsTest, allocationTest )
+TYPED_TEST( ArrayOperationsTest, allocateMemory_host )
{
- Element* data;
- ArrayOperations< Devices::Host >::allocateMemory( data, getTestSize() );
- ASSERT_EQ( data, ( Element* ) NULL );
+ using ElementType = typename TestFixture::ElementType;
+
+ ElementType* data;
+ ArrayOperations< Devices::Host >::allocateMemory( data, ARRAY_TEST_SIZE );
+ ASSERT_NE( data, nullptr );
ArrayOperations< Devices::Host >::freeMemory( data );
-};
+}
-TEST( ArrayOperationsTest, memorySetTest )
+TYPED_TEST( ArrayOperationsTest, setMemoryElement_host )
{
- const int size = 1024;
- Element *data;
- ArrayOperations< Devices::Host > :: allocateMemory( data, size );
- ArrayOperations< Devices::Host > :: setMemory( data, 13, size );
- for( int i = 0; i < size; i ++ )
- ASSERT_EQ( data[ i ], 13 );
- ArrayOperations< Devices::Host > :: freeMemory( data );
-};
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
+
+ ElementType *data;
+ ArrayOperations< Devices::Host >::allocateMemory( data, size );
+ for( int i = 0; i < size; i++ ) {
+ ArrayOperations< Devices::Host >::setMemoryElement( data + i, (ElementType) i );
+ EXPECT_EQ( data[ i ], i );
+ EXPECT_EQ( ArrayOperations< Devices::Host >::getMemoryElement( data + i ), i );
+ }
+ ArrayOperations< Devices::Host >::freeMemory( data );
+}
-TEST( ArrayOperationsTest, copyMemoryTest )
+TYPED_TEST( ArrayOperationsTest, setMemory_host )
{
- const int size = getTestSize();
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
- Element *data1, *data2;
- ArrayOperations< Devices::Host > :: allocateMemory( data1, size );
- ArrayOperations< Devices::Host > :: allocateMemory( data2, size );
- ArrayOperations< Devices::Host > :: setMemory( data1, 13, size );
- ArrayOperations< Devices::Host > :: copyMemory< Element, Element, int >( data2, data1, size );
+ ElementType *data;
+ ArrayOperations< Devices::Host >::allocateMemory( data, size );
+ ArrayOperations< Devices::Host >::setMemory( data, (ElementType) 13, size );
for( int i = 0; i < size; i ++ )
- ASSERT_EQ( data1[ i ], data2[ i ]);
- ArrayOperations< Devices::Host > :: freeMemory( data1 );
- ArrayOperations< Devices::Host > :: freeMemory( data2 );
-};
+ EXPECT_EQ( data[ i ], 13 );
+ ArrayOperations< Devices::Host >::freeMemory( data );
+}
+
+TYPED_TEST( ArrayOperationsTest, copyMemory_host )
+{
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
+
+ ElementType *data1, *data2;
+ ArrayOperations< Devices::Host >::allocateMemory( data1, size );
+ ArrayOperations< Devices::Host >::allocateMemory( data2, size );
+ ArrayOperations< Devices::Host >::setMemory( data1, (ElementType) 13, size );
+ ArrayOperations< Devices::Host >::copyMemory< ElementType, ElementType >( data2, data1, size );
+ for( int i = 0; i < size; i ++ )
+ EXPECT_EQ( data1[ i ], data2[ i ]);
+ ArrayOperations< Devices::Host >::freeMemory( data1 );
+ ArrayOperations< Devices::Host >::freeMemory( data2 );
+}
-TEST( ArrayOperationsTest, copyMemoryWithConversionTest )
+TYPED_TEST( ArrayOperationsTest, copyMemoryWithConversion_host )
{
- const int size = getTestSize();
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
+
int *data1;
float *data2;
- ArrayOperations< Devices::Host > :: allocateMemory( data1, size );
- ArrayOperations< Devices::Host > :: allocateMemory( data2, size );
- ArrayOperations< Devices::Host > :: setMemory( data1, 13, size );
- ArrayOperations< Devices::Host > :: copyMemory< float, int, int >( data2, data1, size );
+ ArrayOperations< Devices::Host >::allocateMemory( data1, size );
+ ArrayOperations< Devices::Host >::allocateMemory( data2, size );
+ ArrayOperations< Devices::Host >::setMemory( data1, 13, size );
+ ArrayOperations< Devices::Host >::copyMemory< float, int >( data2, data1, size );
for( int i = 0; i < size; i ++ )
- ASSERT_EQ( data1[ i ], data2[ i ] );
- ArrayOperations< Devices::Host > :: freeMemory( data1 );
- ArrayOperations< Devices::Host > :: freeMemory( data2 );
-};
+ EXPECT_EQ( data1[ i ], data2[ i ] );
+ ArrayOperations< Devices::Host >::freeMemory( data1 );
+ ArrayOperations< Devices::Host >::freeMemory( data2 );
+}
-TEST( ArrayOperationsTest, compareMemoryTest )
+TYPED_TEST( ArrayOperationsTest, compareMemory_host )
{
- const int size = getTestSize();
- int *data1, *data2;
- ArrayOperations< Devices::Host > :: allocateMemory( data1, size );
- ArrayOperations< Devices::Host > :: allocateMemory( data2, size );
- ArrayOperations< Devices::Host > :: setMemory( data1, 7, size );
- ASSERT_FALSE( ( ArrayOperations< Devices::Host > :: compareMemory< int, int, int >( data1, data2, size ) ) );
- ArrayOperations< Devices::Host > :: setMemory( data2, 7, size );
- ASSERT_TRUE( ( ArrayOperations< Devices::Host > :: compareMemory< int, int, int >( data1, data2, size ) ) );
-};
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
+
+ ElementType *data1, *data2;
+ ArrayOperations< Devices::Host >::allocateMemory( data1, size );
+ ArrayOperations< Devices::Host >::allocateMemory( data2, size );
+ ArrayOperations< Devices::Host >::setMemory( data1, (ElementType) 7, size );
+ ArrayOperations< Devices::Host >::setMemory( data2, (ElementType) 0, size );
+ EXPECT_FALSE( ( ArrayOperations< Devices::Host >::compareMemory< ElementType, ElementType >( data1, data2, size ) ) );
+ ArrayOperations< Devices::Host >::setMemory( data2, (ElementType) 7, size );
+ EXPECT_TRUE( ( ArrayOperations< Devices::Host >::compareMemory< ElementType, ElementType >( data1, data2, size ) ) );
+ ArrayOperations< Devices::Host >::freeMemory( data1 );
+ ArrayOperations< Devices::Host >::freeMemory( data2 );
+}
-TEST( ArrayOperationsTest, compareMemoryWithConversionTest )
+TYPED_TEST( ArrayOperationsTest, compareMemoryWithConversion_host )
{
- const int size = getTestSize();
+ const int size = ARRAY_TEST_SIZE;
+
int *data1;
float *data2;
- ArrayOperations< Devices::Host > :: allocateMemory( data1, size );
- ArrayOperations< Devices::Host > :: allocateMemory( data2, size );
- ArrayOperations< Devices::Host > :: setMemory( data1, 7, size );
- ASSERT_FALSE( ( ArrayOperations< Devices::Host > :: compareMemory< int, float, int >( data1, data2, size ) ) );
- ArrayOperations< Devices::Host > :: setMemory( data2, ( float ) 7.0, size );
- ASSERT_TRUE( ( ArrayOperations< Devices::Host > :: compareMemory< int, float, int >( data1, data2, size ) ) );
-};
+ ArrayOperations< Devices::Host >::allocateMemory( data1, size );
+ ArrayOperations< Devices::Host >::allocateMemory( data2, size );
+ ArrayOperations< Devices::Host >::setMemory( data1, 7, size );
+ ArrayOperations< Devices::Host >::setMemory( data2, (float) 0.0, size );
+ EXPECT_FALSE( ( ArrayOperations< Devices::Host >::compareMemory< int, float >( data1, data2, size ) ) );
+ ArrayOperations< Devices::Host >::setMemory( data2, (float) 7.0, size );
+ EXPECT_TRUE( ( ArrayOperations< Devices::Host >::compareMemory< int, float >( data1, data2, size ) ) );
+ ArrayOperations< Devices::Host >::freeMemory( data1 );
+ ArrayOperations< Devices::Host >::freeMemory( data2 );
+}
#ifdef HAVE_CUDA
-TEST( ArrayOperationsTest, allocationTest )
+TYPED_TEST( ArrayOperationsTest, allocateMemory_cuda )
{
- int* data;
- ArrayOperations< Devices::Cuda >::allocateMemory( data, getTestSize() );
- ASSERT_TRUE( checkCudaDevice );
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
- ArrayOperations< Devices::Cuda >::freeMemory( data );
- ASSERT_TRUE( checkCudaDevice );
-}
-
-TEST( ArrayOperationsTest, setMemoryElementTest )
-{
- const int size( 1024 );
- int* data;
+ ElementType* data;
ArrayOperations< Devices::Cuda >::allocateMemory( data, size );
- ASSERT_TRUE( checkCudaDevice );
-
- for( int i = 0; i < getTestSize(); i++ )
- ArrayOperations< Devices::Cuda >::setMemoryElement( &data[ i ], i );
-
- for( int i = 0; i < size; i++ )
- {
- int d;
- ASSERT_EQ( cudaMemcpy( &d, &data[ i ], sizeof( int ), cudaMemcpyDeviceToHost ), cudaSuccess );
- ASSERT_EQ( d, i );
- }
+ ASSERT_TRUE( TNL_CHECK_CUDA_DEVICE );
+ ASSERT_NE( data, nullptr );
ArrayOperations< Devices::Cuda >::freeMemory( data );
- ASSERT_TRUE( checkCudaDevice );
+ ASSERT_TRUE( TNL_CHECK_CUDA_DEVICE );
}
-TEST( ArrayOperationsTest, getMemoryElementTest )
+TYPED_TEST( ArrayOperationsTest, setMemoryElement_cuda )
{
- const int size( 1024 );
- int* data;
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
+
+ ElementType* data;
ArrayOperations< Devices::Cuda >::allocateMemory( data, size );
- ASSERT_TRUE( checkCudaDevice );
+ ASSERT_TRUE( TNL_CHECK_CUDA_DEVICE );
- for( int i = 0; i < getTestSize(); i++ )
- ArrayOperations< Devices::Cuda >::setMemoryElement( &data[ i ], i );
+ for( int i = 0; i < size; i++ )
+ ArrayOperations< Devices::Cuda >::setMemoryElement( &data[ i ], (ElementType) i );
for( int i = 0; i < size; i++ )
- ASSERT_EQ( ( ArrayOperations< Devices::Cuda >::getMemoryElement( &data[ i ] ), i ) );
+ {
+ ElementType d;
+ ASSERT_EQ( cudaMemcpy( &d, &data[ i ], sizeof( ElementType ), cudaMemcpyDeviceToHost ), cudaSuccess );
+ EXPECT_EQ( d, i );
+ EXPECT_EQ( ArrayOperations< Devices::Cuda >::getMemoryElement( &data[ i ] ), i );
+ }
ArrayOperations< Devices::Cuda >::freeMemory( data );
- ASSERT_TRUE( checkCudaDevice );
+ ASSERT_TRUE( TNL_CHECK_CUDA_DEVICE );
}
-
-TEST( ArrayOperationsTest, smallMemorySetTest )
+TYPED_TEST( ArrayOperationsTest, setMemory_cuda )
{
- const int size = 1024;
- int *hostData, *deviceData;
- ArrayOperations< Devices::Host >::allocateMemory( hostData, size );
- ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
- ArrayOperations< Devices::Host >::setMemory( hostData, 0, size );
- ArrayOperations< Devices::Cuda >::setMemory( deviceData, 13, size );
- ASSERT_TRUE( checkCudaDevice );
- ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< int, int >( hostData, deviceData, size );
- ASSERT_TRUE( checkCudaDevice );
- for( int i = 0; i < size; i ++ )
- ASSERT_EQ( hostData[ i ], 13 );
- ArrayOperations< Devices::Cuda >::freeMemory( hostData );
- ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
-};
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
-TEST( ArrayOperationsTest, bigMemorySetTest )
-{
- const int size( getTestSize() );
- int *hostData, *deviceData;
+ ElementType *hostData, *deviceData;
ArrayOperations< Devices::Host >::allocateMemory( hostData, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
- ArrayOperations< Devices::Host >::setMemory( hostData, 0, size );
- ArrayOperations< Devices::Cuda >::setMemory( deviceData, 13, size );
- ASSERT_TRUE( checkCudaDevice );
- ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< int, int >( hostData, deviceData, size );
- ASSERT_TRUE( checkCudaDevice );
- for( int i = 0; i < size; i += 100 )
- {
- if( hostData[ i ] != 13 )
- ASSERT_EQ( hostData[ i ], 13 );
- }
+ ArrayOperations< Devices::Host >::setMemory( hostData, (ElementType) 0, size );
+ ArrayOperations< Devices::Cuda >::setMemory( deviceData, (ElementType) 13, size );
+ ASSERT_TRUE( TNL_CHECK_CUDA_DEVICE );
+ ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< ElementType, ElementType >( hostData, deviceData, size );
+ ASSERT_TRUE( TNL_CHECK_CUDA_DEVICE );
+ for( int i = 0; i < size; i++ )
+ EXPECT_EQ( hostData[ i ], 13 );
ArrayOperations< Devices::Host >::freeMemory( hostData );
ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
-};
+}
-TEST( ArrayOperationsTest, copyMemoryTest )
+TYPED_TEST( ArrayOperationsTest, copyMemory_cuda )
{
- const int size = getTestSize();
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
- int *hostData1, *hostData2, *deviceData;
- ArrayOperations< Devices::Host >::allocateMemory( hostData1, size );
+ ElementType *hostData, *hostData2, *deviceData, *deviceData2;
+ ArrayOperations< Devices::Host >::allocateMemory( hostData, size );
ArrayOperations< Devices::Host >::allocateMemory( hostData2, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
- ArrayOperations< Devices::Host >::setMemory( hostData1, 13, size );
- ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< int, int >( deviceData, hostData1, size );
- ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< int, int >( hostData2, deviceData, size );
- ASSERT_TRUE( ( ArrayOperations< Devices::Host >::compareMemory< int, int >( hostData1, hostData2, size) ) );
- ArrayOperations< Devices::Host >::freeMemory( hostData1 );
+ ArrayOperations< Devices::Cuda >::allocateMemory( deviceData2, size );
+ ArrayOperations< Devices::Host >::setMemory( hostData, (ElementType) 13, size );
+ ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< ElementType >( deviceData, hostData, size );
+ ArrayOperations< Devices::Cuda >::copyMemory< ElementType, ElementType >( deviceData2, deviceData, size );
+ ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< ElementType, ElementType >( hostData2, deviceData2, size );
+ EXPECT_TRUE( ( ArrayOperations< Devices::Host >::compareMemory< ElementType, ElementType >( hostData, hostData2, size) ) );
+ ArrayOperations< Devices::Host >::freeMemory( hostData );
ArrayOperations< Devices::Host >::freeMemory( hostData2 );
ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
-};
+ ArrayOperations< Devices::Cuda >::freeMemory( deviceData2 );
+}
-TEST( ArrayOperationsTest, copyMemoryWithConversionHostToCudaTest )
+TYPED_TEST( ArrayOperationsTest, copyMemoryWithConversions_cuda )
{
- const int size = getTestSize();
- int *hostData1;
- float *hostData2, *deviceData;
- ArrayOperations< Devices::Host >::allocateMemory( hostData1, size );
- ArrayOperations< Devices::Host >::allocateMemory( hostData2, size );
- ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
- ArrayOperations< Devices::Host >::setMemory( hostData1, 13, size );
- ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< float, int, int >( deviceData, hostData1, size );
- ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< float, float, int >( hostData2, deviceData, size );
- for( int i = 0; i < size; i ++ )
- ASSERT_EQ( hostData1[ i ], hostData2[ i ] );
- ArrayOperations< Devices::Host >::freeMemory( hostData1 );
- ArrayOperations< Devices::Host >::freeMemory( hostData2 );
- ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
-};
+ const int size = ARRAY_TEST_SIZE;
-TEST( ArrayOperationsTest, copyMemoryWithConversionCudaToHostTest )
-{
- const int size = getTestSize();
- int *hostData1, *deviceData;
- float *hostData2;
- ArrayOperations< Devices::Host >::allocateMemory( hostData1, size );
+ int *hostData;
+ double *hostData2;
+ long *deviceData;
+ float *deviceData2;
+ ArrayOperations< Devices::Host >::allocateMemory( hostData, size );
ArrayOperations< Devices::Host >::allocateMemory( hostData2, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
- ArrayOperations< Devices::Host >::setMemory( hostData1, 13, size );
- ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< int, int >( deviceData, hostData1, size );
- ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< float, int, int >( hostData2, deviceData, size );
- for( int i = 0; i < size; i ++ )
- ASSERT_EQ( hostData1[ i ], hostData2[ i ] );
- ArrayOperations< Devices::Host >::freeMemory( hostData1 );
- ArrayOperations< Devices::Host >::freeMemory( hostData2 );
- ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
-};
-
-TEST( ArrayOperationsTest, copyMemoryWithConversionCudaToCudaTest )
-{
- const int size = getTestSize();
- int *hostData1, *deviceData1;
- float *hostData2, *deviceData2;
- ArrayOperations< Devices::Host >::allocateMemory( hostData1, size );
- ArrayOperations< Devices::Host >::allocateMemory( hostData2, size );
- ArrayOperations< Devices::Cuda >::allocateMemory( deviceData1, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData2, size );
- ArrayOperations< Devices::Host >::setMemory( hostData1, 13, size );
- ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< int, int, int >( deviceData1, hostData1, size );
- ArrayOperations< Devices::Cuda >::copyMemory< float, int, int >( deviceData2, deviceData1, size );
- ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< float, float, int >( hostData2, deviceData2, size );
+ ArrayOperations< Devices::Host >::setMemory( hostData, 13, size );
+ ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< long, int >( deviceData, hostData, size );
+ ArrayOperations< Devices::Cuda >::copyMemory< float, long >( deviceData2, deviceData, size );
+ ArrayOperations< Devices::Host, Devices::Cuda >::copyMemory< double, float >( hostData2, deviceData2, size );
for( int i = 0; i < size; i ++ )
- ASSERT_EQ( hostData1[ i ], hostData2[ i ] );
- ArrayOperations< Devices::Host >::freeMemory( hostData1 );
+ EXPECT_EQ( hostData[ i ], hostData2[ i ] );
+ ArrayOperations< Devices::Host >::freeMemory( hostData );
ArrayOperations< Devices::Host >::freeMemory( hostData2 );
- ArrayOperations< Devices::Cuda >::freeMemory( deviceData1 );
+ ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
ArrayOperations< Devices::Cuda >::freeMemory( deviceData2 );
-};
+}
-TEST( ArrayOperationsTest, compareMemoryHostCudaTest )
+TYPED_TEST( ArrayOperationsTest, compareMemory_cuda )
{
- const int size = getTestSize();
- int *hostData, *deviceData;
+ using ElementType = typename TestFixture::ElementType;
+ const int size = ARRAY_TEST_SIZE;
+
+ ElementType *hostData, *deviceData, *deviceData2;
ArrayOperations< Devices::Host >::allocateMemory( hostData, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
- ArrayOperations< Devices::Host >::setMemory( hostData, 7, size );
- ArrayOperations< Devices::Cuda >::setMemory( deviceData, 8, size );
- ASSERT_FALSE( ( ArrayOperations< Devices::Host, Devices::Cuda >::compareMemory< int, int, int >( hostData, deviceData, size ) ) );
- ArrayOperations< Devices::Cuda >::setMemory( deviceData, 7, size );
- ASSERT_TRUE( ( ArrayOperations< Devices::Host, Devices::Cuda >::compareMemory< int, int, int >( hostData, deviceData, size ) ) );
-};
+ ArrayOperations< Devices::Cuda >::allocateMemory( deviceData2, size );
+
+ ArrayOperations< Devices::Host >::setMemory( hostData, (ElementType) 7, size );
+ ArrayOperations< Devices::Cuda >::setMemory( deviceData, (ElementType) 8, size );
+ ArrayOperations< Devices::Cuda >::setMemory( deviceData2, (ElementType) 9, size );
+ EXPECT_FALSE(( ArrayOperations< Devices::Host, Devices::Cuda >::compareMemory< ElementType, ElementType >( hostData, deviceData, size ) ));
+ EXPECT_FALSE(( ArrayOperations< Devices::Cuda, Devices::Host >::compareMemory< ElementType, ElementType >( deviceData, hostData, size ) ));
+ EXPECT_FALSE(( ArrayOperations< Devices::Cuda >::compareMemory< ElementType, ElementType >( deviceData, deviceData2, size ) ));
+
+ ArrayOperations< Devices::Cuda >::setMemory( deviceData, (ElementType) 7, size );
+ ArrayOperations< Devices::Cuda >::setMemory( deviceData2, (ElementType) 7, size );
+ EXPECT_TRUE(( ArrayOperations< Devices::Host, Devices::Cuda >::compareMemory< ElementType, ElementType >( hostData, deviceData, size ) ));
+ EXPECT_TRUE(( ArrayOperations< Devices::Cuda, Devices::Host >::compareMemory< ElementType, ElementType >( deviceData, hostData, size ) ));
+ EXPECT_TRUE(( ArrayOperations< Devices::Cuda >::compareMemory< ElementType, ElementType >( deviceData, deviceData2, size ) ));
-TEST( ArrayOperationsTest, compareMemoryWithConversionHostCudaTest )
+ ArrayOperations< Devices::Host >::freeMemory( hostData );
+ ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
+ ArrayOperations< Devices::Cuda >::freeMemory( deviceData2 );
+}
+
+TYPED_TEST( ArrayOperationsTest, compareMemoryWithConversions_cuda )
{
- const int size = getTestSize();
+ const int size = ARRAY_TEST_SIZE;
+
int *hostData;
float *deviceData;
+ double *deviceData2;
ArrayOperations< Devices::Host >::allocateMemory( hostData, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
+ ArrayOperations< Devices::Cuda >::allocateMemory( deviceData2, size );
+
ArrayOperations< Devices::Host >::setMemory( hostData, 7, size );
- ArrayOperations< Devices::Cuda >::setMemory( deviceData, ( float ) 8.0, size );
- ASSERT_FALSE( ( ArrayOperations< Devices::Host, Devices::Cuda >::compareMemory< int, float, int >( hostData, deviceData, size ) ) );
- ArrayOperations< Devices::Cuda >::setMemory( deviceData, ( float ) 7.0, size );
- ASSERT_TRUE( ( ArrayOperations< Devices::Host, Devices::Cuda >::compareMemory< int, float, int >( hostData, deviceData, size ) ) );
-};
+ ArrayOperations< Devices::Cuda >::setMemory( deviceData, (float) 8, size );
+ ArrayOperations< Devices::Cuda >::setMemory( deviceData2, (double) 9, size );
+ EXPECT_FALSE(( ArrayOperations< Devices::Host, Devices::Cuda >::compareMemory< int, float >( hostData, deviceData, size ) ));
+ EXPECT_FALSE(( ArrayOperations< Devices::Cuda, Devices::Host >::compareMemory< float, int >( deviceData, hostData, size ) ));
+ // TODO: missing implementation of relevant reduction operation on CUDA with different types
+// EXPECT_FALSE(( ArrayOperations< Devices::Cuda >::compareMemory< float, double >( deviceData, deviceData2, size ) ));
+
+ ArrayOperations< Devices::Cuda >::setMemory( deviceData, (float) 7, size );
+ ArrayOperations< Devices::Cuda >::setMemory( deviceData2, (double) 7, size );
+ EXPECT_TRUE(( ArrayOperations< Devices::Host, Devices::Cuda >::compareMemory< int, float >( hostData, deviceData, size ) ));
+ EXPECT_TRUE(( ArrayOperations< Devices::Cuda, Devices::Host >::compareMemory< float, int >( deviceData, hostData, size ) ));
+ // TODO: missing implementation of relevant reduction operation on CUDA with different types
+// EXPECT_TRUE(( ArrayOperations< Devices::Cuda >::compareMemory< float, double >( deviceData, deviceData2, size ) ));
+
+ ArrayOperations< Devices::Host >::freeMemory( hostData );
+ ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
+ ArrayOperations< Devices::Cuda >::freeMemory( deviceData2 );
+}
#endif // HAVE_CUDA
#endif // HAVE_GTEST
+
+#include "../GtestMissingError.h"
int main( int argc, char* argv[] )
{
#ifdef HAVE_GTEST
::testing::InitGoogleTest( &argc, argv );
return RUN_ALL_TESTS();
#else
- return EXIT_FAILURE;
+ throw GtestMissingError();
#endif
}
-
-
-
diff --git a/src/UnitTests/Containers/ArrayTest.h b/src/UnitTests/Containers/ArrayTest.h
index a930efdbb5e5c15b8fb317c21879864d4e5b1d12..7b9bc2f5659644199c91c3a291857cc5e18e2564 100644
--- a/src/UnitTests/Containers/ArrayTest.h
+++ b/src/UnitTests/Containers/ArrayTest.h
@@ -1,5 +1,5 @@
/***************************************************************************
- ArrayTester.h - description
+ ArrayTest.h - description
-------------------
begin : Jul 4, 2012
copyright : (C) 2012 by Tomas Oberhuber
@@ -10,255 +10,522 @@
#pragma once
+#ifdef HAVE_GTEST
+#include <type_traits>
+
#include <TNL/Containers/Array.h>
-#include <TNL/Devices/Host.h>
-#ifdef HAVE_GTEST
#include "gtest/gtest.h"
-#endif
using namespace TNL;
using namespace TNL::Containers;
-#ifdef HAVE_CUDA
-template< typename ElementType, typename IndexType >
-__global__ void testSetGetElementKernel( Array< ElementType, Devices::Cuda, IndexType >* u );
-#endif
-
-class testingClassForArrayTester
+// minimal custom data structure usable as ElementType in Array
+struct MyData
{
- public:
+ double data;
+
+ __cuda_callable__
+ MyData() : data(0) {}
+
+ template< typename T >
+ __cuda_callable__
+ MyData( T v ) : data(v) {}
- static String getType()
- {
- return String( "testingClassForArrayTester" );
- };
+ bool operator==( const MyData& v ) const { return data == v.data; }
+
+ // operator used in tests, not necessary for Array to work
+ template< typename T >
+ bool operator==( T v ) const { return data == v; }
+
+ static String getType()
+ {
+ return String( "MyData" );
+ }
};
-String getType( const testingClassForArrayTester& c )
+std::ostream& operator<<( std::ostream& str, const MyData& v )
{
- return String( "testingClassForArrayTester" );
-};
+ return str << v.data;
+}
-#ifdef HAVE_GTEST
+// test fixture for typed tests
+template< typename Array >
+class ArrayTest : public ::testing::Test
+{
+protected:
+ using ArrayType = Array;
+};
-// TODO: Fix this
-#if GTEST_HAS_TYPED_TEST
+// types for which ArrayTest is instantiated
+using ArrayTypes = ::testing::Types<
+ Array< short, Devices::Host, short >,
+ Array< int, Devices::Host, short >,
+ Array< long, Devices::Host, short >,
+ Array< float, Devices::Host, short >,
+ Array< double, Devices::Host, short >,
+ Array< MyData, Devices::Host, short >,
+ Array< short, Devices::Host, int >,
+ Array< int, Devices::Host, int >,
+ Array< long, Devices::Host, int >,
+ Array< float, Devices::Host, int >,
+ Array< double, Devices::Host, int >,
+ Array< MyData, Devices::Host, int >,
+ Array< short, Devices::Host, long >,
+ Array< int, Devices::Host, long >,
+ Array< long, Devices::Host, long >,
+ Array< float, Devices::Host, long >,
+ Array< double, Devices::Host, long >,
+ Array< MyData, Devices::Host, long >
+ // FIXME: this segfaults in String::~String()
+// , Array< String, Devices::Host, long >
+#ifdef HAVE_CUDA
+ ,
+ Array< short, Devices::Cuda, short >,
+ Array< int, Devices::Cuda, short >,
+ Array< long, Devices::Cuda, short >,
+ Array< float, Devices::Cuda, short >,
+ Array< double, Devices::Cuda, short >,
+ Array< MyData, Devices::Cuda, short >,
+ Array< short, Devices::Cuda, int >,
+ Array< int, Devices::Cuda, int >,
+ Array< long, Devices::Cuda, int >,
+ Array< float, Devices::Cuda, int >,
+ Array< double, Devices::Cuda, int >,
+ Array< MyData, Devices::Cuda, int >,
+ Array< short, Devices::Cuda, long >,
+ Array< int, Devices::Cuda, long >,
+ Array< long, Devices::Cuda, long >,
+ Array< float, Devices::Cuda, long >,
+ Array< double, Devices::Cuda, long >,
+ Array< MyData, Devices::Cuda, long >
+#endif
+#ifdef HAVE_MIC
+ ,
+ Array< short, Devices::MIC, short >,
+ Array< int, Devices::MIC, short >,
+ Array< long, Devices::MIC, short >,
+ Array< float, Devices::MIC, short >,
+ Array< double, Devices::MIC, short >,
+ // TODO: MyData does not work on MIC
+// Array< MyData, Devices::MIC, short >,
+ Array< short, Devices::MIC, int >,
+ Array< int, Devices::MIC, int >,
+ Array< long, Devices::MIC, int >,
+ Array< float, Devices::MIC, int >,
+ Array< double, Devices::MIC, int >,
+ // TODO: MyData does not work on MIC
+// Array< MyData, Devices::MIC, int >,
+ Array< short, Devices::MIC, long >,
+ Array< int, Devices::MIC, long >,
+ Array< long, Devices::MIC, long >,
+ Array< float, Devices::MIC, long >,
+ Array< double, Devices::MIC, long >
+ // TODO: MyData does not work on MIC
+// Array< MyData, Devices::MIC, long >
+#endif
+>;
-using ::testing::Types;
-typedef Types< int, Devices::Host, int > MyTypes;
-/*TYPED_TEST_CASE( ArrayTest, MyTypes );
+TYPED_TEST_CASE( ArrayTest, ArrayTypes );
-TYPED_TEST( ArrayTest, testConstructorDestructor )
+
+TYPED_TEST( ArrayTest, constructors )
{
- typedef Array< TypeParam > ArrayType;
+ using ArrayType = typename TestFixture::ArrayType;
+
ArrayType u;
+ EXPECT_EQ( u.getSize(), 0 );
+
ArrayType v( 10 );
- ASSERT_EQ( v.getSize(), 10 );
+ EXPECT_EQ( v.getSize(), 10 );
+
+ if( std::is_same< typename ArrayType::DeviceType, Devices::Host >::value ) {
+ typename ArrayType::ElementType data[ 10 ];
+ ArrayType w( data, 10 );
+ EXPECT_EQ( w.getData(), data );
+
+ ArrayType z1( w );
+ EXPECT_EQ( z1.getData(), data );
+ EXPECT_EQ( z1.getSize(), 10 );
+
+ ArrayType z2( w, 1 );
+ EXPECT_EQ( z2.getData(), data + 1 );
+ EXPECT_EQ( z2.getSize(), 9 );
+
+ ArrayType z3( w, 2, 3 );
+ EXPECT_EQ( z3.getData(), data + 2 );
+ EXPECT_EQ( z3.getSize(), 3 );
+ }
}
-TYPED_TEST( ArrayTest, testSetSize )
+TYPED_TEST( ArrayTest, setSize )
{
- typedef Array< TypeParam > ArrayType;
- ArrayType u, v;
- u.setSize( 10 );
- v.setSize( 10 );
- ASSERT_EQ( u.getSize(), 10 );
- ASSERT_EQ( v.getSize(), 10 );
+ using ArrayType = typename TestFixture::ArrayType;
+
+ ArrayType u;
+ const int maxSize = 10;
+ for( int i = 0; i <= maxSize; i ++ ) {
+ u.setSize( i );
+ EXPECT_EQ( u.getSize(), i );
+ }
+
+ ArrayType v( u );
+ EXPECT_EQ( v.getSize(), 10 );
+ EXPECT_EQ( v.getData(), u.getData() );
+ v.setSize( 11 );
+ EXPECT_EQ( u.getSize(), 10 );
+ EXPECT_EQ( v.getSize(), 11 );
+ EXPECT_NE( v.getData(), u.getData() );
+
+ // cast to bool returns true iff size > 0
+ EXPECT_TRUE( (bool) u );
+ EXPECT_FALSE( ! u );
+ u.setSize( 0 );
+ EXPECT_FALSE( (bool) u );
+ EXPECT_TRUE( ! u );
}
-TYPED_TEST( ArrayTest, testBind )
+TYPED_TEST( ArrayTest, setLike )
{
- typedef Array< TypeParam > ArrayType;
+ using ArrayType = typename TestFixture::ArrayType;
+
+ ArrayType u( 10 );
+ EXPECT_EQ( u.getSize(), 10 );
+
+ ArrayType v;
+ v.setLike( u );
+ EXPECT_EQ( v.getSize(), u.getSize() );
+ EXPECT_NE( v.getData(), u.getData() );
+}
+
+TYPED_TEST( ArrayTest, bind )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+
ArrayType u( 10 ), v;
- u.setValue( 27 );
v.bind( u );
- ASSERT_EQ( v.getSize(), u.getSize() );
- ASSERT_EQ( u.getElement( 0 ), 27 );
- v.setValue( 50 );
- ASSERT_EQ( u.getElement( 0 ), 50 );
- u.reset();
- ASSERT_EQ( u.getSize(), 0 );
- ASSERT_EQ( v.getElement( 0 ), 50 );
+ EXPECT_EQ( v.getSize(), u.getSize() );
+ EXPECT_EQ( v.getData(), u.getData() );
- ElementType data[ 10 ] = { 1, 2, 3, 4, 5, 6, 7, 8, 10 };
- u.bind( data, 10 );
- ASSERT_EQ( u.getElement( 1 ), 2 );
- v.bind( u );
- ASSERT_EQ( v.getElement( 1 ), 2 );
+ // bind array with offset and size
+ ArrayType w;
+ w.bind( u, 2, 3 );
+ EXPECT_EQ( w.getSize(), 3 );
+ EXPECT_EQ( w.getData(), u.getData() + 2 );
+
+ // setting values
+ u.setValue( 27 );
+ EXPECT_EQ( u.getElement( 0 ), 27 );
+ v.setValue( 50 );
+ EXPECT_EQ( u.getElement( 0 ), 50 );
u.reset();
- v.setElement( 1, 3 );
- v.reset();
- ASSERT_EQ( data[ 1 ], 3 );
-}*/
+ EXPECT_EQ( u.getSize(), 0 );
+ EXPECT_EQ( v.getElement( 0 ), 50 );
+
+ if( std::is_same< typename ArrayType::DeviceType, Devices::Host >::value ) {
+ typename ArrayType::ElementType data[ 10 ] = { 1, 2, 3, 4, 5, 6, 7, 8, 10 };
+ u.bind( data, 10 );
+ EXPECT_EQ( u.getData(), data );
+ EXPECT_EQ( u.getSize(), 10 );
+ EXPECT_EQ( u.getElement( 1 ), 2 );
+ v.bind( u );
+ EXPECT_EQ( v.getElement( 1 ), 2 );
+ u.reset();
+ v.setElement( 1, 3 );
+ v.reset();
+ EXPECT_EQ( data[ 1 ], 3 );
+ }
+}
-#endif /* GTEST_HAS_TYPED_TEST */
+TYPED_TEST( ArrayTest, swap )
+{
+ using ArrayType = typename TestFixture::ArrayType;
-typedef int ElementType;
-typedef Devices::Host Device;
-typedef int IndexType;
+ ArrayType u( 10 ), v( 20 );
+ u.setValue( 0 );
+ v.setValue( 1 );
+ u.swap( v );
+ EXPECT_EQ( u.getSize(), 20 );
+ EXPECT_EQ( v.getSize(), 10 );
+ for( int i = 0; i < 20; i++ )
+ EXPECT_EQ( u.getElement( i ), 1 );
+ for( int i = 0; i < 10; i++ )
+ EXPECT_EQ( v.getElement( i ), 0 );
+}
-TEST( ArrayTest, testSetGetElement )
+TYPED_TEST( ArrayTest, reset )
{
- using namespace TNL::Containers;
- Array< ElementType, Device, IndexType > u;
- u. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- u. setElement( i, i );
- for( int i = 0; i < 10; i ++ )
- ASSERT_EQ( u. getElement( i ), i );
+ using ArrayType = typename TestFixture::ArrayType;
- u.setValue( 0 );
- if( std::is_same< Device, Devices::Host >::value )
- {
- for( int i = 0; i < 10; i ++ )
- u[ i ] = i;
+ ArrayType u;
+ u.setSize( 100 );
+ EXPECT_EQ( u.getSize(), 100 );
+ EXPECT_NE( u.getData(), nullptr );
+ u.reset();
+ EXPECT_EQ( u.getSize(), 0 );
+ EXPECT_EQ( u.getData(), nullptr );
+ u.setSize( 100 );
+ EXPECT_EQ( u.getSize(), 100 );
+ EXPECT_NE( u.getData(), nullptr );
+ u.reset();
+ EXPECT_EQ( u.getSize(), 0 );
+ EXPECT_EQ( u.getData(), nullptr );
+}
+
+template< typename Element, typename Index >
+void testArrayElementwiseAccess( Array< Element, Devices::Host, Index >&& u )
+{
+ u.setSize( 10 );
+ for( int i = 0; i < 10; i++ ) {
+ u.setElement( i, i );
+ EXPECT_EQ( u.getData()[ i ], i );
+ EXPECT_EQ( u.getElement( i ), i );
+ EXPECT_EQ( u[ i ], i );
}
- if( std::is_same< Device, Devices::Cuda >::value )
- {
+}
+
#ifdef HAVE_CUDA
- Array< ElementType, Device, IndexType >* kernel_u =
- Devices::Cuda::passToDevice( u );
- testSetGetElementKernel<<< 1, 16 >>>( kernel_u );
- Devices::Cuda::freeFromDevice( kernel_u );
- ASSERT_TRUE( checkCudaDevice );
-#endif
+template< typename ElementType, typename IndexType >
+__global__ void testSetGetElementKernel( Array< ElementType, Devices::Cuda, IndexType >* u )
+{
+ if( threadIdx.x < ( *u ).getSize() )
+ ( *u )[ threadIdx.x ] = threadIdx.x;
+}
+#endif /* HAVE_CUDA */
+
+template< typename Element, typename Index >
+void testArrayElementwiseAccess( Array< Element, Devices::Cuda, Index >&& u )
+{
+#ifdef HAVE_CUDA
+ u.setSize( 10 );
+ using ArrayType = Array< Element, Devices::Cuda, Index >;
+ ArrayType* kernel_u = Devices::Cuda::passToDevice( u );
+ testSetGetElementKernel<<< 1, 16 >>>( kernel_u );
+ Devices::Cuda::freeFromDevice( kernel_u );
+ TNL_CHECK_CUDA_DEVICE;
+ for( int i = 0; i < 10; i++ ) {
+ EXPECT_EQ( u.getElement( i ), i );
}
- for( int i = 0; i < 10; i++ )
- ASSERT_EQ( u.getElement( i ), i );
-};
+#endif
+}
-TEST( ArrayTest, testComparisonOperator )
+template< typename Element, typename Index >
+void testArrayElementwiseAccess( Array< Element, Devices::MIC, Index >&& u )
{
- using namespace TNL::Containers;
- Array< ElementType, Device, IndexType > u;
- Array< ElementType, Device, IndexType > v;
- Array< ElementType, Device, IndexType > w;
- u. setSize( 10 );
- v. setSize( 10 );
- w. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- {
- u. setElement( i, i );
- v. setElement( i, i );
- w. setElement( i, 2*1 );
+#ifdef HAVE_MIC
+ // TODO
+#endif
+}
+
+TYPED_TEST( ArrayTest, elementwiseAccess )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+
+ testArrayElementwiseAccess( ArrayType() );
+}
+
+// TODO: comparison with different device
+TYPED_TEST( ArrayTest, comparisonOperator )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+
+ ArrayType u( 10 ), v( 10 ), w( 10 );
+ for( int i = 0; i < 10; i ++ ) {
+ u.setElement( i, i );
+ v.setElement( i, i );
+ w.setElement( i, 2 * i );
}
- ASSERT_TRUE( u == v );
- ASSERT_FALSE( u != v );
- ASSERT_TRUE( u != w );
- ASSERT_FALSE( u == w );
-};
+ EXPECT_TRUE( u == u );
+ EXPECT_TRUE( u == v );
+ EXPECT_TRUE( v == u );
+ EXPECT_FALSE( u != v );
+ EXPECT_FALSE( v != u );
+ EXPECT_TRUE( u != w );
+ EXPECT_TRUE( w != u );
+ EXPECT_FALSE( u == w );
+ EXPECT_FALSE( w == u );
+
+ v.setSize( 0 );
+ EXPECT_FALSE( u == v );
+ u.setSize( 0 );
+ EXPECT_TRUE( u == v );
+}
-TEST( ArrayTest, testAssignmentOperator )
+// TODO: comparison with different device
+// TODO: missing implementation of relevant reduction operation on CUDA with different types
+/*
+TYPED_TEST( ArrayTest, comparisonOperatorWithDifferentType )
{
- using namespace TNL::Containers;
- Array< ElementType, Device, IndexType > u;
- Array< ElementType, Device, IndexType > v;
- u. setSize( 10 );
- v. setSize( 10 );
+ Array< short, typename ArrayType::DeviceType, short > z( 10 );
for( int i = 0; i < 10; i ++ )
- u. setElement( i, i );
- v = u;
- ASSERT_TRUE( u == v );
- ASSERT_TRUE( v == u );
- ASSERT_FALSE( u != v );
- ASSERT_FALSE( v != u );
-
- v.setValue( 0 );
- Array< ElementType, Devices::Host, IndexType > w;
- w.setSize( 10 );
- w = u;
+ z.setElement( i, i );
+ EXPECT_TRUE( u == z );
+ EXPECT_FALSE( u != z );
+ for( int i = 0; i < 10; i ++ )
+ z.setElement( i, 2 * i );
+ EXPECT_FALSE( u == z );
+ EXPECT_TRUE( u != z );
+}
+*/
- ASSERT_TRUE( u == w );
- ASSERT_FALSE( u != w );
+// TODO: assignment from different device
+TYPED_TEST( ArrayTest, assignmentOperator )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ ArrayType u( 10 ), v( 10 );
+ for( int i = 0; i < 10; i++ )
+ u.setElement( i, i );
v.setValue( 0 );
- v = w;
- ASSERT_TRUE( v == w );
- ASSERT_FALSE( v != w );
-};
+ v = u;
+ EXPECT_EQ( u, v );
+}
-TEST( ArrayTest, testGetSize )
+// test works only for arithmetic types
+template< typename ArrayType,
+ typename = typename std::enable_if< std::is_arithmetic< typename ArrayType::ElementType >::value >::type >
+void testArrayAssignmentWithDifferentType()
{
- using namespace TNL::Containers;
- Array< ElementType, Device, IndexType > u;
- const int maxSize = 10;
- for( int i = 0; i < maxSize; i ++ )
- u. setSize( i );
+ ArrayType u( 10 );
+ for( int i = 0; i < 10; i++ )
+ u.setElement( i, i );
+ Array< short, typename ArrayType::DeviceType, short > v( 10 );
+ v.setValue( 0 );
+ v = u;
+// TODO: missing implementation of relevant reduction operation on CUDA with different types
+// EXPECT_EQ( u, v );
+ for( int i = 0; i < 10; i++ )
+ EXPECT_EQ( v.getElement( i ), i );
+}
- ASSERT_EQ( u. getSize(), maxSize - 1 );
-};
+template< typename ArrayType,
+ typename = typename std::enable_if< ! std::is_arithmetic< typename ArrayType::ElementType >::value >::type,
+ typename = void >
+void testArrayAssignmentWithDifferentType()
+{
+}
-TEST( ArrayTest, testReset )
+// TODO: assignment from different device
+TYPED_TEST( ArrayTest, assignmentOperatorWithDifferentType )
{
- using namespace TNL::Containers;
- Array< ElementType, Device, IndexType > u;
- u. setSize( 100 );
- ASSERT_EQ( u. getSize(), 100 );
- u. reset();
- ASSERT_EQ( u. getSize(), 0 );
- u. setSize( 100 );
- ASSERT_EQ( u. getSize(), 100 );
- u. reset();
- ASSERT_EQ( u. getSize(), 0 );
+ using ArrayType = typename TestFixture::ArrayType;
-};
+ testArrayAssignmentWithDifferentType< ArrayType >();
+}
-TEST( ArrayTest, testSetSizeAndDestructor )
+TYPED_TEST( ArrayTest, SaveAndLoad )
{
- using namespace TNL::Containers;
+ using ArrayType = typename TestFixture::ArrayType;
+
+ ArrayType u, v;
+ v.setSize( 100 );
for( int i = 0; i < 100; i ++ )
- {
- Array< ElementType, Device, IndexType > u;
- u. setSize( i );
- }
+ v.setElement( i, 3.14147 );
+ File file;
+ file.open( "test-file.tnl", IOMode::write );
+ EXPECT_TRUE( v.save( file ) );
+ file.close();
+ file.open( "test-file.tnl", IOMode::read );
+ EXPECT_TRUE( u.load( file ) );
+ EXPECT_EQ( u, v );
+
+ EXPECT_EQ( std::remove( "test-file.tnl" ), 0 );
}
-TEST( ArrayTest, testSaveAndLoad )
+TYPED_TEST( ArrayTest, boundLoad )
{
- using namespace TNL::Containers;
- Array< ElementType, Device, IndexType > v;
- v. setSize( 100 );
+ using ArrayType = typename TestFixture::ArrayType;
+
+ ArrayType u, v, w;
+ v.setSize( 100 );
for( int i = 0; i < 100; i ++ )
- v. setElement( i, 3.14147 );
+ v.setElement( i, 3.14147 );
File file;
- file. open( "test-file.tnl", tnlWriteMode );
- v. save( file );
- file. close();
- Array< ElementType, Device, IndexType > u;
- file. open( "test-file.tnl", tnlReadMode );
- u. load( file );
- file. close();
- ASSERT_TRUE( u == v );
+ file.open( "test-file.tnl", IOMode::write );
+ EXPECT_TRUE( v.save( file ) );
+ file.close();
+
+ w.setSize( 100 );
+ u.bind( w );
+ file.open( "test-file.tnl", IOMode::read );
+ EXPECT_TRUE( u.boundLoad( file ) );
+ EXPECT_EQ( u, v );
+ EXPECT_EQ( u.getData(), w.getData() );
+
+ u.setSize( 50 );
+ file.open( "test-file.tnl", IOMode::read );
+ EXPECT_FALSE( u.boundLoad( file ) );
+
+ u.reset();
+ file.open( "test-file.tnl", IOMode::read );
+ EXPECT_TRUE( u.boundLoad( file ) );
+
+ EXPECT_EQ( std::remove( "test-file.tnl" ), 0 );
}
-TEST( ArrayTest, testUnusualStructures )
+TYPED_TEST( ArrayTest, referenceCountingConstructors )
{
- using namespace TNL::Containers;
- Array< testingClassForArrayTester >u;
-};
+ using ArrayType = typename TestFixture::ArrayType;
+
+ // copies of a dynamic array
+ ArrayType u( 10 );
+ ArrayType v( u );
+ ArrayType w( v );
+ EXPECT_EQ( v.getData(), u.getData() );
+ EXPECT_EQ( w.getData(), u.getData() );
+
+ // copies of a static array
+ if( std::is_same< typename ArrayType::DeviceType, Devices::Host >::value ) {
+ typename ArrayType::ElementType data[ 10 ] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
+ ArrayType u( data, 10 );
+ ArrayType v( u );
+ ArrayType w( v );
+ EXPECT_EQ( u.getData(), data );
+ EXPECT_EQ( v.getData(), data );
+ EXPECT_EQ( w.getData(), data );
+ }
+}
-#endif /* HAVE_GTEST */
-
-#ifdef HAVE_CUDA
-template< typename ElementType, typename IndexType >
-__global__ void testSetGetElementKernel( Array< ElementType, Devices::Cuda, IndexType >* u )
+TYPED_TEST( ArrayTest, referenceCountingBind )
{
- if( threadIdx.x < ( *u ).getSize() )
- ( *u )[ threadIdx.x ] = threadIdx.x;
+ using ArrayType = typename TestFixture::ArrayType;
+
+ // copies of a dynamic array
+ ArrayType u( 10 );
+ ArrayType v;
+ v.bind( u );
+ ArrayType w;
+ w.bind( v );
+ EXPECT_EQ( v.getData(), u.getData() );
+ EXPECT_EQ( w.getData(), u.getData() );
+
+ // copies of a static array
+ if( std::is_same< typename ArrayType::DeviceType, Devices::Host >::value ) {
+ typename ArrayType::ElementType data[ 10 ] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
+ ArrayType u( data, 10 );
+ ArrayType v;
+ v.bind( u );
+ ArrayType w;
+ w.bind( v );
+ EXPECT_EQ( u.getData(), data );
+ EXPECT_EQ( v.getData(), data );
+ EXPECT_EQ( w.getData(), data );
+ }
}
-#endif /* HAVE_CUDA */
+// TODO: test all __cuda_callable__ methods from a CUDA kernel
+#endif // HAVE_GTEST
+
+
+#include "../GtestMissingError.h"
int main( int argc, char* argv[] )
{
#ifdef HAVE_GTEST
::testing::InitGoogleTest( &argc, argv );
return RUN_ALL_TESTS();
#else
- return EXIT_FAILURE;
+ throw GtestMissingError();
#endif
}
-
-
diff --git a/src/UnitTests/Containers/CMakeLists.txt b/src/UnitTests/Containers/CMakeLists.txt
old mode 100755
new mode 100644
index 3f6b1c4c2d8227734baad4de62a3893c15cbce73..3aa747857a93471dfb06375d746689d1dab9b25e
--- a/src/UnitTests/Containers/CMakeLists.txt
+++ b/src/UnitTests/Containers/CMakeLists.txt
@@ -1,67 +1,83 @@
-IF( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE( ArrayOperationsTest${mpiExt}${debugExt} ArrayOperationsTest.h ArrayOperationsTest.cu )
- SET_TARGET_PROPERTIES( ArrayOperationsTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( ArrayOperationsTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
-ELSE( BUILD_CUDA )
- ADD_EXECUTABLE( ArrayOperationsTest${mpiExt}${debugExt} ArrayOperationsTest.h ArrayOperationsTest.cpp )
- SET_TARGET_PROPERTIES( ArrayOperationsTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( ArrayOperationsTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
-ENDIF( BUILD_CUDA )
+ADD_EXECUTABLE( ListTest${mpiExt}${debugExt} ListTest.cpp )
+TARGET_COMPILE_OPTIONS( ListTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+TARGET_LINK_LIBRARIES( ListTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
IF( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE( ArrayTest${mpiExt}${debugExt} ArrayTest.h ArrayTest.cu )
- SET_TARGET_PROPERTIES( ArrayTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( ArrayTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
-ELSE( BUILD_CUDA )
- ADD_EXECUTABLE( ArrayTest${mpiExt}${debugExt} ArrayTest.h ArrayTest.cpp )
- SET_TARGET_PROPERTIES( ArrayTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( ArrayTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
+ CUDA_ADD_EXECUTABLE( ArrayOperationsTest${mpiExt}${debugExt} ArrayOperationsTest.cu
+ OPTIONS ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( ArrayOperationsTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
+ELSE( BUILD_CUDA )
+ ADD_EXECUTABLE( ArrayOperationsTest${mpiExt}${debugExt} ArrayOperationsTest.cpp )
+ TARGET_COMPILE_OPTIONS( ArrayOperationsTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( ArrayOperationsTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
ENDIF( BUILD_CUDA )
IF( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE( MultiArrayTest${mpiExt}${debugExt} MultiArrayTest.h MultiArrayTest.cu )
- SET_TARGET_PROPERTIES( MultiArrayTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( MultiArrayTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
-ELSE( BUILD_CUDA )
- ADD_EXECUTABLE( StaticArrayTest${mpiExt}${debugExt} StaticArrayTest.h StaticArrayTest.cpp )
- SET_TARGET_PROPERTIES( StaticArrayTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( StaticArrayTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
+ CUDA_ADD_EXECUTABLE( ArrayTest${mpiExt}${debugExt} ArrayTest.cu
+ OPTIONS ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( ArrayTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
+ELSE( BUILD_CUDA )
+ ADD_EXECUTABLE( ArrayTest${mpiExt}${debugExt} ArrayTest.cpp )
+ TARGET_COMPILE_OPTIONS( ArrayTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( ArrayTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
ENDIF( BUILD_CUDA )
-IF( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE( VectorOperationsTest${mpiExt}${debugExt} VectorOperationsTest.h VectorOperationsTest.cu )
- SET_TARGET_PROPERTIES( VectorOperationsTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( VectorOperationsTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
-ELSE( BUILD_CUDA )
- ADD_EXECUTABLE( VectorOperationsTest${mpiExt}${debugExt} VectorOperationsTest.h VectorOperationsTest.cpp )
- SET_TARGET_PROPERTIES( VectorOperationsTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( VectorOperationsTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
-ENDIF( BUILD_CUDA )
+ADD_EXECUTABLE( StaticArrayTest${mpiExt}${debugExt} StaticArrayTest.cpp )
+TARGET_COMPILE_OPTIONS( StaticArrayTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+TARGET_LINK_LIBRARIES( StaticArrayTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
+# NOTE: Vector = Array + VectorOperations, so we test Vector and VectorOperations at the same time
IF( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE( StaticVectorTest${mpiExt}${debugExt} StaticVectorTest.h StaticVectorTest.cu )
- SET_TARGET_PROPERTIES( StaticVectorTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( StaticVectorTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
-ELSE( BUILD_CUDA )
- ADD_EXECUTABLE( StaticVectorTest${mpiExt}${debugExt} StaticVectorTest.h StaticVectorTest.cpp )
- SET_TARGET_PROPERTIES( StaticVectorTest${mpiExt}${debugExt} PROPERTIES COMPILE_FLAGS "${CXX_TESTS_FLAGS}" )
- TARGET_LINK_LIBRARIES( StaticVectorTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
- tnl${mpiExt}${debugExt}-${tnlVersion} )
+ CUDA_ADD_EXECUTABLE( VectorTest${mpiExt}${debugExt} VectorTest.cu
+ OPTIONS ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( VectorTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
+ELSE( BUILD_CUDA )
+ ADD_EXECUTABLE( VectorTest${mpiExt}${debugExt} VectorTest.cpp )
+ TARGET_COMPILE_OPTIONS( VectorTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( VectorTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
ENDIF( BUILD_CUDA )
+ADD_EXECUTABLE( StaticVectorTest${mpiExt}${debugExt} StaticVectorTest.cpp )
+TARGET_COMPILE_OPTIONS( StaticVectorTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+TARGET_LINK_LIBRARIES( StaticVectorTest${mpiExt}${debugExt}
+ ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
+
+#IF( BUILD_CUDA )
+# CUDA_ADD_EXECUTABLE( MultiArrayTest${mpiExt}${debugExt} MultiArrayTest.cu
+# OPTIONS ${CXX_TESTS_FLAGS} )
+# TARGET_LINK_LIBRARIES( MultiArrayTest${mpiExt}${debugExt}
+# ${GTEST_BOTH_LIBRARIES}
+# tnl${mpiExt}${debugExt}-${tnlVersion} )
+#ELSE( BUILD_CUDA )
+# ADD_EXECUTABLE( MultiArrayTest${mpiExt}${debugExt} MultiArrayTest.cpp )
+# TARGET_COMPILE_OPTIONS( MultiArrayTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+# TARGET_LINK_LIBRARIES( MultiArrayTest${mpiExt}${debugExt}
+# ${GTEST_BOTH_LIBRARIES}
+# tnl${mpiExt}${debugExt}-${tnlVersion} )
+#ENDIF( BUILD_CUDA )
+
+ADD_TEST( ListTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/ListTest${mpiExt}${debugExt} )
ADD_TEST( ArrayOperationsTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/ArrayOperationsTest${mpiExt}${debugExt} )
ADD_TEST( ArrayTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/ArrayTest${mpiExt}${debugExt} )
-ADD_TEST( MultiArrayTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/MultiArrayTest${mpiExt}${debugExt} )
ADD_TEST( StaticArrayTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/StaticArrayTest${mpiExt}${debugExt} )
-ADD_TEST( VectorOperationsTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/VectorOperationsTest${mpiExt}${debugExt} )
-ADD_TEST( StaticVectorTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/StaticVectorTest${mpiExt}${debugExt} )
\ No newline at end of file
+ADD_TEST( VectorTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/VectorTest${mpiExt}${debugExt} )
+ADD_TEST( StaticVectorTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/StaticVectorTest${mpiExt}${debugExt} )
+#ADD_TEST( MultiArrayTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/MultiArrayTest${mpiExt}${debugExt} )
diff --git a/src/UnitTests/Containers/ListTest.cpp b/src/UnitTests/Containers/ListTest.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..ec81c1f0bf8f3d25be8d0daf98e2625cdc65ab69
--- /dev/null
+++ b/src/UnitTests/Containers/ListTest.cpp
@@ -0,0 +1,125 @@
+/***************************************************************************
+ ListTest.cpp - description
+ -------------------
+ begin : Feb 15, 2014
+ copyright : (C) 2014 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#ifdef HAVE_GTEST
+#include <gtest/gtest.h>
+
+#include <TNL/Containers/List.h>
+
+using namespace TNL;
+using namespace TNL::Containers;
+
+
+// test fixture for typed tests
+template< typename List >
+class ListTest : public ::testing::Test
+{
+protected:
+ using ListType = List;
+};
+
+// types for which ListTest is instantiated
+using ListTypes = ::testing::Types<
+ List< short >,
+ List< int >,
+ List< long >,
+ List< float >,
+ List< double >,
+ List< String >
+>;
+
+TYPED_TEST_CASE( ListTest, ListTypes );
+
+
+TYPED_TEST( ListTest, constructor )
+{
+ using ListType = typename TestFixture::ListType;
+
+ ListType list;
+ EXPECT_TRUE( list.isEmpty() );
+ EXPECT_EQ( list.getSize(), 0 );
+
+ list.Append( 0 );
+ EXPECT_EQ( list.getSize(), 1 );
+
+ ListType copy( list );
+ list.Append( 0 );
+ EXPECT_EQ( list.getSize(), 2 );
+ EXPECT_EQ( copy.getSize(), 1 );
+ EXPECT_EQ( copy[ 0 ], list[ 0 ] );
+}
+
+TYPED_TEST( ListTest, operations )
+{
+ using ListType = typename TestFixture::ListType;
+ using ElementType = typename ListType::ElementType;
+
+ ListType a, b;
+
+ a.Append( 0 );
+ a.Append( 1 );
+ a.Prepend( 2 );
+ a.Insert( 3, 1 );
+ EXPECT_EQ( a.getSize(), 4 );
+ EXPECT_EQ( a[ 0 ], (ElementType) 2 );
+ EXPECT_EQ( a[ 1 ], (ElementType) 3 );
+ EXPECT_EQ( a[ 2 ], (ElementType) 0 );
+ EXPECT_EQ( a[ 3 ], (ElementType) 1 );
+
+ b = a;
+ EXPECT_EQ( b.getSize(), 4 );
+ EXPECT_EQ( a, b );
+
+ b.Insert( 4, 4 );
+ EXPECT_NE( a, b );
+ EXPECT_EQ( b[ 4 ], (ElementType) 4 );
+
+ a.AppendList( b );
+ EXPECT_EQ( a.getSize(), 9 );
+ EXPECT_EQ( a[ 0 ], (ElementType) 2 );
+ EXPECT_EQ( a[ 1 ], (ElementType) 3 );
+ EXPECT_EQ( a[ 2 ], (ElementType) 0 );
+ EXPECT_EQ( a[ 3 ], (ElementType) 1 );
+ EXPECT_EQ( a[ 4 ], (ElementType) 2 );
+ EXPECT_EQ( a[ 5 ], (ElementType) 3 );
+ EXPECT_EQ( a[ 6 ], (ElementType) 0 );
+ EXPECT_EQ( a[ 7 ], (ElementType) 1 );
+ EXPECT_EQ( a[ 8 ], (ElementType) 4 );
+
+ a.PrependList( b );
+ EXPECT_EQ( a.getSize(), 14 );
+ EXPECT_EQ( a[ 0 ], (ElementType) 2 );
+ EXPECT_EQ( a[ 1 ], (ElementType) 3 );
+ EXPECT_EQ( a[ 2 ], (ElementType) 0 );
+ EXPECT_EQ( a[ 3 ], (ElementType) 1 );
+ EXPECT_EQ( a[ 4 ], (ElementType) 4 );
+ EXPECT_EQ( a[ 5 ], (ElementType) 2 );
+ EXPECT_EQ( a[ 6 ], (ElementType) 3 );
+ EXPECT_EQ( a[ 7 ], (ElementType) 0 );
+ EXPECT_EQ( a[ 8 ], (ElementType) 1 );
+ EXPECT_EQ( a[ 9 ], (ElementType) 2 );
+ EXPECT_EQ( a[ 10 ], (ElementType) 3 );
+ EXPECT_EQ( a[ 11 ], (ElementType) 0 );
+ EXPECT_EQ( a[ 12 ], (ElementType) 1 );
+ EXPECT_EQ( a[ 13 ], (ElementType) 4 );
+}
+#endif
+
+
+#include "../GtestMissingError.h"
+int main( int argc, char* argv[] )
+{
+#ifdef HAVE_GTEST
+ ::testing::InitGoogleTest( &argc, argv );
+ return RUN_ALL_TESTS();
+#else
+ throw GtestMissingError();
+#endif
+}
diff --git a/src/UnitTests/Containers/MultiArrayTest.h b/src/UnitTests/Containers/MultiArrayTest.h
index a14fb919e4cbb34b7aaec0f1cb219280f803b06b..c1a506c26a2d451de59819c70cc3faf7c09552ab 100644
--- a/src/UnitTests/Containers/MultiArrayTest.h
+++ b/src/UnitTests/Containers/MultiArrayTest.h
@@ -119,7 +119,7 @@ TEST( MultiArrayTest, testSetGetElement )
Devices::Cuda::passToDevice( u );
testSetGetElementKernel<<< 1, 16 >>>( kernel_u );
Devices::Cuda::freeFromDevice( kernel_u );
- ASSERT_TRUE( checkCudaDevice );
+ ASSERT_TRUE( TNL_CHECK_CUDA_DEVICE );
#endif
}
for( int i = 0; i < 10; i ++ )
@@ -207,14 +207,16 @@ TEST( MultiArrayTest, testSaveAndLoad )
for( int i = 0; i < size; i ++ )
setDiagonalElement( v, i, 3.14147 );
File file;
- file. open( "test-file.tnl", tnlWriteMode );
+ file. open( "test-file.tnl", IOMode::write );
ASSERT_TRUE( v. save( file ) );
file. close();
MultiArray< Dimension, ElementType, Device, IndexType > u;
- file. open( "test-file.tnl", tnlReadMode );
+ file. open( "test-file.tnl", IOMode::read );
ASSERT_TRUE( u. load( file ) );
file. close();
ASSERT_TRUE( u == v );
+
+ EXPECT_EQ( std::remove( "test-file.tnl" ), 0 );
}
#endif /* HAVE_GTEST */
diff --git a/src/UnitTests/Containers/StaticArrayTest.cpp b/src/UnitTests/Containers/StaticArrayTest.cpp
index afb3017f0909920565046725a8b38a332117bc21..d6dedf6236caf347dba2c458e9ef992cfdec5565 100644
--- a/src/UnitTests/Containers/StaticArrayTest.cpp
+++ b/src/UnitTests/Containers/StaticArrayTest.cpp
@@ -8,4 +8,304 @@
/* See Copyright Notice in tnl/Copyright */
-#include "StaticArrayTest.h"
+#ifdef HAVE_GTEST
+#include <TNL/Containers/StaticArray.h>
+#include <TNL/Containers/Array.h>
+
+#include "gtest/gtest.h"
+
+using namespace TNL;
+using namespace TNL::Containers;
+
+
+// test fixture for typed tests
+template< typename Array >
+class StaticArrayTest : public ::testing::Test
+{
+protected:
+ using ArrayType = Array;
+ using ElementType = typename Array::ElementType;
+};
+
+// types for which ArrayTest is instantiated
+using StaticArrayTypes = ::testing::Types<
+ StaticArray< 1, short >,
+ StaticArray< 2, short >,
+ StaticArray< 3, short >,
+ StaticArray< 4, short >,
+ StaticArray< 5, short >,
+ StaticArray< 1, int >,
+ StaticArray< 2, int >,
+ StaticArray< 3, int >,
+ StaticArray< 4, int >,
+ StaticArray< 5, int >,
+ StaticArray< 1, long >,
+ StaticArray< 2, long >,
+ StaticArray< 3, long >,
+ StaticArray< 4, long >,
+ StaticArray< 5, long >,
+ StaticArray< 1, float >,
+ StaticArray< 2, float >,
+ StaticArray< 3, float >,
+ StaticArray< 4, float >,
+ StaticArray< 5, float >,
+ StaticArray< 1, double >,
+ StaticArray< 2, double >,
+ StaticArray< 3, double >,
+ StaticArray< 4, double >,
+ StaticArray< 5, double >
+>;
+
+TYPED_TEST_CASE( StaticArrayTest, StaticArrayTypes );
+
+TYPED_TEST( StaticArrayTest, constructors )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ using ElementType = typename TestFixture::ElementType;
+ constexpr int Size = ArrayType::size;
+
+ ElementType data[ Size ];
+ for( int i = 0; i < Size; i++ )
+ data[ i ] = i;
+
+ ArrayType u0;
+ EXPECT_TRUE( u0.getData() );
+
+ ArrayType u1( data );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( u1[ i ], data[ i ] );
+
+ ArrayType u2( 7 );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( u2[ i ], 7 );
+
+ ArrayType u3( u1 );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( u3[ i ], u1[ i ] );
+
+ // initialization with 0 requires special treatment to avoid ambiguity,
+ // see https://stackoverflow.com/q/4610503
+ ArrayType v( 0 );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( v[ i ], 0 );
+}
+
+TYPED_TEST( StaticArrayTest, getSize )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ constexpr int Size = ArrayType::size;
+
+ ArrayType u;
+ EXPECT_EQ( u.getSize(), Size );
+}
+
+TYPED_TEST( StaticArrayTest, getData )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+
+ ArrayType u1;
+ EXPECT_TRUE( u1.getData() );
+
+ const ArrayType u2;
+ EXPECT_TRUE( u2.getData() );
+}
+
+template< typename Element >
+void checkCoordinates( StaticArray< 1, Element >& u )
+{
+ EXPECT_EQ( u.x(), 0 );
+ u.x() += 1;
+ EXPECT_EQ( u.x(), 1 );
+}
+
+template< typename Element >
+void checkCoordinates( StaticArray< 2, Element >& u )
+{
+ EXPECT_EQ( u.x(), 0 );
+ EXPECT_EQ( u.y(), 1 );
+ u.x() += 1;
+ u.y() += 1;
+ EXPECT_EQ( u.x(), 1 );
+ EXPECT_EQ( u.y(), 2 );
+}
+
+template< typename Element >
+void checkCoordinates( StaticArray< 3, Element >& u )
+{
+ EXPECT_EQ( u.x(), 0 );
+ EXPECT_EQ( u.y(), 1 );
+ EXPECT_EQ( u.z(), 2 );
+ u.x() += 1;
+ u.y() += 1;
+ u.z() += 1;
+ EXPECT_EQ( u.x(), 1 );
+ EXPECT_EQ( u.y(), 2 );
+ EXPECT_EQ( u.z(), 3 );
+}
+
+template< int _Size, typename Element >
+void checkCoordinates( StaticArray< _Size, Element >& u )
+{
+}
+
+TYPED_TEST( StaticArrayTest, CoordinatesGetter )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ constexpr int Size = ArrayType::size;
+
+ ArrayType u;
+ for( int i = 0; i < Size; i++ )
+ u[ i ] = i;
+
+ checkCoordinates( u );
+}
+
+TYPED_TEST( StaticArrayTest, ComparisonOperator )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ constexpr int Size = ArrayType::size;
+
+ ArrayType u1, u2, u3;
+
+ for( int i = 0; i < Size; i++ ) {
+ u1[ i ] = 1;
+ u2[ i ] = i;
+ u3[ i ] = i;
+ }
+
+ EXPECT_TRUE( u1 == u1 );
+ EXPECT_TRUE( u1 != u2 );
+ EXPECT_TRUE( u2 == u3 );
+
+ // comparison with different type
+ StaticArray< Size, char > u4( 1 );
+ EXPECT_TRUE( u1 == u4 );
+ EXPECT_TRUE( u2 != u4 );
+ EXPECT_TRUE( u3 != u4 );
+
+ for( int i = 0; i < Size; i++ )
+ u4[ i ] = i;
+ EXPECT_TRUE( u1 != u4 );
+ EXPECT_TRUE( u2 == u4 );
+ EXPECT_TRUE( u3 == u4 );
+}
+
+TYPED_TEST( StaticArrayTest, AssignmentOperator )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ constexpr int Size = ArrayType::size;
+
+ ArrayType u1, u2, u3;
+
+ for( int i = 0; i < Size; i++ )
+ {
+ u1[ i ] = 1;
+ u2[ i ] = i;
+ }
+
+ u3 = u1;
+ EXPECT_TRUE( u3 == u1 );
+ EXPECT_TRUE( u3 != u2 );
+
+ u3 = u2;
+ EXPECT_TRUE( u3 == u2 );
+ EXPECT_TRUE( u3 != u1 );
+
+ // assignment from different type
+ StaticArray< Size, char > u4( 128 );
+ u3 = u4;
+ EXPECT_TRUE( u3 == u4 );
+}
+
+TYPED_TEST( StaticArrayTest, setValue )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ constexpr int Size = ArrayType::size;
+
+ ArrayType u;
+ u.setValue( 42 );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( u[ i ], 42 );
+}
+
+TYPED_TEST( StaticArrayTest, CastToDifferentStaticArray )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ constexpr int Size = ArrayType::size;
+ using OtherArray = StaticArray< Size, char >;
+
+ ArrayType u1( 1 );
+ OtherArray u2( 1 );
+ EXPECT_EQ( (OtherArray) u1, u2 );
+ EXPECT_EQ( u1, (ArrayType) u2 );
+}
+
+TYPED_TEST( StaticArrayTest, SaveAndLoad )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+
+ ArrayType u1( 7 ), u2;
+ File file;
+ file.open( "tnl-static-array-test.tnl", IOMode::write );
+ u1.save( file );
+ file.close();
+ file.open( "tnl-static-array-test.tnl", IOMode::read );
+ u2.load( file );
+ file.close();
+
+ EXPECT_EQ( u1, u2 );
+
+ EXPECT_EQ( std::remove( "tnl-static-array-test.tnl" ), 0 );
+}
+
+TYPED_TEST( StaticArrayTest, sort )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ constexpr int Size = ArrayType::size;
+
+ ArrayType u;
+ for( int i = 0; i < Size; i++ )
+ u[ i ] = Size - i - 1;
+ u.sort();
+
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( u[ i ], i );
+}
+
+TYPED_TEST( StaticArrayTest, streamOperator )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+
+ ArrayType u;
+ std::stringstream testStream;
+ testStream << u;
+}
+
+TYPED_TEST( StaticArrayTest, BindToArray )
+{
+ using ArrayType = typename TestFixture::ArrayType;
+ using ElementType = typename TestFixture::ElementType;
+ constexpr int Size = ArrayType::size;
+
+ ArrayType a;
+ for( int i = 0; i < Size; i++ )
+ a[ i ] = i+1;
+
+ Array< ElementType, Devices::Host > sharedArray;
+ sharedArray.bind( a );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( a[ i ], sharedArray[ i ] );
+}
+#endif // HAVE_GTEST
+
+
+#include "../GtestMissingError.h"
+int main( int argc, char* argv[] )
+{
+#ifdef HAVE_GTEST
+ ::testing::InitGoogleTest( &argc, argv );
+ return RUN_ALL_TESTS();
+#else
+ throw GtestMissingError();
+#endif
+}
diff --git a/src/UnitTests/Containers/StaticArrayTest.h b/src/UnitTests/Containers/StaticArrayTest.h
deleted file mode 100644
index b26af6f60673f037c4cea63d3a6d1eddc9b06c42..0000000000000000000000000000000000000000
--- a/src/UnitTests/Containers/StaticArrayTest.h
+++ /dev/null
@@ -1,188 +0,0 @@
-/***************************************************************************
- StaticArrayTester.h - description
- -------------------
- begin : Feb 10, 2014
- copyright : (C) 2014 by Tomas Oberhuber
- email : tomas.oberhuber@fjfi.cvut.cz
- ***************************************************************************/
-
-/* See Copyright Notice in tnl/Copyright */
-
-#pragma once
-
-#include <TNL/Containers/StaticArray.h>
-#include <TNL/Containers/Array.h>
-
-#ifdef HAVE_GTEST
-#include "gtest/gtest.h"
-#endif
-
-using namespace TNL;
-using namespace TNL::Containers;
-
-class testingClassForStaticArrayTester
-{
- public:
-
- static String getType()
- {
- return String( "testingClassForStaticArrayTester" );
- };
-};
-
-String getType( const testingClassForStaticArrayTester& c )
-{
- return String( "testingClassForStaticArrayTester" );
-};
-
-#ifdef HAVE_GTEST
-
-typedef int ElementType;
-const int Size( 16 );
-
-
-TEST( StaticArrayTest, testConstructors )
-{
- ElementType data[ Size ];
- for( int i = 0; i < Size; i++ )
- data[ i ] = i;
-
- StaticArray< Size, ElementType > u1( data );
- for( int i = 0; i < Size; i++ )
- ASSERT_EQ( u1[ i ], data[ i ] );
-
- StaticArray< Size, ElementType > u2( 7 );
- for( int i = 0; i < Size; i++ )
- ASSERT_EQ( u2[ i ], 7 );
-
- StaticArray< Size, ElementType > u3( u1 );
- for( int i = 0; i < Size; i++ )
- ASSERT_EQ( u3[ i ], u1[ i ] );
-}
-
-template< typename Element >
-void checkCoordinates( const StaticArray< 1, Element >& u )
-{
- ASSERT_EQ( u.x(), 0 );
-}
-
-template< typename Element >
-void checkCoordinates( const StaticArray< 2, Element >& u )
-{
- ASSERT_EQ( u.x(), 0 );
- ASSERT_EQ( u.y(), 1 );
-}
-
-template< typename Element >
-void checkCoordinates( const StaticArray< 3, Element >& u )
-{
- ASSERT_EQ( u.x(), 0 );
- ASSERT_EQ( u.y(), 1 );
- ASSERT_EQ( u.z(), 2 );
-}
-
-template< int _Size, typename Element >
-void checkCoordinates( const StaticArray< _Size, Element >& u )
-{
-}
-
-TEST( StaticArrayTest, testCoordinatesGetter )
-{
- StaticArray< Size, ElementType > u;
- for( int i = 0; i < Size; i++ )
- u[ i ] = i;
-
- checkCoordinates( u );
-}
-
-TEST( StaticArrayTest, testComparisonOperator )
-{
- StaticArray< Size, ElementType > u1, u2, u3;
-
- for( int i = 0; i < Size; i++ )
- {
- u1[ i ] = 1;
- u2[ i ] = i;
- u3[ i ] = i;
- }
-
- ASSERT_TRUE( u1 == u1 );
- ASSERT_TRUE( u1 != u2 );
- ASSERT_TRUE( u2 == u3 );
-}
-
-TEST( StaticArrayTest, testAssignmentOperator )
-{
- StaticArray< Size, ElementType > u1, u2, u3;
-
- for( int i = 0; i < Size; i++ )
- {
- u1[ i ] = 1;
- u2[ i ] = i;
- }
-
- u3 = u1;
- ASSERT_TRUE( u3 == u1 );
- ASSERT_TRUE( u3 != u2 );
-
- u3 = u2;
- ASSERT_TRUE( u3 == u2 );
- ASSERT_TRUE( u3 != u1 );
-}
-
-TEST( StaticArrayTest, testLoadAndSave )
-{
- StaticArray< Size, ElementType > u1( 7 ), u2( 0 );
- File file;
- file.open( "tnl-static-array-test.tnl", tnlWriteMode );
- u1.save( file );
- file.close();
- file.open( "tnl-static-array-test.tnl", tnlReadMode );
- u2.load( file );
- file.close();
-
- ASSERT_EQ( u1, u2 );
-}
-
-TEST( StaticArrayTest, testSort )
-{
- StaticArray< Size, ElementType > u;
- for( int i = 0; i < Size; i++ )
- u[ i ] = Size - i - 1;
- u.sort();
-
- for( int i = 0; i < Size; i++ )
- ASSERT_EQ( u[ i ], i );
-}
-
-TEST( StaticArrayTest, testStreamOperator )
-{
- StaticArray< Size, ElementType > u;
- std::stringstream testStream;
- testStream << u;
-}
-
-TEST( StaticArrayTest, testBindToArray )
-{
- StaticArray< Size, ElementType > a;
- for( int i = 0; i < Size; i++ )
- a[ i ] = i+1;
-
- Array< ElementType, Devices::Host > sharedArray;
- sharedArray.bind( a );
- for( int i = 0; i < Size; i++ )
- ASSERT_EQ( a[ i ], sharedArray[ i ] );
-
-}
-#endif /* HAVE_GTEST */
-
-int main( int argc, char* argv[] )
-{
-#ifdef HAVE_GTEST
- ::testing::InitGoogleTest( &argc, argv );
- return RUN_ALL_TESTS();
-#else
- return EXIT_FAILURE;
-#endif
-}
-
diff --git a/src/UnitTests/Containers/StaticVectorTest.cpp b/src/UnitTests/Containers/StaticVectorTest.cpp
index 3505faf2c9ea1cc85566d7f1347f1fb3a8ec2ff5..e34db3f33833dc3254e13bc0b2c650a2613abd22 100644
--- a/src/UnitTests/Containers/StaticVectorTest.cpp
+++ b/src/UnitTests/Containers/StaticVectorTest.cpp
@@ -8,4 +8,196 @@
/* See Copyright Notice in tnl/Copyright */
-#include "StaticVectorTest.h"
+#ifdef HAVE_GTEST
+#include <TNL/Containers/StaticVector.h>
+
+#include "gtest/gtest.h"
+
+using namespace TNL;
+using namespace TNL::Containers;
+
+// test fixture for typed tests
+template< typename Vector >
+class StaticVectorTest : public ::testing::Test
+{
+protected:
+ using VectorType = Vector;
+ using RealType = typename VectorType::RealType;
+};
+
+// types for which VectorTest is instantiated
+using StaticVectorTypes = ::testing::Types<
+ StaticVector< 1, short >,
+ StaticVector< 1, int >,
+ StaticVector< 1, long >,
+ StaticVector< 1, float >,
+ StaticVector< 1, double >,
+ StaticVector< 2, short >,
+ StaticVector< 2, int >,
+ StaticVector< 2, long >,
+ StaticVector< 2, float >,
+ StaticVector< 2, double >,
+ StaticVector< 3, short >,
+ StaticVector< 3, int >,
+ StaticVector< 3, long >,
+ StaticVector< 3, float >,
+ StaticVector< 3, double >,
+ StaticVector< 4, short >,
+ StaticVector< 4, int >,
+ StaticVector< 4, long >,
+ StaticVector< 4, float >,
+ StaticVector< 4, double >,
+ StaticVector< 5, short >,
+ StaticVector< 5, int >,
+ StaticVector< 5, long >,
+ StaticVector< 5, float >,
+ StaticVector< 5, double >
+>;
+
+TYPED_TEST_CASE( StaticVectorTest, StaticVectorTypes );
+
+
+TYPED_TEST( StaticVectorTest, constructors )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using RealType = typename TestFixture::RealType;
+ constexpr int Size = VectorType::size;
+
+ RealType data[ Size ];
+ for( int i = 0; i < Size; i++ )
+ data[ i ] = i;
+
+ VectorType u0;
+ EXPECT_TRUE( u0.getData() );
+
+ VectorType u1( data );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( u1[ i ], data[ i ] );
+
+ VectorType u2( 7 );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( u2[ i ], 7 );
+
+ VectorType u3( u1 );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( u3[ i ], u1[ i ] );
+
+ // initialization with 0 requires special treatment to avoid ambiguity,
+ // see https://stackoverflow.com/q/4610503
+ VectorType v( 0 );
+ for( int i = 0; i < Size; i++ )
+ EXPECT_EQ( v[ i ], 0 );
+}
+
+TYPED_TEST( StaticVectorTest, operators )
+{
+ using VectorType = typename TestFixture::VectorType;
+ constexpr int size = VectorType::size;
+
+ VectorType u1( 1 ), u2( 2 ), u3( 3 );
+
+ u1 += u2;
+ EXPECT_EQ( u1[ 0 ], 3 );
+ EXPECT_EQ( u1[ size - 1 ], 3 );
+
+ u1 -= u2;
+ EXPECT_EQ( u1[ 0 ], 1 );
+ EXPECT_EQ( u1[ size - 1 ], 1 );
+
+ u1 *= 2;
+ EXPECT_EQ( u1[ 0 ], 2 );
+ EXPECT_EQ( u1[ size - 1 ], 2 );
+
+ u3 = u1 + u2;
+ EXPECT_EQ( u3[ 0 ], 4 );
+ EXPECT_EQ( u3[ size - 1 ], 4 );
+
+ u3 = u1 - u2;
+ EXPECT_EQ( u3[ 0 ], 0 );
+ EXPECT_EQ( u3[ size - 1 ], 0 );
+
+ u3 = 2 * u1;
+ EXPECT_EQ( u3[ 0 ], 4 );
+ EXPECT_EQ( u3[ size - 1 ], 4 );
+
+ EXPECT_EQ( u1 * u2, 4 * size );
+}
+
+TYPED_TEST( StaticVectorTest, comparisons )
+{
+ using VectorType = typename TestFixture::VectorType;
+ constexpr int size = VectorType::size;
+
+ VectorType u1( 1 ), u2( 2 ), u3( 3 ), u4;
+ for( int i = 0; i < size; i++ )
+ u4[ i ] = i;
+
+ EXPECT_TRUE( u1 < u3 );
+ EXPECT_TRUE( u1 <= u3 );
+ EXPECT_TRUE( u1 < u2 );
+ EXPECT_TRUE( u1 <= u2 );
+ EXPECT_TRUE( u3 > u1 );
+ EXPECT_TRUE( u3 >= u1 );
+ EXPECT_TRUE( u2 > u1 );
+ EXPECT_TRUE( u2 >= u1 );
+
+ if( size > 2 ) {
+ EXPECT_FALSE( u1 < u4 );
+ EXPECT_FALSE( u1 <= u4 );
+ EXPECT_FALSE( u1 > u4 );
+ EXPECT_FALSE( u1 >= u4 );
+ }
+}
+
+TYPED_TEST( StaticVectorTest, cast )
+{
+ using VectorType = typename TestFixture::VectorType;
+ constexpr int size = VectorType::size;
+
+ VectorType u( 1 );
+ EXPECT_EQ( (StaticVector< size, double >) u, u );
+}
+
+TYPED_TEST( StaticVectorTest, abs )
+{
+ using VectorType = typename TestFixture::VectorType;
+ constexpr int size = VectorType::size;
+
+ VectorType u;
+ for( int i = 0; i < size; i++ )
+ u[ i ] = i;
+
+ // TODO: implement unary minus operator
+ VectorType v = - 1 * u;
+ EXPECT_EQ( v.abs(), u );
+}
+
+TYPED_TEST( StaticVectorTest, lpNorm )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using RealType = typename TestFixture::RealType;
+ constexpr int size = VectorType::size;
+ const RealType epsilon = std::numeric_limits< RealType >::epsilon();
+
+ VectorType v( 1 );
+
+ const RealType expectedL1norm = size;
+ const RealType expectedL2norm = std::sqrt( size );
+ const RealType expectedL3norm = std::cbrt( size );
+ EXPECT_EQ( v.lpNorm( 1.0 ), expectedL1norm );
+ EXPECT_EQ( v.lpNorm( 2.0 ), expectedL2norm );
+ EXPECT_NEAR( v.lpNorm( 3.0 ), expectedL3norm, epsilon );
+}
+#endif
+
+
+#include "../GtestMissingError.h"
+int main( int argc, char* argv[] )
+{
+#ifdef HAVE_GTEST
+ ::testing::InitGoogleTest( &argc, argv );
+ return RUN_ALL_TESTS();
+#else
+ throw GtestMissingError();
+#endif
+}
diff --git a/src/UnitTests/Containers/StaticVectorTest.cu b/src/UnitTests/Containers/StaticVectorTest.cu
deleted file mode 100644
index 693ace5c43590a06a2f8c7c087a0d1480a35d02a..0000000000000000000000000000000000000000
--- a/src/UnitTests/Containers/StaticVectorTest.cu
+++ /dev/null
@@ -1,11 +0,0 @@
-/***************************************************************************
- StaticVectorTest.cu - description
- -------------------
- begin : Feb 10, 2014
- copyright : (C) 2014 by Tomas Oberhuber
- email : tomas.oberhuber@fjfi.cvut.cz
- ***************************************************************************/
-
-/* See Copyright Notice in tnl/Copyright */
-
-#include "StaticVectorTest.h"
diff --git a/src/UnitTests/Containers/StaticVectorTest.h b/src/UnitTests/Containers/StaticVectorTest.h
deleted file mode 100644
index 038ea0f4c7959a36e6cb3bcb8852267ca321e013..0000000000000000000000000000000000000000
--- a/src/UnitTests/Containers/StaticVectorTest.h
+++ /dev/null
@@ -1,74 +0,0 @@
-/***************************************************************************
- StaticVectorTester.h - description
- -------------------
- begin : Feb 10, 2014
- copyright : (C) 2014 by Tomas Oberhuber
- email : tomas.oberhuber@fjfi.cvut.cz
- ***************************************************************************/
-
-/* See Copyright Notice in tnl/Copyright */
-
-#pragma once
-
-#include <TNL/Containers/StaticVector.h>
-
-#ifdef HAVE_GTEST
-#include "gtest/gtest.h"
-#endif
-
-using namespace TNL;
-
-#ifdef HAVE_GTEST
-
-const int Size( 16 );
-typedef double RealType;
-
-TEST( StaticVectorTest, testOperators )
-{
- Containers::StaticVector< Size, RealType > u1( 1.0 ), u2( 2.0 ), u3( 3.0 );
-
- u1 += u2;
- ASSERT_TRUE( u1[ 0 ] == 3.0 );
- ASSERT_TRUE( u1[ Size - 1 ] == 3.0 );
-
- u1 -= u2;
- ASSERT_TRUE( u1[ 0 ] == 1.0 );
- ASSERT_TRUE( u1[ Size - 1 ] == 1.0 );
-
- u1 *= 2.0;
- ASSERT_TRUE( u1[ 0 ] == 2.0 );
- ASSERT_TRUE( u1[ Size - 1 ] == 2.0 );
-
- u3 = u1 + u2;
- ASSERT_TRUE( u3[ 0 ] == 4.0 );
- ASSERT_TRUE( u3[ Size - 1 ] == 4.0 );
-
- u3 = u1 - u2;
- ASSERT_TRUE( u3[ 0 ] == 0.0 );
- ASSERT_TRUE( u3[ Size - 1 ] == 0.0 );
-
- u3 = u1 * 2.0;
- ASSERT_TRUE( u3[ 0 ] == 4.0 );
- ASSERT_TRUE( u3[ Size - 1 ] == 4.0 );
-
- ASSERT_TRUE( u1 * u2 == 4.0 * Size );
-
- ASSERT_TRUE( u1 < u3 );
- ASSERT_TRUE( u1 <= u3 );
- ASSERT_TRUE( u1 <= u2 );
- ASSERT_TRUE( u3 > u1 );
- ASSERT_TRUE( u3 >= u1 );
- ASSERT_TRUE( u2 >= u1 );
-}
-
-#endif
-
-int main( int argc, char* argv[] )
-{
-#ifdef HAVE_GTEST
- ::testing::InitGoogleTest( &argc, argv );
- return RUN_ALL_TESTS();
-#else
- return EXIT_FAILURE;
-#endif
-}
diff --git a/src/UnitTests/Containers/VectorOperationsTest.cu b/src/UnitTests/Containers/VectorOperationsTest.cu
deleted file mode 100644
index 61d3b0357b9fb1c3e5b8decfe74ed6cae9446c25..0000000000000000000000000000000000000000
--- a/src/UnitTests/Containers/VectorOperationsTest.cu
+++ /dev/null
@@ -1,11 +0,0 @@
-/***************************************************************************
- Devices::CudaVectorOperationsTest.cu - description
- -------------------
- begin : Mar 31, 2013
- copyright : (C) 2013 by Tomas Oberhuber
- email : tomas.oberhuber@fjfi.cvut.cz
- ***************************************************************************/
-
-/* See Copyright Notice in tnl/Copyright */
-
-#include "VectorOperationsTest.h"
diff --git a/src/UnitTests/Containers/VectorOperationsTest.h b/src/UnitTests/Containers/VectorOperationsTest.h
deleted file mode 100644
index 08d18b1704ca0c23ee32928551b55a5feb96c772..0000000000000000000000000000000000000000
--- a/src/UnitTests/Containers/VectorOperationsTest.h
+++ /dev/null
@@ -1,342 +0,0 @@
-/***************************************************************************
- VectorOperationsTester.h - description
- -------------------
- begin : Mar 30, 2013
- copyright : (C) 2013 by Tomas Oberhuber
- email : tomas.oberhuber@fjfi.cvut.cz
- ***************************************************************************/
-
-/* See Copyright Notice in tnl/Copyright */
-
-#pragma once
-
-#ifdef HAVE_GTEST
-#include "gtest/gtest.h"
-#endif
-
-#include <TNL/Containers/Vector.h>
-#include <TNL/Containers/VectorOperations.h>
-
-using namespace TNL;
-using namespace TNL::Containers;
-using namespace TNL::Containers::Algorithms;
-
-#ifdef HAVE_GTEST
-
-typedef double Real;
-typedef Devices::Host Device;
-typedef int Index;
-
-template< typename Vector >
-void setLinearSequence( Vector& deviceVector )
-{
- Containers::Vector< typename Vector :: RealType, Devices::Host > a;
- a. setSize( deviceVector. getSize() );
- for( int i = 0; i < a. getSize(); i ++ )
- a. getData()[ i ] = i;
-
- ArrayOperations< typename Vector::DeviceType,
- Devices::Host >::
- template copyMemory< typename Vector::RealType,
- typename Vector::RealType,
- typename Vector::IndexType >
- ( deviceVector.getData(),
- a.getData(),
- a.getSize() );
-}
-
-
-template< typename Vector >
-void setOnesSequence( Vector& deviceVector )
-{
- Containers::Vector< typename Vector :: RealType, Devices::Host > a;
- a. setSize( deviceVector. getSize() );
- for( int i = 0; i < a. getSize(); i ++ )
- a. getData()[ i ] = 1;
-
- ArrayOperations< typename Vector::DeviceType,
- Devices::Host >::
- template copyMemory< typename Vector::RealType,
- typename Vector::RealType,
- typename Vector::IndexType >
- ( deviceVector.getData(),
- a.getData(),
- a.getSize() );
-}
-
-
-template< typename Vector >
-void setNegativeLinearSequence( Vector& deviceVector )
-{
- Containers::Vector< typename Vector :: RealType, Devices::Host > a;
- a. setSize( deviceVector. getSize() );
- for( int i = 0; i < a. getSize(); i ++ )
- a. getData()[ i ] = -i;
-
- ArrayOperations< typename Vector::DeviceType,
- Devices::Host >::
- template copyMemory< typename Vector::RealType,
- typename Vector::RealType,
- typename Vector::IndexType >
- ( deviceVector.getData(),
- a.getData(),
- a.getSize() );
-}
-
-template< typename Vector >
-void setOscilatingSequence( Vector& deviceVector,
- typename Vector::RealType v )
-{
- Containers::Vector< typename Vector::RealType, Devices::Host > a;
- a.setSize( deviceVector. getSize() );
- a[ 0 ] = v;
- for( int i = 1; i < a. getSize(); i ++ )
- a.getData()[ i ] = a.getData()[ i-1 ] * -1;
-
- ArrayOperations< typename Vector::DeviceType,
- Devices::Host >::
- template copyMemory< typename Vector::RealType,
- typename Vector::RealType,
- typename Vector::IndexType >
- ( deviceVector.getData(),
- a.getData(),
- a.getSize() );
-}
-
-
-
-TEST( VectorOperationsTest, getVectorMaxTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > v;
- v. setSize( size );
- setLinearSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorMax( v ) == size - 1 );
-}
-
-TEST( VectorOperationsTest, getVectorMinTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > v;
- v. setSize( size );
- setLinearSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorMin( v ) == 0 );
-}
-
-TEST( VectorOperationsTest, getVectorAbsMaxTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > v;
- v. setSize( size );
- setNegativeLinearSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorAbsMax( v ) == size - 1 );
-}
-
-TEST( VectorOperationsTest, getVectorAbsMinTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > v;
- v. setSize( size );
- setNegativeLinearSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorAbsMin( v ) == 0 );
-}
-
-TEST( VectorOperationsTest, getVectorLpNormTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > v;
- v. setSize( size );
- setOnesSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorLpNorm( v, 2.0 ) == ::sqrt( size ) );
-}
-
-TEST( VectorOperationsTest, getVectorSumTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > v;
- v. setSize( size );
- setOnesSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorSum( v ) == size );
-
- setLinearSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorSum( v ) == ( ( Real ) size ) * ( ( Real ) size - 1 ) / 2 );
-}
-
-TEST( VectorOperationsTest, getVectorDifferenceMaxTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > u, v;
- u. setSize( size );
- v. setSize( size );
- setLinearSequence( u );
- setOnesSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorDifferenceMax( u, v ) == size - 2 );
-}
-
-TEST( VectorOperationsTest, getVectorDifferenceMinTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > u, v;
- u. setSize( size );
- v. setSize( size );
- setLinearSequence( u );
- setOnesSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorDifferenceMin( u, v ) == -1 );
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorDifferenceMin( v, u ) == -123454 );
-}
-
-TEST( VectorOperationsTest, getVectorDifferenceAbsMaxTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > u, v;
- u. setSize( size );
- v. setSize( size );
- setNegativeLinearSequence( u );
- setOnesSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorDifferenceAbsMax( u, v ) == size );
-}
-
-TEST( VectorOperationsTest, getVectorDifferenceAbsMinTest )
-{
- const int size( 123456 );
- Containers::Vector< Real, Device > u, v;
- u. setSize( size );
- v. setSize( size );
- setLinearSequence( u );
- setOnesSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorDifferenceAbsMin( u, v ) == 0 );
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorDifferenceAbsMin( v, u ) == 0 );
-}
-
-TEST( VectorOperationsTest, getVectorDifferenceLpNormTest )
-{
- const int size( 1024 );
- Containers::Vector< Real, Device > u, v;
- u. setSize( size );
- v. setSize( size );
- u. setValue( 3.0 );
- v. setValue( 1.0 );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorDifferenceLpNorm( u, v, 1.0 ) == 2.0 * size );
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorDifferenceLpNorm( u, v, 2.0 ) == ::sqrt( 4.0 * size ) );
-}
-
-TEST( VectorOperationsTest, getVectorDifferenceSumTest )
-{
- const int size( 1024 );
- Containers::Vector< Real, Device > u, v;
- u. setSize( size );
- v. setSize( size );
- u. setValue( 3.0 );
- v. setValue( 1.0 );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getVectorDifferenceSum( u, v ) == 2.0 * size );
-}
-
-TEST( VectorOperationsTest, vectorScalarMultiplicationTest )
-{
- const int size( 1025 );
- Containers::Vector< Real, Device > u;
- u. setSize( size );
- setLinearSequence( u );
-
- Containers::VectorOperations< Device >::vectorScalarMultiplication( u, 3.0 );
-
- for( int i = 0; i < size; i++ )
- ASSERT_TRUE( u.getElement( i ) == 3.0 * i );
-}
-
-TEST( VectorOperationsTest, getVectorScalarProductTest )
-{
- const int size( 1025 );
- Containers::Vector< Real, Device > u, v;
- u. setSize( size );
- v. setSize( size );
- setOscilatingSequence( u, 1.0 );
- setOnesSequence( v );
-
- ASSERT_TRUE( Containers::VectorOperations< Device > :: getScalarProduct( u, v ) == 1.0 );
-}
-
-TEST( VectorOperationsTest, addVectorTest )
-{
- const int size( 10000 );
- Containers::Vector< Real, Device > x, y;
- x.setSize( size );
- y.setSize( size );
- setLinearSequence( x );
- setOnesSequence( y );
- Containers::VectorOperations< Device >::addVector( y, x, 3.0 );
-
- for( int i = 0; i < size; i ++ )
- ASSERT_TRUE( y.getElement( i ) == 1.0 + 3.0 * i );
-};
-
-TEST( VectorOperationsTest, prefixSumTest )
-{
- const int size( 10000 );
- Containers::Vector< Real, Device > v;
- v.setSize( size );
-
- setOnesSequence( v );
- v.computePrefixSum();
- for( int i = 0; i < size; i++ )
- ASSERT_TRUE( v.getElement( i ) == i + 1 );
-
- v.setValue( 0 );
- v.computePrefixSum();
- for( int i = 0; i < size; i++ )
- ASSERT_TRUE( v.getElement( i ) == 0 );
-
- setLinearSequence( v );
- v.computePrefixSum();
- for( int i = 1; i < size; i++ )
- ASSERT_TRUE( v.getElement( i ) - v.getElement( i - 1 ) == i );
-
-};
-
-TEST( VectorOperationsTest, exclusivePrefixSumTest )
-{
- const int size( 10000 );
- Containers::Vector< Real, Device > v;
- v.setSize( size );
-
- setOnesSequence( v );
- v.computeExclusivePrefixSum();
- for( int i = 0; i < size; i++ )
- ASSERT_TRUE( v.getElement( i ) == i );
-
- v.setValue( 0 );
- v.computeExclusivePrefixSum();
- for( int i = 0; i < size; i++ )
- ASSERT_TRUE( v.getElement( i ) == 0 );
-
- setLinearSequence( v );
- v.computeExclusivePrefixSum();
- for( int i = 1; i < size; i++ )
- ASSERT_TRUE( v.getElement( i ) - v.getElement( i - 1 ) == i - 1 );
-
-};
-
-#endif /* HAVE_GTEST */
-
-int main( int argc, char* argv[] )
-{
-#ifdef HAVE_GTEST
- ::testing::InitGoogleTest( &argc, argv );
- return RUN_ALL_TESTS();
-#else
- return EXIT_FAILURE;
-#endif
-}
diff --git a/src/UnitTests/Containers/VectorTest.cu b/src/UnitTests/Containers/VectorTest.cu
index e36b670972b5cf89558a9a2e13a3db817631ea31..f173d4a5ee71749ebf318fba56430101125c0bd4 100644
--- a/src/UnitTests/Containers/VectorTest.cu
+++ b/src/UnitTests/Containers/VectorTest.cu
@@ -8,4 +8,4 @@
/* See Copyright Notice in tnl/Copyright */
-#include "tnlVectorTest.h"
+#include "VectorTest.h"
diff --git a/src/UnitTests/Containers/VectorTest.h b/src/UnitTests/Containers/VectorTest.h
index 56966b64a2620c3fa4c7d6aef21a805aa7b0b857..2f3832fcda5439e617dfe9cea27bd8a11dadd6cb 100644
--- a/src/UnitTests/Containers/VectorTest.h
+++ b/src/UnitTests/Containers/VectorTest.h
@@ -1,5 +1,5 @@
/***************************************************************************
- VectorTester.h - description
+ VectorTest.h - description
-------------------
begin : Oct 25, 2010
copyright : (C) 2010 by Tomas Oberhuber
@@ -8,206 +8,526 @@
/* See Copyright Notice in tnl/Copyright */
+// NOTE: Vector = Array + VectorOperations, so we test Vector and VectorOperations at the same time
+
#pragma once
+#ifdef HAVE_GTEST
+#include <limits>
+
#include <TNL/Containers/Vector.h>
#include <TNL/File.h>
#include <TNL/Math.h>
+#include "gtest/gtest.h"
+
using namespace TNL;
+using namespace TNL::Containers;
+using namespace TNL::Containers::Algorithms;
+
+// should be small enough to have fast tests, but larger than minGPUReductionDataSize
+// and large enough to require multiple CUDA blocks for reduction
+constexpr int VECTOR_TEST_SIZE = 5000;
-#ifdef HAVE_GTEST
-TEST( VectorTest, testMax )
+template< typename Vector >
+void setLinearSequence( Vector& deviceVector )
{
- Containers::Vector< RealType, Device, IndexType > v;
- v. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- v. setElement( i, i );
- ASSERT_TRUE( v. max() == 9 );
-};
+ typename Vector::HostType a;
+ a.setLike( deviceVector );
+ for( int i = 0; i < a.getSize(); i++ )
+ a[ i ] = i;
+ deviceVector = a;
+}
-TEST( VectorTest, testMin )
+template< typename Vector >
+void setConstantSequence( Vector& deviceVector,
+ typename Vector::RealType v )
{
- Containers::Vector< RealType, Device, IndexType > v;
- v. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- v. setElement( i, i );
- ASSERT_TRUE( v. min() == 0 );
-};
+ deviceVector.setValue( v );
+}
-TEST( VectorTest, testAbsMax )
+template< typename Vector >
+void setNegativeLinearSequence( Vector& deviceVector )
{
- Containers::Vector< RealType, Device, IndexType > v;
- v. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- v.setElement( i, -i );
- ASSERT_TRUE( v. absMax() == 9 );
-};
+ typename Vector::HostType a;
+ a.setLike( deviceVector );
+ for( int i = 0; i < a.getSize(); i++ )
+ a[ i ] = -i;
+ deviceVector = a;
+}
-TEST( VectorTest, testAbsMin )
+template< typename Vector >
+void setOscilatingSequence( Vector& deviceVector,
+ typename Vector::RealType v )
{
- Containers::Vector< RealType, Device, IndexType > v;
- v. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- v.setElement( i, -i );
- ASSERT_TRUE( v. absMin() == 0 );
-};
+ typename Vector::HostType a;
+ a.setLike( deviceVector );
+ a[ 0 ] = v;
+ for( int i = 1; i < a.getSize(); i++ )
+ a[ i ] = a[ i-1 ] * -1;
+ deviceVector = a;
+}
+
+
+// TODO: test everything with OpenMP with different number of threads
-TEST( VectorTest, testLpNorm )
+// test fixture for typed tests
+template< typename Vector >
+class VectorTest : public ::testing::Test
{
- Containers::Vector< RealType, Device, IndexType > v;
- v. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- v.setElement( i, -2 );
- ASSERT_TRUE( isSmall( v.lpNorm( 1 ) - 20.0 ) );
- ASSERT_TRUE( isSmall( v.lpNorm( 2 ) - ::sqrt( 40.0 ) ) );
- ASSERT_TRUE( isSmall( v.lpNorm( 3 ) - ::pow( 80.0, 1.0/3.0 ) ) );
+protected:
+ using VectorType = Vector;
+ using VectorOperations = Algorithms::VectorOperations< typename VectorType::DeviceType >;
};
-TEST( VectorTest, testSum )
+// types for which VectorTest is instantiated
+using VectorTypes = ::testing::Types<
+ Vector< int, Devices::Host, short >,
+ Vector< long, Devices::Host, short >,
+ Vector< float, Devices::Host, short >,
+ Vector< double, Devices::Host, short >,
+ Vector< int, Devices::Host, int >,
+ Vector< long, Devices::Host, int >,
+ Vector< float, Devices::Host, int >,
+ Vector< double, Devices::Host, int >,
+ Vector< int, Devices::Host, long >,
+ Vector< long, Devices::Host, long >,
+ Vector< float, Devices::Host, long >,
+ Vector< double, Devices::Host, long >
+#ifdef HAVE_CUDA
+ ,
+ Vector< int, Devices::Cuda, short >,
+ Vector< long, Devices::Cuda, short >,
+ Vector< float, Devices::Cuda, short >,
+ Vector< double, Devices::Cuda, short >,
+ Vector< int, Devices::Cuda, int >,
+ Vector< long, Devices::Cuda, int >,
+ Vector< float, Devices::Cuda, int >,
+ Vector< double, Devices::Cuda, int >,
+ Vector< int, Devices::Cuda, long >,
+ Vector< long, Devices::Cuda, long >,
+ Vector< float, Devices::Cuda, long >,
+ Vector< double, Devices::Cuda, long >
+#endif
+#ifdef HAVE_MIC
+ ,
+ Vector< int, Devices::MIC, short >,
+ Vector< long, Devices::MIC, short >,
+ Vector< float, Devices::MIC, short >,
+ Vector< double, Devices::MIC, short >,
+ Vector< int, Devices::MIC, int >,
+ Vector< long, Devices::MIC, int >,
+ Vector< float, Devices::MIC, int >,
+ Vector< double, Devices::MIC, int >,
+ Vector< int, Devices::MIC, long >,
+ Vector< long, Devices::MIC, long >,
+ Vector< float, Devices::MIC, long >,
+ Vector< double, Devices::MIC, long >
+#endif
+>;
+
+TYPED_TEST_CASE( VectorTest, VectorTypes );
+
+
+TYPED_TEST( VectorTest, max )
{
- Containers::Vector< RealType, Device, IndexType > v;
- v. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- v.setElement( i, -2 );
- ASSERT_TRUE( v. sum() == -20.0 );
- for( int i = 0; i < 10; i ++ )
- v.setElement( i, 2 );
- ASSERT_TRUE( v. sum() == 20.0 );
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
-};
+ VectorType v;
+ v.setSize( size );
+ setLinearSequence( v );
+
+ EXPECT_EQ( v.max(), size - 1 );
+ EXPECT_EQ( VectorOperations::getVectorMax( v ), size - 1 );
+}
-TEST( VectorTest, testDifferenceMax )
+TYPED_TEST( VectorTest, min )
{
- Containers::Vector< RealType, Device, IndexType > v1, v2;
- v1. setSize( 10 );
- v2. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- {
- v1.setElement( i, i );
- v2.setElement( i, -i );
- }
- ASSERT_TRUE( v1. differenceMax( v2 ) == 18.0 );
-};
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+
+ VectorType v;
+ v.setSize( size );
+ setLinearSequence( v );
+
+ EXPECT_EQ( v.min(), 0 );
+ EXPECT_EQ( VectorOperations::getVectorMin( v ), 0 );
+}
-TEST( VectorTest, testDifferenceMin )
+TYPED_TEST( VectorTest, absMax )
{
- Containers::Vector< RealType, Device, IndexType > v1, v2;
- v1. setSize( 10 );
- v2. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- {
- v1.setElement( i, i );
- v2.setElement( i, -i );
- }
- ASSERT_TRUE( v1. differenceMin( v2 ) == 0.0 );
-};
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+
+ VectorType v;
+ v.setSize( size );
+ setNegativeLinearSequence( v );
-TEST( VectorTest, testDifferenceAbsMax )
+ EXPECT_EQ( v.absMax(), size - 1 );
+ EXPECT_EQ( VectorOperations::getVectorAbsMax( v ), size - 1 );
+}
+
+TYPED_TEST( VectorTest, absMin )
{
- Containers::Vector< RealType, Device, IndexType > v1, v2;
- v1. setSize( 10 );
- v2. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- {
- v1.setElement( i, -i );
- v2.setElement( i, i );
- }
- ASSERT_TRUE( v1. differenceAbsMax( v2 ) == 18.0 );
-};
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
-TEST( VectorTest, testDifferenceAbsMin )
+ VectorType v;
+ v.setSize( size );
+ setNegativeLinearSequence( v );
+
+ EXPECT_EQ( v.absMin(), 0 );
+ EXPECT_EQ( VectorOperations::getVectorAbsMin( v ), 0 );
+}
+
+TYPED_TEST( VectorTest, lpNorm )
{
- Containers::Vector< RealType, Device, IndexType > v1, v2;
- v1. setSize( 10 );
- v2. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- {
- v1.setElement( i, -i );
- v2.setElement( i, i );
- }
- ASSERT_TRUE( v1. differenceAbsMin( v2 ) == 0.0 );
-};
+ using VectorType = typename TestFixture::VectorType;
+ using RealType = typename VectorType::RealType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+ const RealType epsilon = 64 * std::numeric_limits< RealType >::epsilon();
+
+ VectorType v;
+ v.setSize( size );
+ setConstantSequence( v, 1 );
+
+ const RealType expectedL1norm = size;
+ const RealType expectedL2norm = std::sqrt( size );
+ const RealType expectedL3norm = std::cbrt( size );
+ EXPECT_EQ( v.lpNorm( 1.0 ), expectedL1norm );
+ EXPECT_EQ( v.lpNorm( 2.0 ), expectedL2norm );
+ EXPECT_NEAR( v.lpNorm( 3.0 ), expectedL3norm, epsilon );
+ EXPECT_EQ( VectorOperations::getVectorLpNorm( v, 1.0 ), expectedL1norm );
+ EXPECT_EQ( VectorOperations::getVectorLpNorm( v, 2.0 ), expectedL2norm );
+ EXPECT_NEAR( VectorOperations::getVectorLpNorm( v, 3.0 ), expectedL3norm, epsilon );
+}
-TEST( VectorTest, testDifferenceLpNorm )
+TYPED_TEST( VectorTest, sum )
{
- Containers::Vector< RealType, Device, IndexType > v1, v2;
- v1. setSize( 10 );
- v2. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- {
- v1.setElement( i, -1 );
- v2.setElement( i, 1 );
- }
- ASSERT_TRUE( isSmall( v1.differenceLpNorm( v2, 1.0 ) - 20.0 ) );
- ASSERT_TRUE( isSmall( v1.differenceLpNorm( v2, 2.0 ) - ::sqrt( 40.0 ) ) );
- ASSERT_TRUE( isSmall( v1.differenceLpNorm( v2, 3.0 ) - ::pow( 80.0, 1.0/3.0 ) ) );
-};
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ // this test expect an even size
+ const int size = VECTOR_TEST_SIZE % 2 ? VECTOR_TEST_SIZE - 1 : VECTOR_TEST_SIZE;
+
+ VectorType v;
+ v.setSize( size );
+
+ setConstantSequence( v, 1 );
+ EXPECT_EQ( v.sum(), size );
+ EXPECT_EQ( VectorOperations::getVectorSum( v ), size );
-TEST( VectorTest, testDifferenceSum )
+ setLinearSequence( v );
+ EXPECT_EQ( v.sum(), 0.5 * size * ( size - 1 ) );
+ EXPECT_EQ( VectorOperations::getVectorSum( v ), 0.5 * size * ( size - 1 ) );
+
+ setNegativeLinearSequence( v );
+ EXPECT_EQ( v.sum(), - 0.5 * size * ( size - 1 ) );
+ EXPECT_EQ( VectorOperations::getVectorSum( v ), - 0.5 * size * ( size - 1 ) );
+
+ setOscilatingSequence( v, 1.0 );
+ EXPECT_EQ( v.sum(), 0 );
+ EXPECT_EQ( VectorOperations::getVectorSum( v ), 0 );
+}
+
+TYPED_TEST( VectorTest, differenceMax )
{
- Containers::Vector< RealType, Device, IndexType > v1, v2;
- v1. setSize( 10 );
- v2. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- {
- v1.setElement( i, -1 );
- v2.setElement( i, 1 );
- }
- ASSERT_TRUE( v1. differenceSum( v2 ) == -20.0 );
-};
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
-TEST( VectorTest, testScalarMultiplication )
+ VectorType u, v;
+ u.setSize( size );
+ v.setSize( size );
+ setLinearSequence( u );
+ setConstantSequence( v, size / 2 );
+
+ EXPECT_EQ( u.differenceMax( v ), size - 1 - size / 2 );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceMax( u, v ), size - 1 - size / 2 );
+}
+
+TYPED_TEST( VectorTest, differenceMin )
{
- Containers::Vector< RealType, Device, IndexType > v;
- v. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- v.setElement( i, i );
- v. scalarMultiplication( 5.0 );
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
- for( int i = 0; i < 10; i ++ )
- ASSERT_TRUE( v. getElement( i ) == 5 * i );
-};
+ VectorType u, v;
+ u.setSize( size );
+ v.setSize( size );
+ setLinearSequence( u );
+ setConstantSequence( v, size / 2 );
+
+ EXPECT_EQ( u.differenceMin( v ), - size / 2 );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceMin( u, v ), - size / 2 );
+ EXPECT_EQ( v.differenceMin( u ), size / 2 - size + 1 );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceMin( v, u ), size / 2 - size + 1 );
+}
+
+TYPED_TEST( VectorTest, differenceAbsMax )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ // this test expects an odd size
+ const int size = VECTOR_TEST_SIZE % 2 ? VECTOR_TEST_SIZE : VECTOR_TEST_SIZE - 1;
+
+ VectorType u, v;
+ u.setSize( size );
+ v.setSize( size );
+ setNegativeLinearSequence( u );
+ setConstantSequence( v, - size / 2 );
-TEST( VectorTest, testScalarProduct )
-{
- Containers::Vector< RealType, Device, IndexType > v1, v2;
- v1. setSize( 10 );
- v2. setSize( 10 );
- v1.setElement( 0, -1 );
- v2.setElement( 0, 1 );
- for( int i = 1; i < 10; i ++ )
- {
- v1.setElement( i, v1.getElement( i - 1 ) * -1 );
- v2.setElement( i, v2.getElement( i - 1 ) );
+ EXPECT_EQ( u.differenceAbsMax( v ), size - 1 - size / 2 );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceAbsMax( u, v ), size - 1 - size / 2 );
+}
+
+TYPED_TEST( VectorTest, differenceAbsMin )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+
+ VectorType u, v;
+ u.setSize( size );
+ v.setSize( size );
+ setNegativeLinearSequence( u );
+ setConstantSequence( v, - size / 2 );
+
+ EXPECT_EQ( u.differenceAbsMin( v ), 0 );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceAbsMin( u, v ), 0 );
+ EXPECT_EQ( v.differenceAbsMin( u ), 0 );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceAbsMin( v, u ), 0 );
+}
+
+TYPED_TEST( VectorTest, differenceLpNorm )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using RealType = typename VectorType::RealType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+ const RealType epsilon = 64 * std::numeric_limits< RealType >::epsilon();
+
+ VectorType u, v;
+ u.setSize( size );
+ v.setSize( size );
+ u.setValue( 3.0 );
+ v.setValue( 1.0 );
+
+ const RealType expectedL1norm = 2.0 * size;
+ const RealType expectedL2norm = std::sqrt( 4.0 * size );
+ const RealType expectedL3norm = std::cbrt( 8.0 * size );
+ EXPECT_EQ( u.differenceLpNorm( v, 1.0 ), expectedL1norm );
+ EXPECT_EQ( u.differenceLpNorm( v, 2.0 ), expectedL2norm );
+ EXPECT_NEAR( u.differenceLpNorm( v, 3.0 ), expectedL3norm, epsilon );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceLpNorm( u, v, 1.0 ), expectedL1norm );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceLpNorm( u, v, 2.0 ), expectedL2norm );
+ EXPECT_NEAR( VectorOperations::getVectorDifferenceLpNorm( u, v, 3.0 ), expectedL3norm, epsilon );
+}
+
+TYPED_TEST( VectorTest, differenceSum )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ // this test expect an even size
+ const int size = VECTOR_TEST_SIZE % 2 ? VECTOR_TEST_SIZE - 1 : VECTOR_TEST_SIZE;
+
+ VectorType u, v;
+ u.setSize( size );
+ v.setSize( size );
+ v.setValue( 1.0 );
+
+ setConstantSequence( u, 2 );
+ EXPECT_EQ( u.differenceSum( v ), size );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceSum( u, v ), size );
+
+ setLinearSequence( u );
+ EXPECT_EQ( u.differenceSum( v ), 0.5 * size * ( size - 1 ) - size );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceSum( u, v ), 0.5 * size * ( size - 1 ) - size );
+
+ setNegativeLinearSequence( u );
+ EXPECT_EQ( u.differenceSum( v ), - 0.5 * size * ( size - 1 ) - size );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceSum( u, v ), - 0.5 * size * ( size - 1 ) - size );
+
+ setOscilatingSequence( u, 1.0 );
+ EXPECT_EQ( u.differenceSum( v ), - size );
+ EXPECT_EQ( VectorOperations::getVectorDifferenceSum( u, v ), - size );
+}
+
+TYPED_TEST( VectorTest, scalarMultiplication )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+
+ VectorType u;
+ u.setSize( size );
+
+ typename VectorType::HostType expected;
+ expected.setSize( size );
+ for( int i = 0; i < size; i++ )
+ expected[ i ] = 2.0 * i;
+
+ setLinearSequence( u );
+ VectorOperations::vectorScalarMultiplication( u, 2.0 );
+ EXPECT_EQ( u, expected );
+
+ setLinearSequence( u );
+ u.scalarMultiplication( 2.0 );
+ EXPECT_EQ( u, expected );
+
+ setLinearSequence( u );
+ u *= 2.0;
+ EXPECT_EQ( u, expected );
+}
+
+TYPED_TEST( VectorTest, scalarProduct )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ // this test expects an odd size
+ const int size = VECTOR_TEST_SIZE % 2 ? VECTOR_TEST_SIZE : VECTOR_TEST_SIZE - 1;
+
+ VectorType u, v;
+ u.setSize( size );
+ v.setSize( size );
+ setOscilatingSequence( u, 1.0 );
+ setConstantSequence( v, 1 );
+
+ EXPECT_EQ( u.scalarProduct( v ), 1.0 );
+ EXPECT_EQ( VectorOperations::getScalarProduct( u, v ), 1.0 );
+}
+
+TYPED_TEST( VectorTest, addVector )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+
+ VectorType x, y;
+ x.setSize( size );
+ y.setSize( size );
+
+ typename VectorType::HostType expected1, expected2;
+ expected1.setSize( size );
+ expected2.setSize( size );
+ for( int i = 0; i < size; i++ ) {
+ expected1[ i ] = 2.0 + 3.0 * i;
+ expected2[ i ] = 1.0 + 3.0 * i;
}
- ASSERT_TRUE( v1. scalarProduct( v2 ) == 0.0 );
-};
-TEST( VectorTest, addVectorTest )
+ setConstantSequence( x, 1 );
+ setLinearSequence( y );
+ VectorOperations::addVector( x, y, 3.0, 2.0 );
+ EXPECT_EQ( x, expected1 );
+
+ setConstantSequence( x, 1 );
+ setLinearSequence( y );
+ x.addVector( y, 3.0, 1.0 );
+ EXPECT_EQ( x, expected2 );
+}
+
+TYPED_TEST( VectorTest, addVectors )
{
- Containers::Vector< RealType, Device, IndexType > v1, v2;
- v1. setSize( 10 );
- v2. setSize( 10 );
- for( int i = 0; i < 10; i ++ )
- {
- v1.setElement( i, i );
- v2.setElement( i, 2.0 * i );
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+
+ VectorType x, y, z;
+ x.setSize( size );
+ y.setSize( size );
+ z.setSize( size );
+
+ typename VectorType::HostType expected1, expected2;
+ expected1.setSize( size );
+ expected2.setSize( size );
+ for( int i = 0; i < size; i++ ) {
+ expected1[ i ] = 1.0 + 3.0 * i + 2.0;
+ expected2[ i ] = 2.0 + 3.0 * i + 2.0;
}
- v1. addVector( v2, 2.0 );
- for( int i = 0; i < 10; i ++ )
- ASSERT_TRUE( v1. getElement( i ) == 5.0 * i );
-};
-#endif /* HAVE_GTEST */
+ setConstantSequence( x, 1 );
+ setLinearSequence( y );
+ setConstantSequence( z, 2 );
+ VectorOperations::addVectors( x, y, 3.0, z, 1.0, 1.0 );
+ EXPECT_EQ( x, expected1 );
+
+ setConstantSequence( x, 1 );
+ setLinearSequence( y );
+ setConstantSequence( z, 2 );
+ x.addVectors( y, 3.0, z, 1.0, 2.0 );
+ EXPECT_EQ( x, expected2 );
+}
+
+// TODO: fix the CUDA implementations
+TYPED_TEST( VectorTest, prefixSum )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+
+ VectorType v;
+ v.setSize( size );
+
+ setConstantSequence( v, 1 );
+ v.computePrefixSum();
+ for( int i = 0; i < size; i++ )
+ EXPECT_EQ( v.getElement( i ), i + 1 );
+
+ v.setValue( 0 );
+ v.computePrefixSum();
+ for( int i = 0; i < size; i++ )
+ EXPECT_EQ( v.getElement( i ), 0 );
+
+ setLinearSequence( v );
+ v.computePrefixSum();
+ for( int i = 1; i < size; i++ )
+ EXPECT_EQ( v.getElement( i ) - v.getElement( i - 1 ), i );
+}
+
+// TODO: fix the CUDA implementations
+TYPED_TEST( VectorTest, exclusivePrefixSum )
+{
+ using VectorType = typename TestFixture::VectorType;
+ using VectorOperations = typename TestFixture::VectorOperations;
+ const int size = VECTOR_TEST_SIZE;
+
+ VectorType v;
+ v.setSize( size );
+
+ setConstantSequence( v, 1 );
+ v.computeExclusivePrefixSum();
+ for( int i = 0; i < size; i++ )
+ EXPECT_EQ( v.getElement( i ), i );
+
+ v.setValue( 0 );
+ v.computeExclusivePrefixSum();
+ for( int i = 0; i < size; i++ )
+ EXPECT_EQ( v.getElement( i ), 0 );
+
+ setLinearSequence( v );
+ v.computeExclusivePrefixSum();
+ for( int i = 1; i < size; i++ )
+ EXPECT_EQ( v.getElement( i ) - v.getElement( i - 1 ), i - 1 );
+}
+
+// TODO: test prefix sum with custom begin and end parameters
+
+#endif // HAVE_GTEST
+
+#include "../GtestMissingError.h"
int main( int argc, char* argv[] )
{
#ifdef HAVE_GTEST
::testing::InitGoogleTest( &argc, argv );
return RUN_ALL_TESTS();
#else
- return EXIT_FAILURE;
+ throw GtestMissingError();
#endif
}
diff --git a/src/UnitTests/FileTest.h b/src/UnitTests/FileTest.h
index 82b8b286c6ba576b95569b0ab7fc5d2ec4e1cda8..b1385f4122924185cfa898b569238e0118aa84af 100644
--- a/src/UnitTests/FileTest.h
+++ b/src/UnitTests/FileTest.h
@@ -1,8 +1,8 @@
/***************************************************************************
- tnlFileTester.h - description
+ FileTest.h - description
-------------------
begin : Oct 24, 2010
- copyright : (C) 2010 by Tomas Oberhuber
+ copyright : (C) 2010 by Tomas Oberhuber et al.
email : tomas.oberhuber@fjfi.cvut.cz
***************************************************************************/
@@ -11,60 +11,43 @@
#include <TNL/File.h>
#ifdef HAVE_GTEST
-#include "gtest/gtest.h"
-#endif
-
-#ifdef HAVE_CUDA
-#include <cuda.h>
-#endif
+#include <gtest/gtest.h>
using namespace TNL;
-#ifdef HAVE_GTEST
+TEST( FileTest, CloseEmpty )
+{
+ File file;
+ ASSERT_TRUE( file.close() );
+}
+
TEST( FileTest, WriteAndRead )
{
File file;
- if( ! file. open( String( "test-file.tnl" ), tnlWriteMode ) )
- {
- std::cerr << "Unable to create file test-file.tnl for the testing." << std::endl;
- return;
- }
+ ASSERT_TRUE( file.open( String( "test-file.tnl" ), IOMode::write ) );
+
int intData( 5 );
-#ifdef HAVE_NOT_CXX11
- file. write< int, Devices::Host >( &intData );
-#else
- file. write( &intData );
-#endif
double doubleData[ 3 ] = { 1.0, 2.0, 3.0 };
-#ifdef HAVE_NOT_CXX11
- file. write< double, Devices::Host >( doubleData, 3 );
-#else
- file. write( doubleData, 3 );
-#endif
- if( ! file. close() )
- {
- std::cerr << "Unable to close the file test-file.tnl" << std::endl;
- return;
- }
-
- if( ! file. open( String( "test-file.tnl" ), tnlReadMode ) )
- {
- std::cerr << "Unable to open the file test-file.tnl for the testing." << std::endl;
- return;
- }
+ const double constDoubleData = 3.14;
+ ASSERT_TRUE( file.write( &intData ) );
+ ASSERT_TRUE( file.write( doubleData, 3 ) );
+ ASSERT_TRUE( file.write( &constDoubleData ) );
+ ASSERT_TRUE( file.close() );
+
+ ASSERT_TRUE( file.open( String( "test-file.tnl" ), IOMode::read ) );
int newIntData;
double newDoubleData[ 3 ];
-#ifdef HAVE_NOT_CXX11
- file. read< int, Devices::Host >( &newIntData );
- file. read< double, Devices::Host >( newDoubleData, 3 );
-#else
- file. read( &newIntData, 1 );
- file. read( newDoubleData, 3 );
-#endif
+ double newConstDoubleData;
+ ASSERT_TRUE( file.read( &newIntData, 1 ) );
+ ASSERT_TRUE( file.read( newDoubleData, 3 ) );
+ ASSERT_TRUE( file.read( &newConstDoubleData, 1 ) );
- ASSERT_EQ( newIntData, intData );
+ EXPECT_EQ( newIntData, intData );
for( int i = 0; i < 3; i ++ )
- ASSERT_EQ( newDoubleData[ i ], doubleData[ i ] );
+ EXPECT_EQ( newDoubleData[ i ], doubleData[ i ] );
+ EXPECT_EQ( newConstDoubleData, constDoubleData );
+
+ EXPECT_EQ( std::remove( "test-file.tnl" ), 0 );
};
#ifdef HAVE_CUDA
@@ -72,11 +55,14 @@ TEST( FileTest, WriteAndReadCUDA )
{
int intData( 5 );
float floatData[ 3 ] = { 1.0, 2.0, 3.0 };
+ const double constDoubleData = 3.14;
int* cudaIntData;
float* cudaFloatData;
+ const double* cudaConstDoubleData;
cudaMalloc( ( void** ) &cudaIntData, sizeof( int ) );
cudaMalloc( ( void** ) &cudaFloatData, 3 * sizeof( float ) );
+ cudaMalloc( ( void** ) &cudaConstDoubleData, sizeof( double ) );
cudaMemcpy( cudaIntData,
&intData,
sizeof( int ),
@@ -85,34 +71,38 @@ TEST( FileTest, WriteAndReadCUDA )
floatData,
3 * sizeof( float ),
cudaMemcpyHostToDevice );
+ cudaMemcpy( (void*) cudaConstDoubleData,
+ &constDoubleData,
+ sizeof( double ),
+ cudaMemcpyHostToDevice );
+
File file;
- if( ! file. open( String( "test-file.tnl" ), tnlWriteMode ) )
- {
- std::cerr << "Unable to create file test-file.tnl for the testing." << std::endl;
- return;
- }
-
- file. write< int, Devices::Cuda >( cudaIntData );
- file. write< float, Devices::Cuda, int >( cudaFloatData, 3 );
- if( ! file. close() )
- {
- std::cerr << "Unable to close the file test-file.tnl" << std::endl;
- return;
- }
-
- if( ! file. open( String( "test-file.tnl" ), tnlReadMode ) )
- {
- std::cerr << "Unable to open the file test-file.tnl for the testing." << std::endl;
- return;
- }
+ ASSERT_TRUE( file.open( String( "test-file.tnl" ), IOMode::write ) );
+
+ bool status = file.write< int, Devices::Cuda >( cudaIntData );
+ ASSERT_TRUE( status );
+ status = file.write< float, Devices::Cuda, int >( cudaFloatData, 3 );
+ ASSERT_TRUE( status );
+ status = file.write< const double, Devices::Cuda >( cudaConstDoubleData );
+ ASSERT_TRUE( status );
+ ASSERT_TRUE( file.close() );
+
+ ASSERT_TRUE( file.open( String( "test-file.tnl" ), IOMode::read ) );
int newIntData;
float newFloatData[ 3 ];
+ double newDoubleData;
int* newCudaIntData;
float* newCudaFloatData;
+ double* newCudaDoubleData;
cudaMalloc( ( void** ) &newCudaIntData, sizeof( int ) );
cudaMalloc( ( void** ) &newCudaFloatData, 3 * sizeof( float ) );
- file. read< int, Devices::Cuda >( newCudaIntData, 1 );
- file. read< float, Devices::Cuda, int >( newCudaFloatData, 3 );
+ cudaMalloc( ( void** ) &newCudaDoubleData, sizeof( double ) );
+ status = file.read< int, Devices::Cuda >( newCudaIntData, 1 );
+ ASSERT_TRUE( status );
+ status = file.read< float, Devices::Cuda, int >( newCudaFloatData, 3 );
+ ASSERT_TRUE( status );
+ status = file.read< double, Devices::Cuda >( newCudaDoubleData, 1 );
+ ASSERT_TRUE( status );
cudaMemcpy( &newIntData,
newCudaIntData,
sizeof( int ),
@@ -121,20 +111,28 @@ TEST( FileTest, WriteAndReadCUDA )
newCudaFloatData,
3 * sizeof( float ),
cudaMemcpyDeviceToHost );
+ cudaMemcpy( &newDoubleData,
+ newCudaDoubleData,
+ sizeof( double ),
+ cudaMemcpyDeviceToHost );
- ASSERT_EQ( newIntData, intData );
+ EXPECT_EQ( newIntData, intData );
for( int i = 0; i < 3; i ++ )
- ASSERT_EQ( newFloatData[ i ], floatData[ i ] );
+ EXPECT_EQ( newFloatData[ i ], floatData[ i ] );
+ EXPECT_EQ( newDoubleData, constDoubleData );
+
+ EXPECT_EQ( std::remove( "test-file.tnl" ), 0 );
};
#endif
#endif
+#include "GtestMissingError.h"
int main( int argc, char* argv[] )
{
#ifdef HAVE_GTEST
::testing::InitGoogleTest( &argc, argv );
return RUN_ALL_TESTS();
#else
- return EXIT_FAILURE;
+ throw GtestMissingError();
#endif
}
diff --git a/src/UnitTests/GtestMissingError.h b/src/UnitTests/GtestMissingError.h
new file mode 100644
index 0000000000000000000000000000000000000000..b308a16c8bb02d6afa38f097c48a9242c0512e08
--- /dev/null
+++ b/src/UnitTests/GtestMissingError.h
@@ -0,0 +1,21 @@
+/***************************************************************************
+ GtestMissingError.h - description
+ -------------------
+ begin : Jul 2, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#pragma once
+
+#include <stdexcept>
+
+struct GtestMissingError
+ : public std::runtime_error
+{
+ GtestMissingError()
+ : std::runtime_error( "The GTest library is needed to run the tests." )
+ {}
+};
diff --git a/src/UnitTests/ListTest.cpp b/src/UnitTests/ListTest.cpp
deleted file mode 100644
index 6774d29f578e911d42e919d2dd8ff0b9f81a1610..0000000000000000000000000000000000000000
--- a/src/UnitTests/ListTest.cpp
+++ /dev/null
@@ -1,33 +0,0 @@
-/***************************************************************************
- tnlListTest.cpp - description
- -------------------
- begin : Feb 15, 2014
- copyright : (C) 2014 by Tomas Oberhuber
- email : tomas.oberhuber@fjfi.cvut.cz
- ***************************************************************************/
-
-/* See Copyright Notice in tnl/Copyright */
-
-#ifdef HAVE_GTEST
-#include "gtest/gtest.h"
-#endif
-
-#include <TNL/Containers/List.h>
-
-using namespace TNL;
-
-#ifdef HAVE_GTEST
-#endif
-
-
-int main( int argc, char* argv[] )
-{
-#ifdef HAVE_GTEST
- ::testing::InitGoogleTest( &argc, argv );
- return RUN_ALL_TESTS();
-#else
- return EXIT_FAILURE;
-#endif
-}
-
-
diff --git a/src/UnitTests/Matrices/CMakeLists.txt b/src/UnitTests/Matrices/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..7d1a78b2bb0699f5adb68af4c6eda8d5b9791dbf
--- /dev/null
+++ b/src/UnitTests/Matrices/CMakeLists.txt
@@ -0,0 +1,13 @@
+IF( BUILD_CUDA )
+ CUDA_ADD_EXECUTABLE( SparseMatrixCopyTest${mpiExt}${debugExt} SparseMatrixCopyTest.h SparseMatrixCopyTest.cu OPTIONS ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( SparseMatrixCopyTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
+ELSE( BUILD_CUDA )
+ ADD_EXECUTABLE( SparseMatrixCopyTest${mpiExt}${debugExt} SparseMatrixCopyTest.h SparseMatrixCopyTest.cpp )
+ TARGET_COMPILE_OPTIONS( SparseMatrixCopyTest${mpiExt}${debugExt} PRIVATE ${CXX_TESTS_FLAGS} )
+ TARGET_LINK_LIBRARIES( SparseMatrixCopyTest${mpiExt}${debugExt} ${GTEST_BOTH_LIBRARIES}
+ tnl${mpiExt}${debugExt}-${tnlVersion} )
+ENDIF( BUILD_CUDA )
+
+
+ADD_TEST( SparseMatrixCopyTest${mpiExt}${debugExt} ${EXECUTABLE_OUTPUT_PATH}/SparseMatrixCopyTest${mpiExt}${debugExt} )
diff --git a/src/UnitTests/Containers/VectorOperationsTest.cpp b/src/UnitTests/Matrices/SparseMatrixCopyTest.cpp
similarity index 60%
rename from src/UnitTests/Containers/VectorOperationsTest.cpp
rename to src/UnitTests/Matrices/SparseMatrixCopyTest.cpp
index b5a1b57f466d5d3ee768ab2247fa8989bc294266..30b8f64ecfdbf228856d272a71d3de08980f3987 100644
--- a/src/UnitTests/Containers/VectorOperationsTest.cpp
+++ b/src/UnitTests/Matrices/SparseMatrixCopyTest.cpp
@@ -1,12 +1,11 @@
/***************************************************************************
- VectorOperationsTest.cpp - description
+ SparseMatrixCopyTest.cpp - description
-------------------
- begin : Jul 15, 2013
- copyright : (C) 2013 by Tomas Oberhuber
+ begin : Jun 25, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
email : tomas.oberhuber@fjfi.cvut.cz
***************************************************************************/
/* See Copyright Notice in tnl/Copyright */
-#include "VectorOperationsTest.h"
-
+#include "SparseMatrixCopyTest.h"
diff --git a/src/UnitTests/Containers/StaticArrayTest.cu b/src/UnitTests/Matrices/SparseMatrixCopyTest.cu
similarity index 60%
rename from src/UnitTests/Containers/StaticArrayTest.cu
rename to src/UnitTests/Matrices/SparseMatrixCopyTest.cu
index 00fa118fd23a8e5898f6c8736172cc3a70a4fe46..431fe481c2db1d5b18cfa849e882c0ed836463c1 100644
--- a/src/UnitTests/Containers/StaticArrayTest.cu
+++ b/src/UnitTests/Matrices/SparseMatrixCopyTest.cu
@@ -1,12 +1,11 @@
/***************************************************************************
- StaticArrayTest.cu - description
+ SparseMatrixCopyTest.cu - description
-------------------
- begin : Feb 10, 2014
- copyright : (C) 2014 by Tomas Oberhuber
+ begin : Jun 25, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
email : tomas.oberhuber@fjfi.cvut.cz
***************************************************************************/
/* See Copyright Notice in tnl/Copyright */
-#include "StaticArrayTest.h"
-
+#include "SparseMatrixCopyTest.h"
diff --git a/src/UnitTests/Matrices/SparseMatrixCopyTest.h b/src/UnitTests/Matrices/SparseMatrixCopyTest.h
new file mode 100644
index 0000000000000000000000000000000000000000..a11a8b4442527b371603b8f5d43b70b8e4ff558d
--- /dev/null
+++ b/src/UnitTests/Matrices/SparseMatrixCopyTest.h
@@ -0,0 +1,281 @@
+/***************************************************************************
+ SparseMatrixCopyTest.h - description
+ -------------------
+ begin : Jun 25, 2017
+ copyright : (C) 2017 by Tomas Oberhuber et al.
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#include <TNL/Matrices/CSR.h>
+#include <TNL/Matrices/Ellpack.h>
+#include <TNL/Matrices/SlicedEllpack.h>
+
+using CSR_host = TNL::Matrices::CSR< int, TNL::Devices::Host, int >;
+using CSR_cuda = TNL::Matrices::CSR< int, TNL::Devices::Cuda, int >;
+using E_host = TNL::Matrices::Ellpack< int, TNL::Devices::Host, int >;
+using E_cuda = TNL::Matrices::Ellpack< int, TNL::Devices::Cuda, int >;
+using SE_host = TNL::Matrices::SlicedEllpack< int, TNL::Devices::Host, int, 2 >;
+using SE_cuda = TNL::Matrices::SlicedEllpack< int, TNL::Devices::Cuda, int, 2 >;
+
+#ifdef HAVE_GTEST
+#include <gtest/gtest.h>
+
+/*
+ * Sets up the following 7x6 sparse matrix:
+ *
+ * / 1 2 \
+ * | 3 4 5 |
+ * | 6 7 8 |
+ * | 9 10 11 |
+ * | 12 13 14 |
+ * | 15 16 |
+ * \ 17 /
+ */
+template< typename Matrix >
+void setupMatrix( Matrix& m )
+{
+ const int rows = 7;
+ const int cols = 6;
+ m.reset();
+ m.setDimensions( rows, cols );
+ typename Matrix::CompressedRowLengthsVector rowLengths;
+ rowLengths.setSize( rows );
+ rowLengths.setValue( 3 );
+ rowLengths.setElement( 0 , 4 );
+ rowLengths.setElement( 1, 4 );
+ m.setCompressedRowLengths( rowLengths );
+
+ int value = 1;
+ for( int i = 0; i < rows; i++ )
+ for( int j = 0; j < 3; j++ )
+ if( i + j - 1 >= 0 && i + j - 1 < cols )
+ m.setElement( i, i + j - 1, value++ );
+}
+
+template< typename Matrix >
+void checkMatrix( Matrix& m )
+{
+ ASSERT_EQ( m.getRows(), 7 );
+ ASSERT_EQ( m.getColumns(), 6 );
+
+ EXPECT_EQ( m.getElement( 0, 0 ), 1 );
+ EXPECT_EQ( m.getElement( 0, 1 ), 2 );
+ EXPECT_EQ( m.getElement( 0, 2 ), 0 );
+ EXPECT_EQ( m.getElement( 0, 3 ), 0 );
+ EXPECT_EQ( m.getElement( 0, 4 ), 0 );
+ EXPECT_EQ( m.getElement( 0, 5 ), 0 );
+
+ EXPECT_EQ( m.getElement( 1, 0 ), 3 );
+ EXPECT_EQ( m.getElement( 1, 1 ), 4 );
+ EXPECT_EQ( m.getElement( 1, 2 ), 5 );
+ EXPECT_EQ( m.getElement( 1, 3 ), 0 );
+ EXPECT_EQ( m.getElement( 1, 4 ), 0 );
+ EXPECT_EQ( m.getElement( 1, 5 ), 0 );
+
+ EXPECT_EQ( m.getElement( 2, 0 ), 0 );
+ EXPECT_EQ( m.getElement( 2, 1 ), 6 );
+ EXPECT_EQ( m.getElement( 2, 2 ), 7 );
+ EXPECT_EQ( m.getElement( 2, 3 ), 8 );
+ EXPECT_EQ( m.getElement( 2, 4 ), 0 );
+ EXPECT_EQ( m.getElement( 2, 5 ), 0 );
+
+ EXPECT_EQ( m.getElement( 3, 0 ), 0 );
+ EXPECT_EQ( m.getElement( 3, 1 ), 0 );
+ EXPECT_EQ( m.getElement( 3, 2 ), 9 );
+ EXPECT_EQ( m.getElement( 3, 3 ), 10 );
+ EXPECT_EQ( m.getElement( 3, 4 ), 11 );
+ EXPECT_EQ( m.getElement( 3, 5 ), 0 );
+
+ EXPECT_EQ( m.getElement( 4, 0 ), 0 );
+ EXPECT_EQ( m.getElement( 4, 1 ), 0 );
+ EXPECT_EQ( m.getElement( 4, 2 ), 0 );
+ EXPECT_EQ( m.getElement( 4, 3 ), 12 );
+ EXPECT_EQ( m.getElement( 4, 4 ), 13 );
+ EXPECT_EQ( m.getElement( 4, 5 ), 14 );
+
+ EXPECT_EQ( m.getElement( 5, 0 ), 0 );
+ EXPECT_EQ( m.getElement( 5, 1 ), 0 );
+ EXPECT_EQ( m.getElement( 5, 2 ), 0 );
+ EXPECT_EQ( m.getElement( 5, 3 ), 0 );
+ EXPECT_EQ( m.getElement( 5, 4 ), 15 );
+ EXPECT_EQ( m.getElement( 5, 5 ), 16 );
+
+ EXPECT_EQ( m.getElement( 6, 0 ), 0 );
+ EXPECT_EQ( m.getElement( 6, 1 ), 0 );
+ EXPECT_EQ( m.getElement( 6, 2 ), 0 );
+ EXPECT_EQ( m.getElement( 6, 3 ), 0 );
+ EXPECT_EQ( m.getElement( 6, 4 ), 0 );
+ EXPECT_EQ( m.getElement( 6, 5 ), 17 );
+}
+
+template< typename Matrix1, typename Matrix2 >
+void testCopyAssignment()
+{
+ Matrix1 m1;
+ setupMatrix( m1 );
+ checkMatrix( m1 );
+
+ Matrix2 m2;
+ m2 = m1;
+ checkMatrix( m2 );
+}
+
+template< typename Matrix1, typename Matrix2 >
+void testConversion()
+{
+ Matrix1 m1;
+ setupMatrix( m1 );
+ checkMatrix( m1 );
+
+ Matrix2 m2;
+ TNL::Matrices::copySparseMatrix( m2, m1 );
+ checkMatrix( m2 );
+}
+
+
+TEST( SparseMatrixCopyTest, CSR_HostToHost )
+{
+ testCopyAssignment< CSR_host, CSR_host >();
+}
+
+#ifdef HAVE_CUDA
+TEST( SparseMatrixCopyTest, CSR_HostToCuda )
+{
+ testCopyAssignment< CSR_host, CSR_cuda >();
+}
+
+TEST( SparseMatrixCopyTest, CSR_CudaToHost )
+{
+ testCopyAssignment< CSR_cuda, CSR_host >();
+}
+
+TEST( SparseMatrixCopyTest, CSR_CudaToCuda )
+{
+ testCopyAssignment< CSR_cuda, CSR_cuda >();
+}
+#endif
+
+
+TEST( SparseMatrixCopyTest, Ellpack_HostToHost )
+{
+ testCopyAssignment< E_host, E_host >();
+}
+
+#ifdef HAVE_CUDA
+TEST( SparseMatrixCopyTest, Ellpack_HostToCuda )
+{
+ testCopyAssignment< E_host, E_cuda >();
+}
+
+TEST( SparseMatrixCopyTest, Ellpack_CudaToHost )
+{
+ testCopyAssignment< E_cuda, E_host >();
+}
+
+TEST( SparseMatrixCopyTest, Ellpack_CudaToCuda )
+{
+ testCopyAssignment< E_cuda, E_cuda >();
+}
+#endif
+
+
+TEST( SparseMatrixCopyTest, SlicedEllpack_HostToHost )
+{
+ testCopyAssignment< SE_host, SE_host >();
+}
+
+#ifdef HAVE_CUDA
+TEST( SparseMatrixCopyTest, SlicedEllpack_HostToCuda )
+{
+ testCopyAssignment< SE_host, SE_cuda >();
+}
+
+TEST( SparseMatrixCopyTest, SlicedEllpack_CudaToHost )
+{
+ testCopyAssignment< SE_cuda, SE_host >();
+}
+
+TEST( SparseMatrixCopyTest, SlicedEllpack_CudaToCuda )
+{
+ testCopyAssignment< SE_cuda, SE_cuda >();
+}
+#endif
+
+
+// test conversion between formats
+TEST( SparseMatrixCopyTest, CSR_to_Ellpack_host )
+{
+ testConversion< CSR_host, E_host >();
+}
+
+TEST( SparseMatrixCopyTest, Ellpack_to_CSR_host )
+{
+ testConversion< E_host, CSR_host >();
+}
+
+TEST( SparseMatrixCopyTest, CSR_to_SlicedEllpack_host )
+{
+ testConversion< CSR_host, SE_host >();
+}
+
+TEST( SparseMatrixCopyTest, SlicedEllpack_to_CSR_host )
+{
+ testConversion< SE_host, CSR_host >();
+}
+
+TEST( SparseMatrixCopyTest, Ellpack_to_SlicedEllpack_host )
+{
+ testConversion< E_host, SE_host >();
+}
+
+TEST( SparseMatrixCopyTest, SlicedEllpack_to_Ellpack_host )
+{
+ testConversion< SE_host, E_host >();
+}
+
+#ifdef HAVE_CUDA
+TEST( SparseMatrixCopyTest, CSR_to_Ellpack_cuda )
+{
+ testConversion< CSR_cuda, E_cuda >();
+}
+
+TEST( SparseMatrixCopyTest, Ellpack_to_CSR_cuda )
+{
+ testConversion< E_cuda, CSR_cuda >();
+}
+
+TEST( SparseMatrixCopyTest, CSR_to_SlicedEllpack_cuda )
+{
+ testConversion< CSR_cuda, SE_cuda >();
+}
+
+TEST( SparseMatrixCopyTest, SlicedEllpack_to_CSR_cuda )
+{
+ testConversion< SE_cuda, CSR_cuda >();
+}
+
+TEST( SparseMatrixCopyTest, Ellpack_to_SlicedEllpack_cuda )
+{
+ testConversion< E_cuda, SE_cuda >();
+}
+
+TEST( SparseMatrixCopyTest, SlicedEllpack_to_Ellpack_cuda )
+{
+ testConversion< SE_cuda, E_cuda >();
+}
+#endif
+
+#endif
+
+#include "../GtestMissingError.h"
+int main( int argc, char* argv[] )
+{
+#ifdef HAVE_GTEST
+ ::testing::InitGoogleTest( &argc, argv );
+ return RUN_ALL_TESTS();
+#else
+ throw GtestMissingError();
+#endif
+}
diff --git a/src/UnitTests/ObjectTest.cpp b/src/UnitTests/ObjectTest.cpp
index 3c495846d3de7b62e8c0efb5050970b2fc275fdc..7b9badd8f9d14ce3a9af45249efe9360b07b2e69 100644
--- a/src/UnitTests/ObjectTest.cpp
+++ b/src/UnitTests/ObjectTest.cpp
@@ -14,7 +14,7 @@
#include <TNL/Containers/Array.h>
#ifdef HAVE_GTEST
-#include "gtest/gtest.h"
+#include <gtest/gtest.h>
#endif
using namespace TNL;
@@ -24,11 +24,13 @@ TEST( ObjectTest, SaveAndLoadTest )
{
Object testObject;
File file;
- file.open( "test-file.tnl", tnlWriteMode );
+ file.open( "test-file.tnl", IOMode::write );
ASSERT_TRUE( testObject.save( file ) );
file.close();
- file.open( "test-file.tnl", tnlReadMode );
+ file.open( "test-file.tnl", IOMode::read );
ASSERT_TRUE( testObject.load( file ) );
+
+ EXPECT_EQ( std::remove( "test-file.tnl" ), 0 );
}
TEST( ObjectTest, parseObjectTypeTest )
@@ -104,14 +106,13 @@ TEST( ObjectTest, parseObjectTypeTest )
#endif
+#include "GtestMissingError.h"
int main( int argc, char* argv[] )
{
#ifdef HAVE_GTEST
::testing::InitGoogleTest( &argc, argv );
return RUN_ALL_TESTS();
#else
- return EXIT_FAILURE;
+ throw GtestMissingError();
#endif
}
-
-
diff --git a/src/UnitTests/StringTest.cpp b/src/UnitTests/StringTest.cpp
index 3a0444ed64801402a95cec5a2ba80aa739e05163..6651455589ceeea3681930e1efb621259381fb46 100644
--- a/src/UnitTests/StringTest.cpp
+++ b/src/UnitTests/StringTest.cpp
@@ -9,11 +9,12 @@
/* See Copyright Notice in tnl/Copyright */
#ifdef HAVE_GTEST
-#include "gtest/gtest.h"
+#include <gtest/gtest.h>
#endif
#include <TNL/String.h>
#include <TNL/File.h>
+#include <TNL/Containers/List.h>
using namespace TNL;
@@ -21,7 +22,7 @@ using namespace TNL;
TEST( StringTest, BasicConstructor )
{
String str;
- ASSERT_EQ( strcmp( str. getString(), "" ), 0 );
+ EXPECT_EQ( strcmp( str.getString(), "" ), 0 );
}
TEST( StringTest, ConstructorWithChar )
@@ -31,10 +32,10 @@ TEST( StringTest, ConstructorWithChar )
String str3( "string3xxx", 0, 3 );
String str4( "xxxstring4xxx", 3, 3 );
- ASSERT_EQ( strcmp( str1. getString(), "string1" ), 0 );
- ASSERT_EQ( strcmp( str2. getString(), "string2" ), 0 );
- ASSERT_EQ( strcmp( str3. getString(), "string3" ), 0 );
- ASSERT_EQ( strcmp( str4. getString(), "string4" ), 0 );
+ EXPECT_EQ( strcmp( str1.getString(), "string1" ), 0 );
+ EXPECT_EQ( strcmp( str2.getString(), "string2" ), 0 );
+ EXPECT_EQ( strcmp( str3.getString(), "string3" ), 0 );
+ EXPECT_EQ( strcmp( str4.getString(), "string4" ), 0 );
}
TEST( StringTest, CopyConstructor )
@@ -44,103 +45,255 @@ TEST( StringTest, CopyConstructor )
String string2( string );
String emptyString2( emptyString );
- ASSERT_EQ( strcmp( string2. getString(), "string1" ), 0 );
- ASSERT_EQ( strcmp( emptyString2. getString(), "" ), 0 );
+ EXPECT_EQ( strcmp( string2.getString(), "string1" ), 0 );
+ EXPECT_EQ( strcmp( emptyString2.getString(), "" ), 0 );
}
TEST( StringTest, ConstructorWithNumber )
{
String string1( 10 );
String string2( -5 );
+ String string3( true );
+ String string4( false );
- ASSERT_EQ( strcmp( string1. getString(), "10" ), 0 );
- ASSERT_EQ( strcmp( string2. getString(), "-5" ), 0 );
+ EXPECT_EQ( strcmp( string1.getString(), "10" ), 0 );
+ EXPECT_EQ( strcmp( string2.getString(), "-5" ), 0 );
+ EXPECT_EQ( strcmp( string3.getString(), "true" ), 0 );
+ EXPECT_EQ( strcmp( string4.getString(), "false" ), 0 );
+}
+
+TEST( StringTest, SetSize )
+{
+ String str;
+ str.setSize( 42 );
+ EXPECT_EQ( str.getAllocatedSize(), 256 );
+ // it allocates one more byte for the terminating 0
+ str.setSize( 256 );
+ EXPECT_EQ( str.getAllocatedSize(), 512 );
}
TEST( StringTest, SetString )
{
String str1, str2, str3, str4;
- str1. setString( "string1" );
- str2. setString( "xxxstring2", 3 );
- str3. setString( "string3xxx", 0, 3 );
- str4. setString( "xxxstring4xxx", 3, 3 );
+ str1.setString( "string1" );
+ str2.setString( "xxxstring2", 3 );
+ str3.setString( "string3xxx", 0, 3 );
+ str4.setString( "xxxstring4xxx", 3, 3 );
- ASSERT_EQ( strcmp( str1. getString(), "string1" ), 0 );
- ASSERT_EQ( strcmp( str2. getString(), "string2" ), 0 );
- ASSERT_EQ( strcmp( str3. getString(), "string3" ), 0 );
- ASSERT_EQ( strcmp( str4. getString(), "string4" ), 0 );
+ EXPECT_EQ( strcmp( str1.getString(), "string1" ), 0 );
+ EXPECT_EQ( strcmp( str2.getString(), "string2" ), 0 );
+ EXPECT_EQ( strcmp( str3.getString(), "string3" ), 0 );
+ EXPECT_EQ( strcmp( str4.getString(), "string4" ), 0 );
}
TEST( StringTest, IndexingOperator )
{
String str( "1234567890" );
- ASSERT_EQ( str[ 0 ], '1' );
- ASSERT_EQ( str[ 1 ], '2' );
- ASSERT_EQ( str[ 2 ], '3' );
- ASSERT_EQ( str[ 3 ], '4' );
- ASSERT_EQ( str[ 4 ], '5' );
- ASSERT_EQ( str[ 5 ], '6' );
- ASSERT_EQ( str[ 6 ], '7' );
- ASSERT_EQ( str[ 7 ], '8' );
- ASSERT_EQ( str[ 8 ], '9' );
- ASSERT_EQ( str[ 9 ], '0' );
+ EXPECT_EQ( str[ 0 ], '1' );
+ EXPECT_EQ( str[ 1 ], '2' );
+ EXPECT_EQ( str[ 2 ], '3' );
+ EXPECT_EQ( str[ 3 ], '4' );
+ EXPECT_EQ( str[ 4 ], '5' );
+ EXPECT_EQ( str[ 5 ], '6' );
+ EXPECT_EQ( str[ 6 ], '7' );
+ EXPECT_EQ( str[ 7 ], '8' );
+ EXPECT_EQ( str[ 8 ], '9' );
+ EXPECT_EQ( str[ 9 ], '0' );
}
-TEST( StringTest, AssignmentOperator )
+TEST( StringTest, CStringOperators )
{
- String string1( "string" );
- String string2;
- string2 = string1;
+ // assignment operator
+ String string1;
+ string1 = "string";
+ EXPECT_EQ( strcmp( string1.getString(), "string" ), 0 );
+
+ // addition
+ string1 += "string2";
+ EXPECT_EQ( strcmp( string1.getString(), "stringstring2" ), 0 );
+
+ // addition that forces a new page allocation
+ string1 += " long long long long long long long long long long long long long long"
+ " long long long long long long long long long long long long long long"
+ " long long long long long long long long long long long long long long"
+ " long long long long long long long long long long long long long long";
+ EXPECT_EQ( strcmp( string1.getString(),
+ "stringstring2"
+ " long long long long long long long long long long long long long long"
+ " long long long long long long long long long long long long long long"
+ " long long long long long long long long long long long long long long"
+ " long long long long long long long long long long long long long long" ),
+ 0 );
- ASSERT_EQ( strcmp( string2. getString(), "string" ), 0 );
+ // addition
+ EXPECT_EQ( strcmp( (String( "foo " ) + "bar").getString(), "foo bar" ), 0 );
+ EXPECT_EQ( strcmp( ("foo" + String( " bar" )).getString(), "foo bar" ), 0 );
+
+ // comparison
+ EXPECT_EQ( String( "foo" ), "foo" );
+ EXPECT_NE( String( "bar" ), "foo" );
+ EXPECT_NE( String( "fooo" ), "foo" );
}
-TEST( StringTest, AdditionAssignmentOperator )
+TEST( StringTest, StringOperators )
{
+ // assignment
String string1( "string" );
String string2;
string2 = string1;
- string2 += "string2";
+ EXPECT_EQ( strcmp( string2.getString(), "string" ), 0 );
+
+ // addition
+ string1.setString( "foo " );
+ string1 += String( "bar" );
+ EXPECT_EQ( strcmp( string1.getString(), "foo bar" ), 0 );
+
+ // comparison
+ EXPECT_EQ( String( "foo bar" ), string1 );
+ EXPECT_NE( String( "bar" ), string1 );
+ EXPECT_NE( String( "bar" ), String( "baz" ) );
+ EXPECT_NE( String( "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long" ),
+ String( "short" ) );
+ String string3( "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long" );
+ string3[ 255 ] = 0;
+ EXPECT_EQ( string3,
+ String( "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long long long long long "
+ "long long long long long long long " ) );
+
+ // addition
+ EXPECT_EQ( String( "foo " ) + String( "bar" ), "foo bar" );
+}
+
+TEST( StringTest, SingleCharacterOperators )
+{
+ // assignment
+ String string1;
+ string1 = 'A';
+ EXPECT_EQ( strcmp( string1.getString(), "A" ), 0 );
+
+ // addition of a single character
+ String string2( "string " );
+ string2 += 'A';
+ EXPECT_EQ( strcmp( string2.getString(), "string A" ), 0 );
- ASSERT_EQ( strcmp( string2. getString(), "stringstring2" ), 0 );
+ // addition of a single character that causes new page allocation
+ string2.setString( "long long long long long long long long long long long long long "
+ "long long long long long long long long long long long long long "
+ "long long long long long long long long long long long long long "
+ "long long long long long long long long long long long long " );
+ ASSERT_EQ( string2.getLength(), 255 );
+ string2 += 'B';
+ EXPECT_EQ( strcmp( string2.getString(),
+ "long long long long long long long long long long long long long "
+ "long long long long long long long long long long long long long "
+ "long long long long long long long long long long long long long "
+ "long long long long long long long long long long long long B" ),
+ 0 );
+
+ // addition
+ EXPECT_EQ( strcmp( (String( "A " ) + 'B').getString(), "A B" ), 0 );
+ EXPECT_EQ( strcmp( ('A' + String( " B" )).getString(), "A B" ), 0 );
+
+ // comparison
+ EXPECT_EQ( String( "A" ), 'A' );
+ EXPECT_NE( String( "B" ), 'A' );
+ EXPECT_NE( String( "AB" ), 'A' );
+}
+
+TEST( StringTest, CastToBoolOperator )
+{
+ String string;
+ EXPECT_TRUE( ! string );
+ EXPECT_FALSE( string );
+ string.setString( "foo" );
+ EXPECT_TRUE( string );
+ EXPECT_FALSE( ! string );
+}
+
+TEST( StringTest, replace )
+{
+ EXPECT_EQ( String( "string" ).replace( "ing", "bc" ), "strbc" );
+ EXPECT_EQ( String( "abracadabra" ).replace( "ab", "CAT" ), "CATracadCATra" );
+ EXPECT_EQ( String( "abracadabra" ).replace( "ab", "CAT", 1 ), "CATracadabra" );
}
TEST( StringTest, strip )
{
- EXPECT_EQ( String( "string" ).strip(), String( "string" ) );
- EXPECT_EQ( String( " string" ).strip(), String( "string" ) );
- EXPECT_EQ( String( "string " ).strip(), String( "string" ) );
- EXPECT_EQ( String( " string " ).strip(), String( "string" ) );
- EXPECT_EQ( String( " string1 string2 " ).strip(), String( "string1 string2" ) );
- EXPECT_EQ( String( "" ).strip(), String( "" ) );
- EXPECT_EQ( String( " " ).strip(), String( "" ) );
+ EXPECT_EQ( String( "string" ).strip(), "string" );
+ EXPECT_EQ( String( " string" ).strip(), "string" );
+ EXPECT_EQ( String( "string " ).strip(), "string" );
+ EXPECT_EQ( String( " string " ).strip(), "string" );
+ EXPECT_EQ( String( " string1 string2 " ).strip(), "string1 string2" );
+ EXPECT_EQ( String( "" ).strip(), "" );
+ EXPECT_EQ( String( " " ).strip(), "" );
}
+TEST( StringTest, split )
+{
+ Containers::List< String > list;
+
+ String( "A B C" ).split( list, ' ' );
+ ASSERT_EQ( list.getSize(), 3 );
+ EXPECT_EQ( list[ 0 ], "A" );
+ EXPECT_EQ( list[ 1 ], "B" );
+ EXPECT_EQ( list[ 2 ], "C" );
+
+ String( "abracadabra" ).split( list, 'a' );
+ ASSERT_EQ( list.getSize(), 4 );
+ EXPECT_EQ( list[ 0 ], "br" );
+ EXPECT_EQ( list[ 1 ], "c" );
+ EXPECT_EQ( list[ 2 ], "d" );
+ EXPECT_EQ( list[ 3 ], "br" );
+
+ String( "abracadabra" ).split( list, 'b' );
+ ASSERT_EQ( list.getSize(), 3 );
+ EXPECT_EQ( list[ 0 ], "a" );
+ EXPECT_EQ( list[ 1 ], "racada" );
+ EXPECT_EQ( list[ 2 ], "ra" );
+
+ String( "abracadabra" ).split( list, 'A' );
+ ASSERT_EQ( list.getSize(), 1 );
+ EXPECT_EQ( list[ 0 ], "abracadabra" );
+}
TEST( StringTest, SaveLoad )
{
String str1( "testing-string" );
File file;
- file.open( "test-file.tnl", tnlWriteMode );
+ file.open( "test-file.tnl", IOMode::write );
ASSERT_TRUE( str1.save( file ) );
file.close();
- file.open( "test-file.tnl", tnlReadMode );
+ file.open( "test-file.tnl", IOMode::read );
String str2;
ASSERT_TRUE( str2.load( file ) );
- ASSERT_EQ( str1, str2 );
-};
+ EXPECT_EQ( str1, str2 );
+ EXPECT_EQ( std::remove( "test-file.tnl" ), 0 );
+};
#endif
+#include "GtestMissingError.h"
int main( int argc, char* argv[] )
{
#ifdef HAVE_GTEST
::testing::InitGoogleTest( &argc, argv );
return RUN_ALL_TESTS();
#else
- return EXIT_FAILURE;
+ throw GtestMissingError();
#endif
}
-
diff --git a/src/UnitTests/UniquePointerTest.cpp b/src/UnitTests/UniquePointerTest.cpp
index 579fd2569df3e9b6b68f7e21d426836a7a1abf12..677b3e2bb98a07508ec059b2e6a350375785c9b1 100644
--- a/src/UnitTests/UniquePointerTest.cpp
+++ b/src/UnitTests/UniquePointerTest.cpp
@@ -21,7 +21,7 @@
#include <TNL/Containers/StaticArray.h>
#ifdef HAVE_GTEST
-#include "gtest/gtest.h"
+#include <gtest/gtest.h>
#endif
using namespace TNL;
@@ -47,12 +47,13 @@ TEST( UniquePointerTest, ConstructorTest )
};
#endif
+#include "GtestMissingError.h"
int main( int argc, char* argv[] )
{
#ifdef HAVE_GTEST
::testing::InitGoogleTest( &argc, argv );
return RUN_ALL_TESTS();
#else
- return EXIT_FAILURE;
+ throw GtestMissingError();
#endif
}
diff --git a/src/core/tnlTypeInfo.h b/src/core/tnlTypeInfo.h
deleted file mode 100644
index 5d5282f3c4ed13bcc190446f7c1026ac28663944..0000000000000000000000000000000000000000
--- a/src/core/tnlTypeInfo.h
+++ /dev/null
@@ -1,39 +0,0 @@
-/*
- * File: tnlTypeInfo.h
- * Author: oberhuber
- *
- * Created on July 14, 2016, 3:46 PM
- */
-
-#pragma once
-
-#include <limits>
-
-template< typename Type >
-class tnlTypeInfo
-{
-};
-
-template<>
-class tnlTypeInfo< double >
-{
- public:
-
- typedef double Type;
-
- static __cuda_callable__
- Type getMaxValue() { return DBL_MAX; };
-};
-
-template<>
-class tnlTypeInfo< float >
-{
- public:
-
- typedef float Type;
-
- static __cuda_callable__
- Type getMaxValue() { return FLT_MAX; };
-};
-
-
diff --git a/src/functions/tnlFunctions.h b/src/functions/tnlFunctions.h
deleted file mode 100644
index 326fa6024b37256360fb6653d195310fe6eb9a98..0000000000000000000000000000000000000000
--- a/src/functions/tnlFunctions.h
+++ /dev/null
@@ -1,52 +0,0 @@
-/*
- * File: tnlFunctions.h
- * Author: oberhuber
- *
- * Created on July 11, 2016, 6:01 PM
- */
-
-#pragma once
-
-#include <core/tnlCuda.h>
-
-template< typename Real >
-__cuda_callable__
-Real sign( const Real& x, const Real& smoothing = 0.0 )
-{
- if( x > smoothing )
- return 1.0;
- else if( x < -smoothing )
- return -1.0;
- if( smoothing == 0.0 )
- return 0.0;
- return sin( ( M_PI * x ) / ( 2.0 * smoothing ) );
-}
-
-template< typename Real >
-__cuda_callable__
-Real positivePart( const Real& arg)
-{
- return arg > 0.0 ? arg : 0.0;
-}
-
-template< typename Real >
-__cuda_callable__
-Real negativePart( const Real& arg)
-{
- return arg < 0.0 ? arg : 0.0;
-}
-
-template< typename Real >
-__cuda_callable__
-Real ArgAbsMin( const Real& x, const Real& y )
-{
- return fabs( x ) < fabs( y ) ? x : y;
-}
-
-template< typename Real >
-__cuda_callable__
-Real ArgAbsMax( const Real& x, const Real& y )
-{
- return fabs( x ) > fabs( y ) ? x : y;
-}
-
diff --git a/src/operators/tnlDirichletBoundaryConditions_impl.h b/src/operators/tnlDirichletBoundaryConditions_impl.h
deleted file mode 100644
index 0fed06fc59345dec8ddd799007d8eed3cb9ba8dd..0000000000000000000000000000000000000000
--- a/src/operators/tnlDirichletBoundaryConditions_impl.h
+++ /dev/null
@@ -1,144 +0,0 @@
-/***************************************************************************
- tnlDirichletBoundaryConditions_impl.h - description
- -------------------
- begin : Nov 17, 2014
- copyright : (C) 2014 by oberhuber
- email : tomas.oberhuber@fjfi.cvut.cz
- ***************************************************************************/
-
-/***************************************************************************
- * *
- * This program is free software; you can redistribute it and/or modify *
- * it under the terms of the GNU General Public License as published by *
- * the Free Software Foundation; either version 2 of the License, or *
- * (at your option) any later version. *
- * *
- ***************************************************************************/
-
-#ifndef TNLDIRICHLETBOUNDARYCONDITIONS_IMPL_H_
-#define TNLDIRICHLETBOUNDARYCONDITIONS_IMPL_H_
-
-#include <functions/tnlFunctionAdapter.h>
-
-template< typename Mesh,
- typename Function,
- int MeshEntitiesDimension,
- typename Real,
- typename Index >
-void
-tnlDirichletBoundaryConditions< Mesh, Function, MeshEntitiesDimension, Real, Index >::
-configSetup( tnlConfigDescription& config,
- const String& prefix )
-{
- Function::configSetup( config );
-}
-
-template< typename Mesh,
- typename Function,
- int MeshEntitiesDimension,
- typename Real,
- typename Index >
-bool
-tnlDirichletBoundaryConditions< Mesh, Function, MeshEntitiesDimension, Real, Index >::
-setup( const Config::ParameterContainer& parameters,
- const String& prefix )
-{
- return this->function.setup( parameters );
-}
-
-template< typename Mesh,
- typename Function,
- int MeshEntitiesDimension,
- typename Real,
- typename Index >
-void
-tnlDirichletBoundaryConditions< Mesh, Function, MeshEntitiesDimension, Real, Index >::
-setFunction( const Function& function )
-{
- this->function = function;
-}
-
-template< typename Mesh,
- typename Function,
- int MeshEntitiesDimension,
- typename Real,
- typename Index >
-Function&
-tnlDirichletBoundaryConditions< Mesh, Function, MeshEntitiesDimension, Real, Index >::
-getFunction()
-{
- return this->function;
-}
-
-template< typename Mesh,
- typename Function,
- int MeshEntitiesDimension,
- typename Real,
- typename Index >
-const Function&
-tnlDirichletBoundaryConditions< Mesh, Function, MeshEntitiesDimension, Real, Index >::
-getFunction() const
-{
- return *this->function;
-}
-
-
-template< typename Mesh,
- typename Function,
- int MeshEntitiesDimension,
- typename Real,
- typename Index >
-template< typename EntityType,
- typename MeshFunction >
-__cuda_callable__
-const Real
-tnlDirichletBoundaryConditions< Mesh, Function, MeshEntitiesDimension, Real, Index >::
-operator()( const MeshFunction& u,
- const EntityType& entity,
- const RealType& time ) const
-{
- //static_assert( EntityType::getDimensions() == MeshEntitiesDimension, "Wrong mesh entity dimensions." );
- return tnlFunctionAdapter< MeshType, Function >::template getValue( this->function, entity, time );
-}
-
-template< typename Mesh,
- typename Function,
- int MeshEntitiesDimension,
- typename Real,
- typename Index >
- template< typename EntityType >
-__cuda_callable__
-Index
-tnlDirichletBoundaryConditions< Mesh, Function, MeshEntitiesDimension, Real, Index >::
-getLinearSystemRowLength( const MeshType& mesh,
- const IndexType& index,
- const EntityType& entity ) const
-{
- return 1;
-}
-
-template< typename Mesh,
- typename Function,
- int MeshEntitiesDimension,
- typename Real,
- typename Index >
- template< typename Matrix,
- typename EntityType,
- typename MeshFunction >
-__cuda_callable__
-void
-tnlDirichletBoundaryConditions< Mesh, Function, MeshEntitiesDimension, Real, Index >::
-updateLinearSystem( const RealType& time,
- const MeshType& mesh,
- const IndexType& index,
- const EntityType& entity,
- const MeshFunction& u,
- DofVectorType& b,
- Matrix& matrix ) const
-{
- typename Matrix::MatrixRow matrixRow = matrix.getRow( index );
- matrixRow.setElement( 0, index, 1.0 );
- b[ index ] = tnlFunctionAdapter< MeshType, Function >::getValue( this->function, entity, time );
-}
-
-#endif /* TNLDIRICHLETBOUNDARYCONDITIONS_IMPL_H_ */
diff --git a/src/operators/tnlNeumannReflectionBoundaryConditions.h b/src/operators/tnlNeumannReflectionBoundaryConditions.h
deleted file mode 100644
index 914d50de797f66c5323be7c0b261b193ef3ede71..0000000000000000000000000000000000000000
--- a/src/operators/tnlNeumannReflectionBoundaryConditions.h
+++ /dev/null
@@ -1,126 +0,0 @@
-#pragma once
-
-#include <TNL/Containers/StaticVector.h>
-#include <core/vectors/tnlSharedVector.h>
-#include <TNL/Config/ParameterContainer.h>
-#include <functions/tnlConstantFunction.h>
-
-template< typename Mesh,
- typename Real = typename Mesh::RealType,
- typename Index = typename Mesh::IndexType >
-class tnlNeumannReflectionBoundaryConditions
-{
-
-};
-
-template< typename MeshReal,
- typename Device,
- typename MeshIndex,
- typename Real,
- typename Index >
-class tnlNeumannReflectionBoundaryConditions< tnlGrid< 1, MeshReal, Device, MeshIndex >, Real, Index >
-{
- public:
-
- typedef tnlGrid< 1, MeshReal, Device, MeshIndex > MeshType;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
-
- typedef tnlSharedVector< RealType, DeviceType, IndexType > SharedVector;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
- typedef Containers::StaticVector< 1, RealType > PointType;
- typedef typename MeshType::CoordinatesType CoordinatesType;
-
- bool setup( const Config::ParameterContainer& parameters,
- const String& prefix = "" );
-
- template< typename EntityType,
- typename MeshFunction >
- __cuda_callable__
- const RealType operator()( const MeshFunction& u,
- const EntityType& entity,
- const RealType& time = 0 ) const;
-
- CoordinatesType tmp;
-
-};
-
-template< typename MeshReal,
- typename Device,
- typename MeshIndex,
- typename Real,
- typename Index >
-class tnlNeumannReflectionBoundaryConditions< tnlGrid< 2, MeshReal, Device, MeshIndex >, Real, Index >
-{
- public:
-
- typedef tnlGrid< 2, MeshReal, Device, MeshIndex > MeshType;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
-
- typedef tnlSharedVector< RealType, DeviceType, IndexType > SharedVector;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
- typedef Containers::StaticVector< 2, RealType > PointType;
- typedef typename MeshType::CoordinatesType CoordinatesType;
-
-
- bool setup( const Config::ParameterContainer& parameters,
- const String& prefix = "" );
-
-
- template< typename EntityType,
- typename MeshFunction >
- __cuda_callable__
- const RealType operator()( const MeshFunction& u,
- const EntityType& entity,
- const RealType& time = 0 ) const;
-
-
- CoordinatesType tmp;
-
-
-};
-
-template< typename MeshReal,
- typename Device,
- typename MeshIndex,
- typename Real,
- typename Index >
-class tnlNeumannReflectionBoundaryConditions< tnlGrid< 3, MeshReal, Device, MeshIndex >, Real, Index >
-{
- public:
-
- typedef tnlGrid< 3, MeshReal, Device, MeshIndex > MeshType;
- typedef Real RealType;
- typedef Device DeviceType;
- typedef Index IndexType;
-
- typedef tnlSharedVector< RealType, DeviceType, IndexType > SharedVector;
- typedef tnlVector< RealType, DeviceType, IndexType> DofVectorType;
- typedef Containers::StaticVector< 3, RealType > PointType;
- typedef typename MeshType::CoordinatesType CoordinatesType;
-
-
- bool setup( const Config::ParameterContainer& parameters,
- const String& prefix = "" );
-
-
- template< typename EntityType,
- typename MeshFunction >
- __cuda_callable__
- const RealType operator()( const MeshFunction& u,
- const EntityType& entity,
- const RealType& time = 0 ) const;
-
- private:
-
- CoordinatesType tmp;
-
-
-};
-
-#include <operators/tnlNeumannReflectionBoundaryConditions_impl.h>
-
-#endif /* TNLNEUMANNREFLECTIONBOUNDARYCONDITIONS_H */
diff --git a/src/operators/tnlNeumannReflectionBoundaryConditions_impl.h b/src/operators/tnlNeumannReflectionBoundaryConditions_impl.h
deleted file mode 100644
index d791383520bd267beae784fe640d012ddd3147e0..0000000000000000000000000000000000000000
--- a/src/operators/tnlNeumannReflectionBoundaryConditions_impl.h
+++ /dev/null
@@ -1,171 +0,0 @@
-#pragma once
-
-template< typename MeshReal,
- typename Device,
- typename MeshIndex,
- typename Real,
- typename Index >
-bool
-tnlNeumannReflectionBoundaryConditions< tnlGrid< 1, MeshReal, Device, MeshIndex >, Real, Index >::
-setup( const Config::ParameterContainer& parameters,
- const String& prefix )
-{
- return true;
-}
-
-
-
-template< typename MeshReal,
- typename Device,
- typename MeshIndex,
- typename Real,
- typename Index >
- template< typename EntityType,
- typename MeshFunction >
-__cuda_callable__
-const Real
-tnlNeumannBoundaryConditions< tnlGrid< 1, MeshReal, Device, MeshIndex >, Function, Real, Index >::
-operator()( const MeshFunction& u,
- const EntityType& entity,
- const RealType& time ) const
-
-#ifdef HAVE_CUDA
- __device__ __host__
-#endif
-void tnlNeumannReflectionBoundaryConditions< tnlGrid< 1,MeshReal, Device, MeshIndex >, Real, Index >::
-setBoundaryConditions( const RealType& time,
- const MeshType& mesh,
- const IndexType index,
- const CoordinatesType& coordinates,
- DofVectorType& u,
- DofVectorType& fu )
-{
- tmp = coordinates;
-
- if(coordinates.x() == 0)
- tmp.x() = 1;
- else if(coordinates.x() == mesh. getDimensions().x() - 1)
- tmp.x() = coordinates.x() - 2;
-
- u[ index ] = u[mesh.getCellIndex( tmp )];
-}
-
-
-
-template< typename MeshReal,
- typename Device,
- typename MeshIndex,
- typename Real,
- typename Index >
-bool
-tnlNeumannReflectionBoundaryConditions< tnlGrid< 2, MeshReal, Device, MeshIndex >, Real, Index >::
-setup( const Config::ParameterContainer& parameters,
- const String& prefix )
-{
- return true;
-}
-
-template< typename MeshReal,
- typename Device,
- typename MeshIndex,
- typename Real,
- typename Index >
- template< typename EntityType,
- typename MeshFunction >
-__cuda_callable__
-const Real
-tnlNeumannBoundaryConditions< tnlGrid< 1, MeshReal, Device, MeshIndex >, Function, Real, Index >::
-operator()( const MeshFunction& u,
- const EntityType& entity,
- const RealType& time ) const
-
-#ifdef HAVE_CUDA
- __device__ __host__
-#endif
-void tnlNeumannReflectionBoundaryConditions< tnlGrid< 2, MeshReal, Device, MeshIndex >, Real, Index >::
-setBoundaryConditions( const RealType& time,
- const MeshType& mesh,
- const IndexType index,
- const CoordinatesType& coordinates,
- DofVectorType& u,
- DofVectorType& fu )
-{
-
- tmp = coordinates;
-
- if(coordinates.x() == 0)
- tmp.x() = coordinates.x() + 2;
- else if(coordinates.x() == mesh. getDimensions().x() - 1)
- tmp.x() = coordinates.x() - 2;
-
- if(coordinates.y() == 0)
- tmp.y() = coordinates.y() + 2;
- else if(coordinates.y() == mesh. getDimensions().y() - 1)
- tmp.y() = coordinates.y() - 2;
-
- u[ index ] = u[mesh.getCellIndex( tmp )];
-}
-
-
-template< typename MeshReal,
- typename Device,
- typename MeshIndex,
- typename Real,
- typename Index >
-bool
-tnlNeumannReflectionBoundaryConditions< tnlGrid< 3, MeshReal, Device, MeshIndex >, Real, Index >::
-setup( const Config::ParameterContainer& parameters,
- const String& prefix )
-{
- return true;
-}
-
-template< typename MeshReal,
- typename Device,
- typename MeshIndex,
- typename Real,
- typename Index >
- template< typename EntityType,
- typename MeshFunction >
-__cuda_callable__
-const Real
-tnlNeumannBoundaryConditions< tnlGrid< 1, MeshReal, Device, MeshIndex >, Function, Real, Index >::
-operator()( const MeshFunction& u,
- const EntityType& entity,
- const RealType& time ) const
-
-#ifdef HAVE_CUDA
- __device__ __host__
-#endif
-void tnlNeumannReflectionBoundaryConditions< tnlGrid< 3, MeshReal, Device, MeshIndex >, Real, Index >::
-setBoundaryConditions( const RealType& time,
- const MeshType& mesh,
- const IndexType index,
- const CoordinatesType& coordinates,
- DofVectorType& u,
- DofVectorType& fu )
-{
- tmp = coordinates;
-
- if(coordinates.x() == 0)
- tmp.x() = coordinates.x() + 2;
- else if(coordinates.x() == mesh. getDimensions().x() - 1)
- tmp.x() = coordinates.x() - 2;
-
- if(coordinates.y() == 0)
- tmp.y() = coordinates.y() + 2;
- else if(coordinates.y() == mesh. getDimensions().y() - 1)
- tmp.y() = coordinates.y() - 2;
-
- if(coordinates.z() == 0)
- tmp.z() = coordinates.z() + 2;
- else if(coordinates.z() == mesh. getDimensions().z() - 1)
- tmp.z() = coordinates.z() - 2;
-
- u[ index ] = u[mesh.getCellIndex( tmp )];
-}
-
-
-
-#endif /* TNLNEUMANNREFLECTIONBOUNDARYCONDITIONS_IMPL_H */
-
diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
old mode 100755
new mode 100644
index d813a07bae2fa2484ed5c853f3eb2e8a9d6153d5..21090988a1322a52fdb10006ef413e95db0504a7
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -4,6 +4,7 @@ ADD_SUBDIRECTORY( data )
ADD_SUBDIRECTORY( benchmarks )
#ADD_SUBDIRECTORY( unit-tests )
ADD_SUBDIRECTORY( long-time-unit-tests )
+ADD_SUBDIRECTORY( mic )
ADD_SUBDIRECTORY( mpi )
-unset( ENABLE_CODECOVERAGE )
\ No newline at end of file
+unset( ENABLE_CODECOVERAGE )
diff --git a/tests/benchmarks/CMakeLists.txt b/tests/benchmarks/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/tests/benchmarks/array-operations.h b/tests/benchmarks/array-operations.h
index a7c1513e47f122adbc7fd334c2535b4d9cb75777..504dcc1da03a91fa87af913008f9355579e62930 100644
--- a/tests/benchmarks/array-operations.h
+++ b/tests/benchmarks/array-operations.h
@@ -36,21 +36,12 @@ benchmarkArrayOperations( Benchmark & benchmark,
HostArray hostArray, hostArray2;
CudaArray deviceArray, deviceArray2;
- if( ! hostArray.setSize( size ) ||
- ! hostArray2.setSize( size )
+ hostArray.setSize( size );
+ hostArray2.setSize( size );
#ifdef HAVE_CUDA
- ||
- ! deviceArray.setSize( size ) ||
- ! deviceArray2.setSize( size )
+ deviceArray.setSize( size );
+ deviceArray2.setSize( size );
#endif
- )
-
- {
- const char* msg = "error: allocation of arrays failed";
- std::cerr << msg << std::endl;
- benchmark.addErrorMessage( msg );
- return false;
- }
Real resultHost, resultDevice;
diff --git a/tests/benchmarks/heat-equation-benchmark/BenchmarkLaplace_impl.h b/tests/benchmarks/heat-equation-benchmark/BenchmarkLaplace_impl.h
index 8c67951c9093d670f310753306e6eea04a9a4858..4e260b0f711851d114f29122e8c02ba299ffcb72 100644
--- a/tests/benchmarks/heat-equation-benchmark/BenchmarkLaplace_impl.h
+++ b/tests/benchmarks/heat-equation-benchmark/BenchmarkLaplace_impl.h
@@ -37,14 +37,14 @@ operator()( const MeshFunction& u,
* The following example is the Laplace operator approximated
* by the Finite difference method.
*/
- static_assert( MeshEntity::entityDimension == 1, "Wrong mesh entity dimension." );
+ static_assert( MeshEntity::getEntityDimension() == 1, "Wrong mesh entity dimension." );
static_assert( MeshFunction::getEntitiesDimension() == 1, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxSquareInverse = entity.getMesh().template getSpaceStepsProducts< -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
return ( u[ west ] - 2.0 * u[ center ] + u[ east ] ) * hxSquareInverse;
}
@@ -95,11 +95,11 @@ setMatrixElements( const RealType& time,
* by the Finite difference method.
*/
- const typename MeshEntity::template NeighbourEntities< 1 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 1 >& neighborEntities = entity.getNeighborEntities();
const RealType& lambdaX = tau * entity.getMesh().template getSpaceStepsProducts< -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1 >();
matrixRow.setElement( 0, west, - lambdaX );
matrixRow.setElement( 1, center, 2.0 * lambdaX );
matrixRow.setElement( 2, east, - lambdaX );
@@ -141,17 +141,17 @@ operator()( const MeshFunction& u,
* The following example is the Laplace operator approximated
* by the Finite difference method.
*/
- /*static_assert( MeshEntity::entityDimension == 2, "Wrong mesh entity dimension." );
+ /*static_assert( MeshEntity::getEntityDimension() == 2, "Wrong mesh entity dimension." );
static_assert( MeshFunction::getEntitiesDimension() == 2, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxSquareInverse = entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const RealType& hySquareInverse = entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >(); */
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >(); */
const IndexType& xSize = entity.getMesh().getDimensions().x();
const IndexType& c = entity.getIndex();
@@ -208,14 +208,14 @@ setMatrixElements( const RealType& time,
* by the Finite difference method.
*/
- const typename MeshEntity::template NeighbourEntities< 2 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 2 >& neighborEntities = entity.getNeighborEntities();
const RealType& lambdaX = tau * entity.getMesh().template getSpaceStepsProducts< -2, 0 >();
const RealType& lambdaY = tau * entity.getMesh().template getSpaceStepsProducts< 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1 >();
matrixRow.setElement( 0, south, -lambdaY );
matrixRow.setElement( 1, west, -lambdaX );
matrixRow.setElement( 2, center, 2.0 * ( lambdaX + lambdaY ) );
@@ -259,20 +259,20 @@ operator()( const MeshFunction& u,
* The following example is the Laplace operator approximated
* by the Finite difference method.
*/
- static_assert( MeshEntity::entityDimension == 3, "Wrong mesh entity dimension." );
+ static_assert( MeshEntity::getEntityDimension() == 3, "Wrong mesh entity dimension." );
static_assert( MeshFunction::getEntitiesDimension() == 3, "Wrong preimage function" );
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& hxSquareInverse = entity.getMesh().template getSpaceStepsProducts< -2, 0, 0 >();
const RealType& hySquareInverse = entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const RealType& hzSquareInverse = entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
return ( u[ west ] - 2.0 * u[ center ] + u[ east ] ) * hxSquareInverse +
( u[ south ] - 2.0 * u[ center ] + u[ north ] ) * hySquareInverse +
( u[ up ] - 2.0 * u[ center ] + u[ down ] ) * hzSquareInverse;
@@ -325,17 +325,17 @@ setMatrixElements( const RealType& time,
* by the Finite difference method.
*/
- const typename MeshEntity::template NeighbourEntities< 3 >& neighbourEntities = entity.getNeighbourEntities();
+ const typename MeshEntity::template NeighborEntities< 3 >& neighborEntities = entity.getNeighborEntities();
const RealType& lambdaX = tau * entity.getMesh().template getSpaceStepsProducts< -2, 0, 0 >();
const RealType& lambdaY = tau * entity.getMesh().template getSpaceStepsProducts< 0, -2, 0 >();
const RealType& lambdaZ = tau * entity.getMesh().template getSpaceStepsProducts< 0, 0, -2 >();
const IndexType& center = entity.getIndex();
- const IndexType& east = neighbourEntities.template getEntityIndex< 1, 0, 0 >();
- const IndexType& west = neighbourEntities.template getEntityIndex< -1, 0, 0 >();
- const IndexType& north = neighbourEntities.template getEntityIndex< 0, 1, 0 >();
- const IndexType& south = neighbourEntities.template getEntityIndex< 0, -1, 0 >();
- const IndexType& up = neighbourEntities.template getEntityIndex< 0, 0, 1 >();
- const IndexType& down = neighbourEntities.template getEntityIndex< 0, 0, -1 >();
+ const IndexType& east = neighborEntities.template getEntityIndex< 1, 0, 0 >();
+ const IndexType& west = neighborEntities.template getEntityIndex< -1, 0, 0 >();
+ const IndexType& north = neighborEntities.template getEntityIndex< 0, 1, 0 >();
+ const IndexType& south = neighborEntities.template getEntityIndex< 0, -1, 0 >();
+ const IndexType& up = neighborEntities.template getEntityIndex< 0, 0, 1 >();
+ const IndexType& down = neighborEntities.template getEntityIndex< 0, 0, -1 >();
matrixRow.setElement( 0, down, -lambdaZ );
matrixRow.setElement( 1, south, -lambdaY );
matrixRow.setElement( 2, west, -lambdaX );
diff --git a/tests/benchmarks/heat-equation-benchmark/HeatEquationBenchmarkProblem_impl.h b/tests/benchmarks/heat-equation-benchmark/HeatEquationBenchmarkProblem_impl.h
index 07e78a73abb8daf9794d49d293afba0e6b59deb6..573eab7885f83c8ffb2f26a6a4438b3129258099 100644
--- a/tests/benchmarks/heat-equation-benchmark/HeatEquationBenchmarkProblem_impl.h
+++ b/tests/benchmarks/heat-equation-benchmark/HeatEquationBenchmarkProblem_impl.h
@@ -459,7 +459,7 @@ getExplicitUpdate( const RealType& time,
}
if( this->cudaKernelType == "templated-compact" )
{
- typedef typename MeshType::MeshEntity< 2 > CellType;
+ typedef typename MeshType::EntityType< 2 > CellType;
//typedef typename MeshType::Cell CellType;
//std::cerr << "Size of entity is ... " << sizeof( TestEntity< MeshType > ) << " vs. " << sizeof( CellType ) << std::endl;
typedef typename CellType::CoordinatesType CoordinatesType;
@@ -494,7 +494,7 @@ getExplicitUpdate( const RealType& time,
gridXIdx,
gridYIdx );
cudaThreadSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
//std::cerr << "Computing the heat equation ..." << std::endl;
for( IndexType gridYIdx = 0; gridYIdx < cudaYGrids; gridYIdx ++ )
@@ -514,7 +514,7 @@ getExplicitUpdate( const RealType& time,
gridXIdx,
gridYIdx );
cudaThreadSynchronize();
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
#endif
if( this->cudaKernelType == "templated" )
diff --git a/tests/benchmarks/heat-equation-benchmark/HeatEquationBenchmarkRhs.h b/tests/benchmarks/heat-equation-benchmark/HeatEquationBenchmarkRhs.h
index 9ba4c06c944e0906399fe3079dc98e2fc2ddbc3f..dd4f33bc2225fe51796654b2b473b62fe4ea4e06 100644
--- a/tests/benchmarks/heat-equation-benchmark/HeatEquationBenchmarkRhs.h
+++ b/tests/benchmarks/heat-equation-benchmark/HeatEquationBenchmarkRhs.h
@@ -2,7 +2,7 @@
#define HeatEquationBenchmarkRHS_H_
#include <TNL/Functions/Domain.h>
template< typename Mesh, typename Real >class HeatEquationBenchmarkRhs
- : public Functions::Domain< Mesh::meshDimension, Functions::MeshDomain >
+ : public Functions::Domain< Mesh::getMeshDimension(), Functions::MeshDomain >
{
public:
diff --git a/tests/benchmarks/heat-equation-benchmark/TestGridEntity.h b/tests/benchmarks/heat-equation-benchmark/TestGridEntity.h
index afc7e5e6f7931a2aaf1b6d3cb40f378ed600f21a..3492b219807f4650ed665b2ee57c77754f5934f1 100644
--- a/tests/benchmarks/heat-equation-benchmark/TestGridEntity.h
+++ b/tests/benchmarks/heat-equation-benchmark/TestGridEntity.h
@@ -18,12 +18,12 @@
#pragma once
template< typename GridEntity >
-class TestNeighbourGridEntitiesStorage
+class TestNeighborGridEntitiesStorage
{
public:
__cuda_callable__
- TestNeighbourGridEntitiesStorage( const GridEntity& entity )
+ TestNeighborGridEntitiesStorage( const GridEntity& entity )
: entity( entity )
{}
@@ -79,13 +79,13 @@ class TestGridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension
typedef Containers::StaticVector< meshDimension, IndexType > EntityOrientationType;
typedef Containers::StaticVector< meshDimension, IndexType > EntityBasisType;
typedef TestGridEntity< GridType, entityDimension > ThisType;
- typedef TestNeighbourGridEntitiesStorage< ThisType > NeighbourGridEntitiesStorageType;
+ typedef TestNeighborGridEntitiesStorage< ThisType > NeighborGridEntitiesStorageType;
__cuda_callable__ inline
TestGridEntity( const GridType& grid )
: grid( grid ),
/*entityIndex( -1 ),*/
- neighbourEntitiesStorage( *this )
+ neighborEntitiesStorage( *this )
{
}
@@ -98,7 +98,7 @@ class TestGridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension
: grid( grid ),
/*entityIndex( -1 ),
coordinates( coordinates ),*/
- neighbourEntitiesStorage( *this )
+ neighborEntitiesStorage( *this )
{
}
@@ -116,7 +116,7 @@ class TestGridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimension
EntityBasisType basis;
- NeighbourGridEntitiesStorageType neighbourEntitiesStorage;
+ NeighborGridEntitiesStorageType neighborEntitiesStorage;
};
diff --git a/tests/benchmarks/heat-equation-benchmark/tnlTestGrid2D.h b/tests/benchmarks/heat-equation-benchmark/tnlTestGrid2D.h
index d54da5db86516dee835a23ead5e672b9e52253f4..c10cec6ee9fa2a2f5762bcac0c80607c2a3d8326 100644
--- a/tests/benchmarks/heat-equation-benchmark/tnlTestGrid2D.h
+++ b/tests/benchmarks/heat-equation-benchmark/tnlTestGrid2D.h
@@ -15,7 +15,7 @@
#include <core/tnlObject.h>
#include <core/Devices::Host.h>
#include <TNL/Containers/StaticVector.h>
-#include <core/vectors/tnlVector.h>
+#include <TNL/Containers/Vector.h>
template< int Dimension,
typename Real = double,
@@ -32,12 +32,12 @@ class Grid : public tnlObject
#include <mesh/grids/GridEntityTopology.h>
#include <mesh/grids/GridEntityGetter.h>
#include <mesh/grids/GridEntityConfig.h>
-#include <mesh/grids/NeighbourGridEntityGetter.h>
+#include <mesh/grids/NeighborGridEntityGetter.h>
#include <core/tnlLogger.h>
// TODO: remove this
//#include <../tests/benchmarks/heat-equation-benchmark/tnlTestGridEntity.h>
-//#include <../tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntityGetter2D_impl.h>
+//#include <../tests/benchmarks/heat-equation-benchmark/tnlTestNeighborGridEntityGetter2D_impl.h>
/////
template< typename Real,
@@ -192,7 +192,7 @@ class Meshes::Grid< 2, Real, Device, Index > : public tnlObject
#include <core/tnlTNL_ASSERT.h>
#include <mesh/GnuplotWriter.h>
#include <mesh/grids/GridEntityGetter_impl.h>
-#include <mesh/grids/NeighbourGridEntityGetter2D_impl.h>
+#include <mesh/grids/NeighborGridEntityGetter2D_impl.h>
#include <mesh/grids/GridEntityMeasureGetter.h>
using namespace std;
@@ -214,11 +214,11 @@ template< typename Real,
typename Index >
String Meshes::Grid< 2, Real, Device, Index > :: getType()
{
- return tnlString( "Meshes::Grid< " ) +
- tnlString( getMeshDimension() ) + ", " +
- tnlString( ::getType< RealType >() ) + ", " +
- tnlString( Device :: getDeviceType() ) + ", " +
- tnlString( ::getType< IndexType >() ) + " >";
+ return TNL::String( "Meshes::Grid< " ) +
+ TNL::String( getMeshDimension() ) + ", " +
+ TNL::String( ::getType< RealType >() ) + ", " +
+ TNL::String( Device :: getDeviceType() ) + ", " +
+ TNL::String( ::getType< IndexType >() ) + " >";
}
template< typename Real,
@@ -979,11 +979,11 @@ template< typename Real,
typename Index >
String Meshes::Grid< 2, Real, Device, Index > :: getType()
{
- return tnlString( "Meshes::Grid< " ) +
- tnlString( getMeshDimension() ) + ", " +
- tnlString( ::getType< RealType >() ) + ", " +
- tnlString( Device :: getDeviceType() ) + ", " +
- tnlString( ::getType< IndexType >() ) + " >";
+ return TNL::String( "Meshes::Grid< " ) +
+ TNL::String( getMeshDimension() ) + ", " +
+ TNL::String( ::getType< RealType >() ) + ", " +
+ TNL::String( Device :: getDeviceType() ) + ", " +
+ TNL::String( ::getType< IndexType >() ) + " >";
}
template< typename Real,
diff --git a/tests/benchmarks/heat-equation-benchmark/tnlTestGridEntity.h b/tests/benchmarks/heat-equation-benchmark/tnlTestGridEntity.h
index d7841576e1e55be8d28c692ac39b1b6fbb7ffa47..aa8bd8d057309b1cd48fe38d71eab7886ccc0d7c 100644
--- a/tests/benchmarks/heat-equation-benchmark/tnlTestGridEntity.h
+++ b/tests/benchmarks/heat-equation-benchmark/tnlTestGridEntity.h
@@ -53,25 +53,25 @@ class tnlTestGridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimensi
constexpr static int getDimension() { return meshDimension; };
- typedef tnlStaticVector< meshDimension, IndexType > EntityOrientationType;
- typedef tnlStaticVector< meshDimension, IndexType > EntityBasisType;
+ typedef TNL::Containers::StaticVector< meshDimension, IndexType > EntityOrientationType;
+ typedef TNL::Containers::StaticVector< meshDimension, IndexType > EntityBasisType;
typedef tnlTestGridEntity< GridType, entityDimension, Config > ThisType;
- //typedef tnlTestNeighbourGridEntitiesStorage< ThisType > NeighbourGridEntitiesStorageType;
+ //typedef tnlTestNeighborGridEntitiesStorage< ThisType > NeighborGridEntitiesStorageType;
- /*template< int NeighbourEntityDimension = entityDimension >
- using NeighbourEntities =
- tnlTestNeighbourGridEntityGetter<
+ /*template< int NeighborEntityDimension = entityDimension >
+ using NeighborEntities =
+ tnlTestNeighborGridEntityGetter<
tnlTestGridEntity< Meshes::Grid< Dimension, Real, Device, Index >,
entityDimension,
Config >,
- NeighbourEntityDimension >;*/
+ NeighborEntityDimension >;*/
__cuda_callable__ inline
tnlTestGridEntity( const GridType& grid )
: grid( grid ),
entityIndex( -1 )/*,
- neighbourEntitiesStorage( *this )*/
+ neighborEntitiesStorage( *this )*/
{
this->coordinates = CoordinatesType( ( Index ) 0 );
this->orientation = EntityOrientationType( ( Index ) 0 );
@@ -87,7 +87,7 @@ class tnlTestGridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimensi
: grid( grid ),
entityIndex( -1 ),
coordinates( coordinates )/*,
- neighbourEntitiesStorage( *this )*/
+ neighborEntitiesStorage( *this )*/
{
this->orientation = EntityOrientationType( ( Index ) 0 );
this->basis = EntityBasisType( ( Index ) 1 );
@@ -107,7 +107,7 @@ class tnlTestGridEntity< Meshes::Grid< Dimension, Real, Device, Index >, Dimensi
EntityBasisType basis;
- //NeighbourGridEntitiesStorageType neighbourEntitiesStorage;
+ //NeighborGridEntitiesStorageType neighborEntitiesStorage;
};
diff --git a/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntitiesStorage.h b/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntitiesStorage.h
index 4d58008765aa4c8e2f613caa6417590248eddc76..f066ef3e3d2ea95531ddbdcbf27714d2f551875c 100644
--- a/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntitiesStorage.h
+++ b/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntitiesStorage.h
@@ -1,5 +1,5 @@
/***************************************************************************
- tnlTestNeighbourGridEntitiesStorage.h - description
+ tnlTestNeighborGridEntitiesStorage.h - description
-------------------
begin : Dec 18, 2015
copyright : (C) 2015 by Tomas Oberhuber
@@ -20,23 +20,23 @@
#include <core/tnlCuda.h>
#include <mesh/MeshDimensionTag.h>
-#include "tnlTestNeighbourGridEntityGetter.h"
+#include "tnlTestNeighborGridEntityGetter.h"
template< typename GridEntity,
- int NeighbourEntityDimension >
-class tnlTestNeighbourGridEntityLayer
-: public tnlTestNeighbourGridEntityLayer< GridEntity, NeighbourEntityDimension - 1 >
+ int NeighborEntityDimension >
+class tnlTestNeighborGridEntityLayer
+: public tnlTestNeighborGridEntityLayer< GridEntity, NeighborEntityDimension - 1 >
{
public:
- typedef tnlTestNeighbourGridEntityLayer< GridEntity, NeighbourEntityDimension - 1 > BaseType;
- typedef tnlTestNeighbourGridEntityGetter< GridEntity, NeighbourEntityDimension > NeighbourEntityGetterType;
+ typedef tnlTestNeighborGridEntityLayer< GridEntity, NeighborEntityDimension - 1 > BaseType;
+ typedef tnlTestNeighborGridEntityGetter< GridEntity, NeighborEntityDimension > NeighborEntityGetterType;
- using BaseType::getNeighbourEntities;
+ using BaseType::getNeighborEntities;
__cuda_callable__
- tnlTestNeighbourGridEntityLayer( const GridEntity& entity )
- : neighbourEntities( entity ),
+ tnlTestNeighborGridEntityLayer( const GridEntity& entity )
+ : neighborEntities( entity ),
BaseType( entity )
{}
@@ -45,51 +45,51 @@ class tnlTestNeighbourGridEntityLayer
const typename GridEntity::GridType::IndexType& entityIndex )
{
BaseType::refresh( grid, entityIndex );
- neighbourEntities.refresh( grid, entityIndex );
+ neighborEntities.refresh( grid, entityIndex );
};
protected:
- NeighbourEntityGetterType neighbourEntities;
+ NeighborEntityGetterType neighborEntities;
};
template< typename GridEntity >
-class tnlTestNeighbourGridEntityLayer< GridEntity, 0 >
+class tnlTestNeighborGridEntityLayer< GridEntity, 0 >
{
public:
- typedef tnlTestNeighbourGridEntityGetter< GridEntity, 0 > NeighbourEntityGetterType;
+ typedef tnlTestNeighborGridEntityGetter< GridEntity, 0 > NeighborEntityGetterType;
__cuda_callable__
- tnlTestNeighbourGridEntityLayer( const GridEntity& entity )
- : neighbourEntities( entity )
+ tnlTestNeighborGridEntityLayer( const GridEntity& entity )
+ : neighborEntities( entity )
{}
__cuda_callable__
void refresh( const typename GridEntity::GridType& grid,
const typename GridEntity::GridType::IndexType& entityIndex )
{
- neighbourEntities.refresh( grid, entityIndex );
+ neighborEntities.refresh( grid, entityIndex );
};
protected:
- NeighbourEntityGetterType neighbourEntities;
+ NeighborEntityGetterType neighborEntities;
};
template< typename GridEntity >
-class tnlTestNeighbourGridEntitiesStorage
-: public tnlTestNeighbourGridEntityLayer< GridEntity, GridEntity::meshDimension >
+class tnlTestNeighborGridEntitiesStorage
+: public tnlTestNeighborGridEntityLayer< GridEntity, GridEntity::meshDimension >
{
- typedef tnlTestNeighbourGridEntityLayer< GridEntity, GridEntity::meshDimension > BaseType;
+ typedef tnlTestNeighborGridEntityLayer< GridEntity, GridEntity::meshDimension > BaseType;
public:
- using BaseType::getNeighbourEntities;
+ using BaseType::getNeighborEntities;
__cuda_callable__
- tnlTestNeighbourGridEntitiesStorage( const GridEntity& entity )
+ tnlTestNeighborGridEntitiesStorage( const GridEntity& entity )
: BaseType( entity )
{}
diff --git a/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntityGetter.h b/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntityGetter.h
index 2aa6f25bb65be792da79252969b4583cf13269e7..521c58ff56902084a602e33bf131808e33d54862 100644
--- a/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntityGetter.h
+++ b/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntityGetter.h
@@ -1,5 +1,5 @@
/***************************************************************************
- tnlTestNeighbourGridEntityGetter.h - description
+ tnlTestNeighborGridEntityGetter.h - description
-------------------
begin : Nov 23, 2015
copyright : (C) 2015 by Tomas Oberhuber
@@ -21,17 +21,17 @@
template< typename GridEntity,
- int NeighbourEntityDimension,
+ int NeighborEntityDimension,
typename EntityStencilTag =
- GridEntityStencilStorageTag< GridEntity::ConfigType::template neighbourEntityStorage< GridEntity >( NeighbourEntityDimension ) > >
-class tnlTestNeighbourGridEntityGetter
+ GridEntityStencilStorageTag< GridEntity::ConfigType::template neighborEntityStorage< GridEntity >( NeighborEntityDimension ) > >
+class tnlTestNeighborGridEntityGetter
{
public:
// TODO: not all specializations are implemented yet
__cuda_callable__
- tnlTestNeighbourGridEntityGetter( const GridEntity& entity )
+ tnlTestNeighborGridEntityGetter( const GridEntity& entity )
{
//tnlTNL_ASSERT( false, );
};
@@ -50,7 +50,7 @@ template< typename Real,
typename Index,
typename Config,
typename StencilStorage >
-class tnlTestNeighbourGridEntityGetter<
+class tnlTestNeighborGridEntityGetter<
GridEntity< Meshes::Grid< 2, Real, Device, Index >, 2, Config >,
2,
StencilStorage >
@@ -58,17 +58,17 @@ class tnlTestNeighbourGridEntityGetter<
public:
static const int EntityDimension = 2;
- static const int NeighbourEntityDimension = 2;
+ static const int NeighborEntityDimension = 2;
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetter;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetter;
__cuda_callable__ inline
- tnlTestNeighbourGridEntityGetter( const GridEntityType& entity )
+ tnlTestNeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
@@ -79,7 +79,7 @@ class tnlTestNeighbourGridEntityGetter<
const GridEntityType& entity;
- //tnlTestNeighbourGridEntityGetter(){};
+ //tnlTestNeighborGridEntityGetter(){};
};
diff --git a/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntityGetter2D_impl.h b/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntityGetter2D_impl.h
index cce9a7ec5a63111b1a0a06b6d4d0288c6de95312..23d35fdae1038c6ec82e65fd69febfb1b0c6779a 100644
--- a/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntityGetter2D_impl.h
+++ b/tests/benchmarks/heat-equation-benchmark/tnlTestNeighbourGridEntityGetter2D_impl.h
@@ -1,5 +1,5 @@
/***************************************************************************
- tnlTestNeighbourGridEntityGetter2D_impl.h - description
+ tnlTestNeighborGridEntityGetter2D_impl.h - description
-------------------
begin : Nov 23, 2015
copyright : (C) 2015 by Tomas Oberhuber
@@ -17,13 +17,13 @@
#pragma once
-#include "tnlTestNeighbourGridEntityGetter.h"
+#include "tnlTestNeighborGridEntityGetter.h"
#include <mesh/grids/Grid2D.h>
#include <core/tnlStaticFor.h>
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stencil Storage |
+ * | EntityDimenions | NeighborEntityDimension | Stencil Storage |
* +-----------------+---------------------------+-------------------+
* | 2 | 2 | No specialization |
* +-----------------+---------------------------+-------------------+
@@ -33,7 +33,7 @@ template< typename Real,
typename Index,
typename Config,
typename StencilStorage >
-class tnlTestNeighbourGridEntityGetter<
+class tnlTestNeighborGridEntityGetter<
GridEntity< Meshes::Grid< 2, Real, Device, Index >, 2, Config >,
2,
StencilStorage >
@@ -41,17 +41,17 @@ class tnlTestNeighbourGridEntityGetter<
public:
static const int EntityDimension = 2;
- static const int NeighbourEntityDimension = 2;
+ static const int NeighborEntityDimension = 2;
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetter;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetter;
__cuda_callable__ inline
- tnlTestNeighbourGridEntityGetter( const GridEntityType& entity )
+ tnlTestNeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
@@ -62,12 +62,12 @@ class tnlTestNeighbourGridEntityGetter<
const GridEntityType& entity;
- //tnlTestNeighbourGridEntityGetter(){};
+ //tnlTestNeighborGridEntityGetter(){};
};
/****
* +-----------------+---------------------------+-------------------+
- * | EntityDimenions | NeighbourEntityDimension | Stencil Storage |
+ * | EntityDimenions | NeighborEntityDimension | Stencil Storage |
* +-----------------+---------------------------+-------------------+
* | 2 | 2 | Cross |
* +-----------------+---------------------------+-------------------+
@@ -76,7 +76,7 @@ template< typename Real,
typename Device,
typename Index,
typename Config >
-class tnlTestNeighbourGridEntityGetter<
+class tnlTestNeighborGridEntityGetter<
GridEntity< Meshes::Grid< 2, Real, Device, Index >, 2, Config >,
2,
GridEntityStencilStorageTag< GridEntityCrossStencil > >
@@ -84,22 +84,22 @@ class tnlTestNeighbourGridEntityGetter<
public:
static const int EntityDimension = 2;
- static const int NeighbourEntityDimension = 2;
+ static const int NeighborEntityDimension = 2;
typedef Meshes::Grid< 2, Real, Device, Index > GridType;
typedef GridEntity< GridType, EntityDimension, Config > GridEntityType;
- typedef GridEntity< GridType, NeighbourEntityDimension, Config > NeighbourGridEntityType;
+ typedef GridEntity< GridType, NeighborEntityDimension, Config > NeighborGridEntityType;
typedef Real RealType;
typedef Index IndexType;
typedef typename GridType::CoordinatesType CoordinatesType;
- typedef GridEntityGetter< GridType, NeighbourGridEntityType > GridEntityGetter;
+ typedef GridEntityGetter< GridType, NeighborGridEntityType > GridEntityGetter;
typedef GridEntityStencilStorageTag< GridEntityCrossStencil > StencilStorage;
- typedef tnlTestNeighbourGridEntityGetter< GridEntityType, 2, StencilStorage > ThisType;
+ typedef tnlTestNeighborGridEntityGetter< GridEntityType, 2, StencilStorage > ThisType;
static const int stencilSize = Config::getStencilSize();
__cuda_callable__ inline
- tnlTestNeighbourGridEntityGetter( const GridEntityType& entity )
+ tnlTestNeighborGridEntityGetter( const GridEntityType& entity )
: entity( entity )
{}
@@ -110,9 +110,9 @@ class tnlTestNeighbourGridEntityGetter<
public:
__cuda_callable__
- static void exec( ThisType& neighbourEntityGetter, const IndexType& entityIndex )
+ static void exec( ThisType& neighborEntityGetter, const IndexType& entityIndex )
{
- neighbourEntityGetter.stencilX[ index + stencilSize ] = entityIndex + index;
+ neighborEntityGetter.stencilX[ index + stencilSize ] = entityIndex + index;
}
};
@@ -122,10 +122,10 @@ class tnlTestNeighbourGridEntityGetter<
public:
__cuda_callable__
- static void exec( ThisType& neighbourEntityGetter, const IndexType& entityIndex )
+ static void exec( ThisType& neighborEntityGetter, const IndexType& entityIndex )
{
- neighbourEntityGetter.stencilY[ index + stencilSize ] =
- entityIndex + index * neighbourEntityGetter.entity.getMesh().getDimensions().x();
+ neighborEntityGetter.stencilY[ index + stencilSize ] =
+ entityIndex + index * neighborEntityGetter.entity.getMesh().getDimensions().x();
}
};
@@ -134,9 +134,9 @@ class tnlTestNeighbourGridEntityGetter<
void refresh( const GridType& grid, const IndexType& entityIndex )
{
#ifndef HAVE_CUDA // TODO: fix this to work with CUDA
- tnlStaticFor< IndexType, -stencilSize, 0, StencilYRefresher >::exec( *this, entityIndex );
- tnlStaticFor< IndexType, 1, stencilSize + 1, StencilYRefresher >::exec( *this, entityIndex );
- tnlStaticFor< IndexType, -stencilSize, stencilSize + 1, StencilXRefresher >::exec( *this, entityIndex );
+ StaticFor< IndexType, -stencilSize, 0, StencilYRefresher >::exec( *this, entityIndex );
+ StaticFor< IndexType, 1, stencilSize + 1, StencilYRefresher >::exec( *this, entityIndex );
+ StaticFor< IndexType, -stencilSize, stencilSize + 1, StencilXRefresher >::exec( *this, entityIndex );
#endif
};
@@ -147,6 +147,6 @@ class tnlTestNeighbourGridEntityGetter<
IndexType stencilX[ 2 * stencilSize + 1 ];
IndexType stencilY[ 2 * stencilSize + 1 ];
- //tnlTestNeighbourGridEntityGetter(){};
+ //tnlTestNeighborGridEntityGetter(){};
};
diff --git a/tests/benchmarks/share/CMakeLists.txt b/tests/benchmarks/share/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/tests/benchmarks/spmv.h b/tests/benchmarks/spmv.h
index 6fbd68dfb2bae574e47a05ea1f7e49f2b90532d4..c04a62eabe58a1474d5b9c8a2c361911d8762693 100644
--- a/tests/benchmarks/spmv.h
+++ b/tests/benchmarks/spmv.h
@@ -84,7 +84,7 @@ void setCudaTestMatrix( Matrix& matrix,
setCudaTestMatrixKernel< Matrix >
<<< cudaGridSize, cudaBlockSize >>>
( kernel_matrix, elementsPerRow, gridIdx );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
}
Devices::Cuda::freeFromDevice( kernel_matrix );
#endif
@@ -118,43 +118,25 @@ benchmarkSpMV( Benchmark & benchmark,
parseObjectType( HostMatrix::getType(), parsedType );
benchmark.createHorizontalGroup( parsedType[ 0 ], 2 );
- if( ! hostRowLengths.setSize( size ) ||
- ! hostMatrix.setDimensions( size, size ) ||
- ! hostVector.setSize( size ) ||
- ! hostVector2.setSize( size )
+ hostRowLengths.setSize( size );
+ hostMatrix.setDimensions( size, size );
+ hostVector.setSize( size );
+ hostVector2.setSize( size );
#ifdef HAVE_CUDA
- ||
- ! deviceRowLengths.setSize( size ) ||
- ! deviceMatrix.setDimensions( size, size ) ||
- ! deviceVector.setSize( size ) ||
- ! deviceVector2.setSize( size )
+ deviceRowLengths.setSize( size );
+ deviceMatrix.setDimensions( size, size );
+ deviceVector.setSize( size );
+ deviceVector2.setSize( size );
#endif
- )
- {
- const char* msg = "error: allocation of vectors failed";
- std::cerr << msg << std::endl;
- benchmark.addErrorMessage( msg, 2 );
- return false;
- }
hostRowLengths.setValue( elementsPerRow );
#ifdef HAVE_CUDA
deviceRowLengths.setValue( elementsPerRow );
#endif
- if( ! hostMatrix.setCompressedRowLengths( hostRowLengths ) ) {
- const char* msg = "error: allocation of host matrix failed";
- std::cerr << msg << std::endl;
- benchmark.addErrorMessage( msg, 2 );
- return false;
- }
+ hostMatrix.setCompressedRowLengths( hostRowLengths );
#ifdef HAVE_CUDA
- if( ! deviceMatrix.setCompressedRowLengths( deviceRowLengths ) ) {
- const char* msg = "error: allocation of device matrix failed";
- std::cerr << msg << std::endl;
- benchmark.addErrorMessage( msg, 2 );
- return false;
- }
+ deviceMatrix.setCompressedRowLengths( deviceRowLengths );
#endif
const int elements = setHostTestMatrix< HostMatrix >( hostMatrix, elementsPerRow );
diff --git a/tests/benchmarks/tnl-benchmark-spmv.h b/tests/benchmarks/tnl-benchmark-spmv.h
index 3fcb997c3f4d0711ca43d847b61aa8be89f7c1de..a7ebf68dc40e368929f7a5c284e65ead69845783 100644
--- a/tests/benchmarks/tnl-benchmark-spmv.h
+++ b/tests/benchmarks/tnl-benchmark-spmv.h
@@ -326,7 +326,6 @@ bool setupBenchmark( const Config::ParameterContainer& parameters )
return false;
}
const int rows = csrMatrix.getRows();
- const int columns = csrMatrix.getColumns();
const long int nonzeroElements = csrMatrix.getNumberOfMatrixElements();
Containers::Vector< int, Devices::Host, int > rowLengthsHost;
rowLengthsHost.setSize( rows );
@@ -363,160 +362,153 @@ bool setupBenchmark( const Config::ParameterContainer& parameters )
typedef CSR< Real, Devices::Cuda, int > CSRCudaType;
CSRCudaType cudaCSR;
//cout << "Copying matrix to GPU... ";
- if( ! cudaCSR.copyFrom( csrMatrix, rowLengthsCuda ) )
- {
- std::cerr << "I am not able to transfer the matrix on GPU." << std::endl;
- writeTestFailed( logFile, 21 );
- }
- else
- {
- ::tnlCusparseCSR< Real > cusparseCSR;
- cusparseCSR.init( cudaCSR, &cusparseHandle );
- benchmarkMatrix( cusparseCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "Cusparse CSR",
- stopTime,
- baseline,
- verbose,
- logFile );
- cusparseDestroy( cusparseHandle );
+ cudaCSR = csrMatrix;
+ ::tnlCusparseCSR< Real > cusparseCSR;
+ cusparseCSR.init( cudaCSR, &cusparseHandle );
+ benchmarkMatrix( cusparseCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "Cusparse CSR",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cusparseDestroy( cusparseHandle );
- std::cout << " done. \r";
- /*cudaCSR.setCudaKernelType( CSRCudaType::scalar );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Scalar",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setCudaKernelType( CSRCudaType::vector );
- cudaCSR.setCudaWarpSize( 1 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Vector 1",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setCudaWarpSize( 2 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Vector 2",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setCudaWarpSize( 4 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Vector 4",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setCudaWarpSize( 8 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Vector 8",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setCudaWarpSize( 16 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Vector 16",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setCudaWarpSize( 32 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Vector 32",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setCudaKernelType( CSRCudaType::hybrid );
- cudaCSR.setHybridModeSplit( 2 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Hyrbid 2",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setHybridModeSplit( 4 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Hyrbid 4",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setHybridModeSplit( 8 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Hyrbid 8",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setHybridModeSplit( 16 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Hyrbid 16",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setHybridModeSplit( 32 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Hyrbid 32",
- stopTime,
- baseline,
- verbose,
- logFile );
- cudaCSR.setHybridModeSplit( 64 );
- benchmarkMatrix( cudaCSR,
- cudaX,
- cudaB,
- nonzeroElements,
- "CSR Cuda Hyrbid 64",
- stopTime,
- baseline,
- verbose,
- logFile );*/
- }
+ std::cout << " done. \r";
+ /*cudaCSR.setCudaKernelType( CSRCudaType::scalar );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Scalar",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setCudaKernelType( CSRCudaType::vector );
+ cudaCSR.setCudaWarpSize( 1 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Vector 1",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setCudaWarpSize( 2 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Vector 2",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setCudaWarpSize( 4 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Vector 4",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setCudaWarpSize( 8 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Vector 8",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setCudaWarpSize( 16 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Vector 16",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setCudaWarpSize( 32 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Vector 32",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setCudaKernelType( CSRCudaType::hybrid );
+ cudaCSR.setHybridModeSplit( 2 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Hyrbid 2",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setHybridModeSplit( 4 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Hyrbid 4",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setHybridModeSplit( 8 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Hyrbid 8",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setHybridModeSplit( 16 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Hyrbid 16",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setHybridModeSplit( 32 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Hyrbid 32",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaCSR.setHybridModeSplit( 64 );
+ benchmarkMatrix( cudaCSR,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "CSR Cuda Hyrbid 64",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );*/
cudaCSR.reset();
#endif
@@ -524,138 +516,105 @@ bool setupBenchmark( const Config::ParameterContainer& parameters )
double padding;
typedef Ellpack< Real, Devices::Host, int > EllpackType;
EllpackType ellpackMatrix;
- if( ! ellpackMatrix.copyFrom( csrMatrix, rowLengthsHost ) )
- writeTestFailed( logFile, 7 );
- else
- {
- allocatedElements = ellpackMatrix.getNumberOfMatrixElements();
- padding = ( double ) allocatedElements / ( double ) nonzeroElements * 100.0 - 100.0;
- logFile << " " << padding << std::endl;
- benchmarkMatrix( ellpackMatrix,
- hostX,
- hostB,
- nonzeroElements,
- "Ellpack Host",
- stopTime,
- baseline,
- verbose,
- logFile );
+ Matrices::copySparseMatrix( ellpackMatrix, csrMatrix );
+ allocatedElements = ellpackMatrix.getNumberOfMatrixElements();
+ padding = ( double ) allocatedElements / ( double ) nonzeroElements * 100.0 - 100.0;
+ logFile << " " << padding << std::endl;
+ benchmarkMatrix( ellpackMatrix,
+ hostX,
+ hostB,
+ nonzeroElements,
+ "Ellpack Host",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
#ifdef HAVE_CUDA
- typedef Ellpack< Real, Devices::Cuda, int > EllpackCudaType;
- EllpackCudaType cudaEllpack;
- std::cout << "Copying matrix to GPU... ";
- if( ! cudaEllpack.copyFrom( ellpackMatrix, rowLengthsCuda ) )
- {
- std::cerr << "I am not able to transfer the matrix on GPU." << std::endl;
- writeTestFailed( logFile, 3 );
- }
- else
- {
- std::cout << " done. \r";
- benchmarkMatrix( cudaEllpack,
- cudaX,
- cudaB,
- nonzeroElements,
- "Ellpack Cuda",
- stopTime,
- baseline,
- verbose,
- logFile );
- }
- cudaEllpack.reset();
+ typedef Ellpack< Real, Devices::Cuda, int > EllpackCudaType;
+ EllpackCudaType cudaEllpack;
+ std::cout << "Copying matrix to GPU... ";
+ cudaEllpack = ellpackMatrix;
+ std::cout << " done. \r";
+ benchmarkMatrix( cudaEllpack,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "Ellpack Cuda",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaEllpack.reset();
#endif
- ellpackMatrix.reset();
- }
+ ellpackMatrix.reset();
typedef SlicedEllpack< Real, Devices::Host, int > SlicedEllpackType;
SlicedEllpackType slicedEllpack;
- if( ! slicedEllpack.copyFrom( csrMatrix, rowLengthsHost ) )
- writeTestFailed( logFile, 7 );
- else
- {
- allocatedElements = slicedEllpack.getNumberOfMatrixElements();
- padding = ( double ) allocatedElements / ( double ) nonzeroElements * 100.0 - 100.0;
- logFile << " " << padding << std::endl;
- benchmarkMatrix( slicedEllpack,
- hostX,
- hostB,
- nonzeroElements,
- "SlicedEllpack Host",
- stopTime,
- baseline,
- verbose,
- logFile );
+ Matrices::copySparseMatrix( slicedEllpack, csrMatrix );
+ allocatedElements = slicedEllpack.getNumberOfMatrixElements();
+ padding = ( double ) allocatedElements / ( double ) nonzeroElements * 100.0 - 100.0;
+ logFile << " " << padding << std::endl;
+ benchmarkMatrix( slicedEllpack,
+ hostX,
+ hostB,
+ nonzeroElements,
+ "SlicedEllpack Host",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
#ifdef HAVE_CUDA
- typedef SlicedEllpack< Real, Devices::Cuda, int > SlicedEllpackCudaType;
- SlicedEllpackCudaType cudaSlicedEllpack;
- std::cout << "Copying matrix to GPU... ";
- if( ! cudaSlicedEllpack.copyFrom( slicedEllpack, rowLengthsCuda ) )
- {
- std::cerr << "I am not able to transfer the matrix on GPU." << std::endl;
- writeTestFailed( logFile, 3 );
- }
- else
- {
- std::cout << " done. \r";
- benchmarkMatrix( cudaSlicedEllpack,
- cudaX,
- cudaB,
- nonzeroElements,
- "SlicedEllpack Cuda",
- stopTime,
- baseline,
- verbose,
- logFile );
- }
- cudaSlicedEllpack.reset();
+ typedef SlicedEllpack< Real, Devices::Cuda, int > SlicedEllpackCudaType;
+ SlicedEllpackCudaType cudaSlicedEllpack;
+ std::cout << "Copying matrix to GPU... ";
+ cudaSlicedEllpack = slicedEllpack;
+ std::cout << " done. \r";
+ benchmarkMatrix( cudaSlicedEllpack,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "SlicedEllpack Cuda",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaSlicedEllpack.reset();
#endif
- slicedEllpack.reset();
- }
+ slicedEllpack.reset();
typedef ChunkedEllpack< Real, Devices::Host, int > ChunkedEllpackType;
ChunkedEllpackType chunkedEllpack;
- if( ! chunkedEllpack.copyFrom( csrMatrix, rowLengthsHost ) )
- writeTestFailed( logFile, 7 );
- else
- {
- allocatedElements = chunkedEllpack.getNumberOfMatrixElements();
- padding = ( double ) allocatedElements / ( double ) nonzeroElements * 100.0 - 100.0;
- logFile << " " << padding << std::endl;
- benchmarkMatrix( chunkedEllpack,
- hostX,
- hostB,
- nonzeroElements,
- "ChunkedEllpack Host",
- stopTime,
- baseline,
- verbose,
- logFile );
+ Matrices::copySparseMatrix( chunkedEllpack, csrMatrix );
+ allocatedElements = chunkedEllpack.getNumberOfMatrixElements();
+ padding = ( double ) allocatedElements / ( double ) nonzeroElements * 100.0 - 100.0;
+ logFile << " " << padding << std::endl;
+ benchmarkMatrix( chunkedEllpack,
+ hostX,
+ hostB,
+ nonzeroElements,
+ "ChunkedEllpack Host",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
#ifdef HAVE_CUDA
- typedef ChunkedEllpack< Real, Devices::Cuda, int > ChunkedEllpackCudaType;
- ChunkedEllpackCudaType cudaChunkedEllpack;
- std::cout << "Copying matrix to GPU... ";
- if( ! cudaChunkedEllpack.copyFrom( chunkedEllpack, rowLengthsCuda ) )
- {
- std::cerr << "I am not able to transfer the matrix on GPU." << std::endl;
- writeTestFailed( logFile, 3 );
- }
- else
- {
- std::cout << " done. \r";
- benchmarkMatrix( cudaChunkedEllpack,
- cudaX,
- cudaB,
- nonzeroElements,
- "ChunkedEllpack Cuda",
- stopTime,
- baseline,
- verbose,
- logFile );
- }
- cudaChunkedEllpack.reset();
+ typedef ChunkedEllpack< Real, Devices::Cuda, int > ChunkedEllpackCudaType;
+ ChunkedEllpackCudaType cudaChunkedEllpack;
+ std::cout << "Copying matrix to GPU... ";
+ cudaChunkedEllpack = chunkedEllpack;
+ std::cout << " done. \r";
+ benchmarkMatrix( cudaChunkedEllpack,
+ cudaX,
+ cudaB,
+ nonzeroElements,
+ "ChunkedEllpack Cuda",
+ stopTime,
+ baseline,
+ verbose,
+ logFile );
+ cudaChunkedEllpack.reset();
#endif
- chunkedEllpack.reset();
- }
+ chunkedEllpack.reset();
}
return true;
}
diff --git a/tests/benchmarks/vector-operations.h b/tests/benchmarks/vector-operations.h
index 1c663ac20de0ecb1bfe20300787a43303b1f65bc..093243c07c1912919c0cd5baae1c0abb9f4a22b4 100644
--- a/tests/benchmarks/vector-operations.h
+++ b/tests/benchmarks/vector-operations.h
@@ -42,20 +42,12 @@ benchmarkVectorOperations( Benchmark & benchmark,
HostVector hostVector, hostVector2;
CudaVector deviceVector, deviceVector2;
- if( ! hostVector.setSize( size ) ||
- ! hostVector2.setSize( size )
+ hostVector.setSize( size );
+ hostVector2.setSize( size );
#ifdef HAVE_CUDA
- ||
- ! deviceVector.setSize( size ) ||
- ! deviceVector2.setSize( size )
+ deviceVector.setSize( size );
+ deviceVector2.setSize( size );
#endif
- )
- {
- const char* msg = "error: allocation of vectors failed";
- std::cerr << msg << std::endl;
- benchmark.addErrorMessage( msg );
- return false;
- }
Real resultHost, resultDevice;
diff --git a/tests/long-time-unit-tests/CMakeLists.txt b/tests/long-time-unit-tests/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/tests/mic/CMakeLists.txt b/tests/mic/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..7e6d9e53437394174a3af73c6c63067a1633b229
--- /dev/null
+++ b/tests/mic/CMakeLists.txt
@@ -0,0 +1,9 @@
+ ADD_EXECUTABLE( tnlMICArrayTest${mpiExt}${debugExt} ${headers} tnlMICArrayTest.cpp )
+ TARGET_LINK_LIBRARIES( tnlMICArrayTest${mpiExt}${debugExt} ${CPPUNIT_LIBRARIES}
+ tnl${mpiExt}${debugExt}-0.1 )
+ TARGET_COMPILE_DEFINITIONS( tnlMICArrayTest${mpiExt}${debugExt} PUBLIC ${MIC_CXX_FLAGS} )
+
+ ADD_EXECUTABLE( tnlMICVectorTest${mpiExt}${debugExt} ${headers} tnlMICVectorTest.cpp )
+ TARGET_LINK_LIBRARIES( tnlMICVectorTest${mpiExt}${debugExt} ${CPPUNIT_LIBRARIES}
+ tnl${mpiExt}${debugExt}-0.1 )
+ TARGET_COMPILE_DEFINITIONS( tnlMICVectorTest${mpiExt}${debugExt} PUBLIC ${MIC_CXX_FLAGS} )
\ No newline at end of file
diff --git a/tests/mic/tnlMICArrayTest.cpp b/tests/mic/tnlMICArrayTest.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..7c932fad62a86d9ffc0b42abdb5494a954940a5c
--- /dev/null
+++ b/tests/mic/tnlMICArrayTest.cpp
@@ -0,0 +1,205 @@
+/***************************************************************************
+ tnlMICArrayTest.cpp -
+ application testing Array implemntation on MIC KNC
+ by hanouvit
+ -------------------
+ copyright : (C) 2004 by Tomas Oberhuber
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#include <iostream>
+#include <TNL/Devices/MIC.h>
+#include <omp.h>
+#include <TNL/Containers/Array.h>
+
+ using namespace std;
+ using namespace TNL;
+ using namespace TNL::Containers;
+
+//TUNE MACROS FOR YOUR FUNKY OUTPUT
+#define TEST_VERBOSE
+
+unsigned int errors=0;
+unsigned int success=0;
+#define TEST_TEST(a) if((a)){cout << __LINE__ <<":\t OK " <<endl;success++;}else{cout << __LINE__<<":\t FAIL" <<endl;errors++;}
+#define TEST_RESULT cout<<"SUCCES: "<<success<<endl<<"ERRRORS: "<<errors<<endl;
+inline void Test_Say( const char * message)
+{
+#ifdef TEST_VERBOSE
+ cout << message <<endl;
+#endif
+}
+
+using namespace std;
+
+
+int main(void)
+{
+ cout << "Array on MIC test by hanouvit:" <<endl;
+
+ #ifdef HAVE_ICPC
+ cout << "ICPC in USE" <<endl;
+ #endif
+
+ #ifdef HAVE_MIC
+ cout << "MIC in USE" <<endl; //LOL
+ #endif
+
+#ifdef HAVE_MIC
+//prepare arrays with data
+
+ Array<double,Devices::MIC,int> aa(10);
+ Array<double,Devices::MIC,int> ee(6);
+ Array<double,Devices::Host,int> cc(5);
+
+//fill it
+Devices::MICHider<double> data_ptr;
+data_ptr.pointer= aa.getData();
+int size=aa.getSize();
+
+#pragma offload target(mic) in(data_ptr,size)
+{
+ for(int i=0;i<size;i++)
+ {
+ data_ptr.pointer[i]=i;
+ }
+}
+
+for(int i=0;i<5;i++)
+{
+ cc[i]=10+i;
+}
+
+//prepare arrays for funky tests
+Array<double,Devices::MIC,int> bb(10);
+Array<double,Devices::MIC,int> dd(0);
+
+//TEST IT!
+Test_Say("aa.getSize():");
+TEST_TEST(aa.getSize()==10);
+
+Test_Say("Is aa filled correctly? (aa.getElement):");
+for(int i=0;i<10;i++)
+ TEST_TEST(aa.getElement(i)==i);
+
+Test_Say("Copy to bb(MIC->MIC) (=):");
+bb=aa;
+TEST_TEST(aa.getSize()==bb.getSize());
+for(int i=0;i<bb.getSize();i++)
+ TEST_TEST(bb.getElement(i)==i);
+
+Test_Say("setLike:");
+bb.setLike(cc);
+TEST_TEST(bb.getSize()==cc.getSize());
+Test_Say("Copy (Host -> MIC) (=)");
+bb=cc;
+for(int i=0;i<bb.getSize();i++)
+ TEST_TEST(bb.getElement(i)==i+10);
+
+Test_Say("setValue:");
+bb.setValue(5);
+for(int i=0;i<bb.getSize();i++)
+ TEST_TEST(bb.getElement(i)==5);
+
+Test_Say("swap:");
+aa.swap(bb);
+TEST_TEST(aa.getSize()==5||bb.getSize()==10);
+for(int i=0;i<aa.getSize();i++)
+{
+ TEST_TEST(aa.getElement(i)==5);
+}
+for(int i=0;i<bb.getSize();i++)
+{
+ TEST_TEST(bb.getElement(i)==i);
+}
+
+Test_Say("(MIC -> MIC) ==");
+aa.setLike(bb);
+aa=bb;
+TEST_TEST(aa==bb);
+TEST_TEST(!(aa!=bb));
+TEST_TEST(aa!=ee);
+bb.setElement(5,66);
+TEST_TEST(aa!=bb);
+
+Test_Say("(Host -> MIC) !=");
+aa.setLike(cc);
+aa=cc;
+TEST_TEST(aa==cc);
+aa.setElement(3,66);
+TEST_TEST(aa!=cc);
+
+Test_Say("bidn (light test)");
+dd.bind(bb,5);
+TEST_TEST(dd.getSize()==5);
+TEST_TEST(dd.getElement(1)==6);
+
+//MylsÃÂm, že nenàzdaleka testováno vÅ¡e...
+
+///////////////////////////////////////////////////////////////////////////////
+
+Test_Say("File Array Test: \n");
+
+//prepare arrays with data
+
+aa.setSize(10);
+ee.setSize(6);
+cc.setSize(5);
+
+//fill it UP
+/*Devices::MICHider<double> data_ptr;*/
+data_ptr.pointer= aa.getData();
+//size=aa.getSize();
+
+#pragma offload target(mic) in(data_ptr,size)
+{
+ for(int i=0;i<size;i++)
+ {
+ data_ptr.pointer[i]=i;
+ }
+}
+
+for(int i=0;i<5;i++)
+{
+ cc[i]=10+i;
+}
+
+File soubor;
+soubor.open("/tmp/tnlArrayExperimentSave_cc.bin",IOMode::write);
+cc.save(soubor);
+soubor.close();
+
+soubor.open("/tmp/tnlArrayExperimentSave_aa.bin",IOMode::write);
+aa.save(soubor);
+soubor.close();
+
+ee.bind(aa,5,5);
+ee.boundLoad("/tmp/tnlArrayExperimentSave_cc.bin");
+
+TEST_TEST( 10 == aa.getSize())
+for(int i=0;i<5;i++)
+{
+ TEST_TEST(aa.getElement(i)==i)
+}
+for(int i=5;i<10;i++)
+{
+ TEST_TEST(aa.getElement(i)==i+5)
+}
+
+soubor.open("/tmp/tnlArrayExperimentSave_aa.bin",IOMode::read);
+cc.load(soubor);
+soubor.close();
+
+TEST_TEST( 10 == cc.getSize())
+for(int i=0;i<cc.getSize();i++)
+{
+ TEST_TEST(cc.getElement(i)==i)
+}
+
+#endif
+
+ TEST_RESULT;
+ return 0;
+}
diff --git a/tests/mic/tnlMICVectorTest.cpp b/tests/mic/tnlMICVectorTest.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..058715b208c229e58b1d50a98e0c747304f679cd
--- /dev/null
+++ b/tests/mic/tnlMICVectorTest.cpp
@@ -0,0 +1,150 @@
+/***************************************************************************
+ tnlMICVectorTest.cpp -
+ application testing Vector implemntation on MIC KNC
+ by hanouvit
+ -------------------
+ copyright : (C) 2004 by Tomas Oberhuber
+ email : tomas.oberhuber@fjfi.cvut.cz
+ ***************************************************************************/
+
+/* See Copyright Notice in tnl/Copyright */
+
+#include <iostream>
+#include <omp.h>
+#include <stdint.h>
+
+#include <TNL/Devices/MIC.h>
+#include <TNL/Containers/Vector.h>
+
+
+using namespace std;
+using namespace TNL;
+using namespace TNL::Containers;
+
+//TUNE MACROS FOR YOUR FUNKY OUTPUT
+#define TEST_VERBOSE
+
+unsigned int errors=0;
+unsigned int success=0;
+#define TEST_TEST(a) if((a)){cout << __LINE__ <<":\t OK" <<endl;success++;}else{cout << __LINE__<<":\t FAIL" <<endl;errors++;}
+#define TEST_RESULT cout<<"SUCCES: "<<success<<endl<<"ERRRORS: "<<errors<<endl;
+inline void Test_Say( const char * message)
+{
+#ifdef TEST_VERBOSE
+ cout << message <<endl;
+#endif
+}
+
+int main(void)
+{
+ cout << "Vector on MIC test by hanouvit:" <<endl;
+
+ #ifdef HAVE_ICPC
+ cout << "ICPC in USE" <<endl;
+ #endif
+
+ #ifdef HAVE_MIC
+ cout << "MIC in USE" <<endl;
+ #endif
+
+#ifdef HAVE_MIC
+ Vector<double,Devices::MIC,int> aa(10);
+ Vector<double,Devices::MIC,int> bb(10);
+ Vector<double,Devices::MIC,int> cc(10);
+
+ Vector<double,Devices::Host,int> aaa(10);
+ Vector<double,Devices::Host,int> bbb(10);
+ Vector<double,Devices::Host,int> ccc(10);
+
+ for(int i=0;i<10;i++)
+ {
+ aa.setElement(i,i-5);
+ aaa.setElement(i,i-5);
+ bb.setElement(i,5-i);
+ bbb.setElement(i,5-i);
+ cc.setElement(i,10+i);
+ ccc.setElement(i,10+i);
+ }
+
+ Test_Say("Is filled correctly?:");
+ for(int i=0;i<10;i++)
+ {
+ TEST_TEST(aa.getElement(i)==aaa.getElement(i));
+ TEST_TEST(bb.getElement(i)==bbb.getElement(i));
+ TEST_TEST(cc.getElement(i)==ccc.getElement(i));
+ }
+
+ Test_Say("min():");
+ TEST_TEST(bb.min()==bbb.min());
+ Test_Say("absMin():");
+ TEST_TEST(bb.absMin()==bbb.absMin());
+ Test_Say("max():");
+ TEST_TEST(bb.max()==bbb.max());
+ Test_Say("absMax():");
+ TEST_TEST(bb.absMax()==bbb.absMax());
+
+ Test_Say("lpNorm( N ):");
+ TEST_TEST(aa.lpNorm(1)==aaa.lpNorm(1));
+ TEST_TEST(aa.lpNorm(2)==aaa.lpNorm(2));
+ TEST_TEST(aa.lpNorm(0.5)==aaa.lpNorm(0.5));
+ TEST_TEST(aa.lpNorm(3)==aaa.lpNorm(3));
+ Test_Say("sum():");
+ TEST_TEST(aa.sum()==aaa.sum());
+
+ Test_Say("differenceMax():");
+ TEST_TEST(aa.differenceMax(bb)==aaa.differenceMax(bbb));
+ Test_Say("differenceMin():");
+ TEST_TEST(aa.differenceMin(bb)==aaa.differenceMin(bbb));
+ Test_Say("differenceAbsMax():");
+ TEST_TEST(aa.differenceAbsMax(bb)==aaa.differenceAbsMax(bbb));
+ Test_Say("differenceAbsMin():");
+ TEST_TEST(aa.differenceAbsMin(bb)==aaa.differenceAbsMin(bbb));
+ Test_Say("differenceSum():");
+ TEST_TEST(aa.differenceSum(bb)==aaa.differenceSum(bbb));
+
+ ////
+ Test_Say("differenceLpNorm( N ):");
+ TEST_TEST(aa.differenceLpNorm(bb,1)==aaa.differenceLpNorm(bbb,1));
+ TEST_TEST(aa.differenceLpNorm(bb,2)==aaa.differenceLpNorm(bbb,2));
+ TEST_TEST(aa.differenceLpNorm(bb,0.5)==aaa.differenceLpNorm(bbb,0.5));
+ TEST_TEST(aa.differenceLpNorm(bb,3)==aaa.differenceLpNorm(bbb,3));
+
+ ////
+ Test_Say("== :");
+ TEST_TEST(aa==aaa);
+ Test_Say("vct*0.5 :");
+ aa*=0.5;
+ aaa*=0.5;
+ TEST_TEST(aa==aaa);
+
+ Test_Say("scalarProduct :");
+ TEST_TEST(aa.scalarProduct(bb) == aaa.scalarProduct(bbb));
+
+ Test_Say("addVector :");
+ aa.addVector(bb,2.0,3.0);
+ aaa.addVector(bbb,2.0,3.0);
+ TEST_TEST(aa==aaa);
+ aa.addVectors(bb,2.0,cc,1.0,-3.0);
+ aaa.addVectors(bbb,2.0,ccc,1.0,-3.0);
+ TEST_TEST(aa==aaa);
+
+ Test_Say("computeExclusivePrefixSum :");
+ aa.computeExclusivePrefixSum();
+ aaa.computeExclusivePrefixSum();
+ TEST_TEST(aa==aaa);
+ bb.computeExclusivePrefixSum(2,4);
+ bbb.computeExclusivePrefixSum(2,4);
+ TEST_TEST(bb==bbb);
+
+ Test_Say("computePrefixSum :");
+ cc.computePrefixSum();
+ ccc.computePrefixSum();
+ TEST_TEST(cc==ccc);
+ cc.computePrefixSum(2,4);
+ ccc.computePrefixSum(2,4);
+ TEST_TEST(cc==ccc);
+#endif
+ TEST_RESULT;
+
+ return 0;
+}
diff --git a/tests/mpi/Functions.h b/tests/mpi/Functions.h
index 21cfeb9276740887960be16cb686be660e7374ae..c74be9e63a70e359eda3fbe5110ed7b9b8c5c52e 100644
--- a/tests/mpi/Functions.h
+++ b/tests/mpi/Functions.h
@@ -80,8 +80,6 @@ class FunctionToEvaluate<Real,2> : public Functions::Domain< 2, Functions::MeshD
{
//return meshEntity.getCoordinates().y()*10+meshEntity.getCoordinates().x();
return meshEntity.getCenter().y()*100+meshEntity.getCenter().x();
-
-
}
};
diff --git a/tests/mpi/MeshFunctionEvauateTest.cpp b/tests/mpi/MeshFunctionEvauateTest.cpp
index fc83f44dc4d55af3812c6917821d55082e59d971..7ab5a7f6ad044f183e308f8cc90fe13f5b35e46d 100644
--- a/tests/mpi/MeshFunctionEvauateTest.cpp
+++ b/tests/mpi/MeshFunctionEvauateTest.cpp
@@ -18,7 +18,7 @@ using namespace std;
#include <TNL/Timer.h>
-#define OUTPUT
+//#define OUTPUT
#include "Functions.h"
diff --git a/tests/mpi/a.out b/tests/mpi/a.out
deleted file mode 100755
index 0228d30c7f9574335ed795dc630c0485e3d15ea2..0000000000000000000000000000000000000000
Binary files a/tests/mpi/a.out and /dev/null differ
diff --git a/tests/unit-tests/CMakeLists.txt b/tests/unit-tests/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/tests/unit-tests/core/cuda/CMakeLists.txt b/tests/unit-tests/core/cuda/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/tests/unit-tests/core/cuda/tnlCudaReductionTester.h b/tests/unit-tests/core/cuda/tnlCudaReductionTester.h
index 193b95a6808b7995d49324ecc179e1ae000287de..313063b22d690c50aa375bc01a7a489500448a99 100644
--- a/tests/unit-tests/core/cuda/tnlCudaReductionTester.h
+++ b/tests/unit-tests/core/cuda/tnlCudaReductionTester.h
@@ -95,7 +95,7 @@ class CudaReductionTester : public CppUnit :: TestCase
for( int i = 0; i < size; i ++ )
hostData[ i ] = value;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< RealType, RealType, int >( deviceData, hostData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
}
template< typename RealType >
@@ -105,7 +105,7 @@ class CudaReductionTester : public CppUnit :: TestCase
RealType *hostData, *deviceData;
ArrayOperations< Devices::Host >::allocateMemory( hostData, shortSequence );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, shortSequence );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
RealType result;
@@ -149,7 +149,7 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Host >::freeMemory( hostData );
ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
}
template< typename RealType >
@@ -159,7 +159,7 @@ class CudaReductionTester : public CppUnit :: TestCase
RealType *hostData, *deviceData;
ArrayOperations< Devices::Host >::allocateMemory( hostData, longSequence );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, longSequence );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
RealType result;
@@ -237,7 +237,7 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Host >::freeMemory( hostData );
ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
}
template< typename RealType >
@@ -247,7 +247,7 @@ class CudaReductionTester : public CppUnit :: TestCase
RealType *hostData, *deviceData;
ArrayOperations< Devices::Host >::allocateMemory( hostData, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
RealType sum( 0.0 );
for( int i = 0; i < size; i ++ )
@@ -256,7 +256,7 @@ class CudaReductionTester : public CppUnit :: TestCase
sum += hostData[ i ];
}
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< RealType, RealType, int >( deviceData, hostData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
tnlParallelReductionSum< RealType, int > sumOperation;
RealType result;
CPPUNIT_ASSERT(
@@ -289,7 +289,7 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Host >::freeMemory( hostData );
ArrayOperations< Devices::Cuda >::freeMemory( deviceData );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
}
template< typename Type >
@@ -299,13 +299,13 @@ class CudaReductionTester : public CppUnit :: TestCase
Type *hostData, *deviceData;
ArrayOperations< Devices::Host >::allocateMemory( hostData, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
for( int i = 0; i < size; i ++ )
hostData[ i ] = 1;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData, hostData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
tnlParallelReductionLogicalAnd< Type, int > andOperation;
tnlParallelReductionLogicalOr< Type, int > orOperation;
@@ -319,7 +319,7 @@ class CudaReductionTester : public CppUnit :: TestCase
hostData[ 0 ] = 0;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData, hostData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
CPPUNIT_ASSERT(
( reductionOnCudaDevice( andOperation, size, deviceData, ( Type* ) 0, result ) ) );
CPPUNIT_ASSERT( result == 0 );
@@ -331,7 +331,7 @@ class CudaReductionTester : public CppUnit :: TestCase
hostData[ i ] = 0;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData, hostData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
CPPUNIT_ASSERT(
( reductionOnCudaDevice( andOperation, size, deviceData, ( Type* ) 0, result ) ) );
CPPUNIT_ASSERT( result == 0 );
@@ -347,13 +347,13 @@ class CudaReductionTester : public CppUnit :: TestCase
Type *hostData, *deviceData;
ArrayOperations< Devices::Host >::allocateMemory( hostData, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
for( int i = 0; i < size; i ++ )
hostData[ i ] = 1;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData, hostData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
tnlParallelReductionLogicalAnd< Type, int > andOperation;
tnlParallelReductionLogicalOr< Type, int > orOperation;
@@ -367,7 +367,7 @@ class CudaReductionTester : public CppUnit :: TestCase
hostData[ 0 ] = 0;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData, hostData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
CPPUNIT_ASSERT(
( reductionOnCudaDevice( andOperation, size, deviceData, ( Type* ) 0, result ) ) );
CPPUNIT_ASSERT( result == 0 );
@@ -379,7 +379,7 @@ class CudaReductionTester : public CppUnit :: TestCase
hostData[ i ] = 0;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData, hostData, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
CPPUNIT_ASSERT(
( reductionOnCudaDevice( andOperation, size, deviceData, ( Type* ) 0, result ) ) );
CPPUNIT_ASSERT( result == 0 );
@@ -398,13 +398,13 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Host >::allocateMemory( hostData2, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData1, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData2, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
for( int i = 0; i < size; i ++ )
hostData1[ i ] = hostData2[ i ] = 1;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData1, hostData1, size );
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData2, hostData2, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
bool result( false );
tnlParallelReductionEqualities< Type, int > equalityOperation;
@@ -420,7 +420,7 @@ class CudaReductionTester : public CppUnit :: TestCase
hostData1[ 0 ] = 0;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData1, hostData1, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
CPPUNIT_ASSERT(
( reductionOnCudaDevice( equalityOperation, size, deviceData1, deviceData2, result ) ) );
@@ -433,7 +433,7 @@ class CudaReductionTester : public CppUnit :: TestCase
for( int i = 0; i < size; i ++ )
hostData1[ i ] = 0;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData1, hostData1, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
CPPUNIT_ASSERT(
( reductionOnCudaDevice( equalityOperation, size, deviceData1, deviceData2, result ) ) );
@@ -454,13 +454,13 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Host >::allocateMemory( hostData2, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData1, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData2, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
for( int i = 0; i < size; i ++ )
hostData1[ i ] = hostData2[ i ] = 1;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData1, hostData1, size );
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData2, hostData2, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
bool result( false );
tnlParallelReductionEqualities< Type, int > equalityOperation;
@@ -476,7 +476,7 @@ class CudaReductionTester : public CppUnit :: TestCase
hostData1[ 0 ] = 0;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData1, hostData1, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
CPPUNIT_ASSERT(
( reductionOnCudaDevice( equalityOperation, size, deviceData1, deviceData2, result ) ) );
@@ -489,7 +489,7 @@ class CudaReductionTester : public CppUnit :: TestCase
for( int i = 0; i < size; i ++ )
hostData1[ i ] = 0;
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData1, hostData1, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
CPPUNIT_ASSERT(
( reductionOnCudaDevice( equalityOperation, size, deviceData1, deviceData2, result ) ) );
@@ -510,7 +510,7 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Host >::allocateMemory( hostData2, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData1, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData2, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
hostData1[ 0 ] = 0;
hostData2[ 0 ] = 1;
@@ -523,7 +523,7 @@ class CudaReductionTester : public CppUnit :: TestCase
}
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData1, hostData1, size );
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData2, hostData2, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
Type result( 0.0 );
tnlParallelReductionScalarProduct< Type, int > scalarProductOperation;
@@ -544,7 +544,7 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Host >::allocateMemory( hostData2, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData1, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceData2, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
hostData1[ 0 ] = 0;
hostData2[ 0 ] = 1;
@@ -557,7 +557,7 @@ class CudaReductionTester : public CppUnit :: TestCase
}
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData1, hostData1, size );
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceData2, hostData2, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
Type result( 0.0 );
tnlParallelReductionScalarProduct< Type, int > scalarProductOperation;
@@ -579,7 +579,7 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Cuda >::allocateMemory( deviceZeros, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceOnes, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceLinear, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
for( int i = 0; i < size; i ++ )
{
@@ -591,7 +591,7 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceZeros, hostZeros, size );
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceOnes, hostOnes, size );
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceLinear, hostLinear, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
tnlParallelReductionDiffSum< Type, int > diffSumOp;
tnlParallelReductionDiffMin< Type, int > diffMinOp;
@@ -696,7 +696,7 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Cuda >::allocateMemory( deviceZeros, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceOnes, size );
ArrayOperations< Devices::Cuda >::allocateMemory( deviceLinear, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
for( int i = 0; i < size; i ++ )
{
@@ -708,7 +708,7 @@ class CudaReductionTester : public CppUnit :: TestCase
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceZeros, hostZeros, size );
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceOnes, hostOnes, size );
ArrayOperations< Devices::Cuda, Devices::Host >::copyMemory< Type, Type, int >( deviceLinear, hostLinear, size );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
tnlParallelReductionDiffSum< Type, int > diffSumOp;
tnlParallelReductionDiffMin< Type, int > diffMinOp;
diff --git a/tests/unit-tests/core/cuda/tnlCudaTester.h b/tests/unit-tests/core/cuda/tnlCudaTester.h
index 5c2024cbcc5dcdb70c4ad90b15d70d400bf36649..3cb033c53b7812510f22d3d1b976b9eeee6bd267 100644
--- a/tests/unit-tests/core/cuda/tnlCudaTester.h
+++ b/tests/unit-tests/core/cuda/tnlCudaTester.h
@@ -59,7 +59,7 @@ class Devices::CudaTester : public CppUnit :: TestCase
blockSize. x = 1;
gridSize. x = 1;
simpleKernel<<< gridSize, blockSize >>>();
- if( ! checkCudaDevice )
+ if( ! TNL_CHECK_CUDA_DEVICE )
{
std::cerr << "Test with simple kernel failed. It seems that the CUDA device does not work properly." << std::endl;
CPPUNIT_ASSERT( false );
diff --git a/tests/unit-tests/core/multimaps/tnlIndexMultimapTester.h b/tests/unit-tests/core/multimaps/tnlIndexMultimapTester.h
index 2e077e9c3d9057e90819073cf48f56743e113096..236a608e5601791fec44f134a00436629e4352d5 100644
--- a/tests/unit-tests/core/multimaps/tnlIndexMultimapTester.h
+++ b/tests/unit-tests/core/multimaps/tnlIndexMultimapTester.h
@@ -208,7 +208,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
tnlIndexMultimapTester__setElementFastTestCudaKernel< IndexMultimapType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -217,7 +217,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -271,7 +271,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
tnlIndexMultimapTester__setElementFast_DiagonalIndexMultimapTestCudaKernel< IndexMultimapType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -280,7 +280,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -355,7 +355,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
tnlIndexMultimapTester__setElementFast_DenseIndexMultimapTestCudaKernel1< IndexMultimapType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -364,7 +364,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -390,7 +390,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
tnlIndexMultimapTester__setElementFast_DenseIndexMultimapTestCudaKernel2< IndexMultimapType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -399,7 +399,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -469,7 +469,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
tnlIndexMultimapTester__setElementFast_LowerTriangularIndexMultimapTestCudaKernel1< IndexMultimapType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -478,7 +478,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -504,7 +504,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
tnlIndexMultimapTester__setElementFast_LowerTriangularIndexMultimapTestCudaKernel2< IndexMultimapType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -513,7 +513,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -615,7 +615,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
tnlIndexMultimapTester__setRowFast_DiagonalIndexMultimapTestCudaKernel< IndexMultimapType >
@@ -625,7 +625,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -722,7 +722,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
tnlIndexMultimapTester__setRowFast_DenseIndexMultimapTestCudaKernel1< IndexMultimapType >
@@ -732,7 +732,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -762,7 +762,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
tnlIndexMultimapTester__setRowFast_DenseIndexMultimapTestCudaKernel2< IndexMultimapType >
@@ -772,7 +772,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -863,7 +863,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
tnlIndexMultimapTester__setRowFast_LowerTriangularIndexMultimapTestCudaKernel< IndexMultimapType >
@@ -873,7 +873,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -903,7 +903,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
IndexMultimapType* kernel_graph = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
tnlIndexMultimapTester__setRowFast_LowerTriangularIndexMultimapTestCudaKernel< IndexMultimapType >
@@ -913,7 +913,7 @@ class tnlIndexMultimapTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_graph );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
diff --git a/tests/unit-tests/matrices/CMakeLists.txt b/tests/unit-tests/matrices/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/tests/unit-tests/matrices/tnlDenseMatrixTester.h b/tests/unit-tests/matrices/tnlDenseMatrixTester.h
index 7e7d0db2ec32600180c573b8ce569f85eb213363..b2c89e27fea7c0fa04371daab0d0f251f3d3cd2a 100644
--- a/tests/unit-tests/matrices/tnlDenseMatrixTester.h
+++ b/tests/unit-tests/matrices/tnlDenseMatrixTester.h
@@ -100,11 +100,11 @@ class DenseTester : public CppUnit :: TestCase
{
#ifdef HAVE_CUDA
MatrixType* kernel_m = Devices::Cuda::passToDevice( m );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
setElementFastTestKernel<<< 1, 16 >>>( kernel_m );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
Devices::Cuda::freeFromDevice( kernel_m );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
#endif
}
for( int i = 0; i < 10; i++ )
@@ -145,11 +145,11 @@ class DenseTester : public CppUnit :: TestCase
{
#ifdef HAVE_CUDA
MatrixType* kernel_m = Devices::Cuda::passToDevice( m );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
addElementFastTestKernel<<< 1, 128 >>>( kernel_m );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
Devices::Cuda::freeFromDevice( kernel_m );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
#endif
}
for( int i = 0; i < 10; i++ )
@@ -216,11 +216,11 @@ class DenseTester : public CppUnit :: TestCase
{
#ifdef HAVE_CUDA
MatrixType* kernel_m = Devices::Cuda::passToDevice( m );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
setRowFastTestKernel<<< 1, 128 >>>( kernel_m, columns.getData(), values.getData(), ( IndexType ) 10 );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
Devices::Cuda::freeFromDevice( kernel_m );
- CPPUNIT_ASSERT( checkCudaDevice );
+ CPPUNIT_ASSERT( TNL_CHECK_CUDA_DEVICE );
#endif
}
diff --git a/tests/unit-tests/matrices/tnlSparseMatrixTester.h b/tests/unit-tests/matrices/tnlSparseMatrixTester.h
index 0bf1da58c83c8df00f55ff6066525cae28e2899d..b2b29c0e4c254a23756488fd57b1a65d151b6f50 100644
--- a/tests/unit-tests/matrices/tnlSparseMatrixTester.h
+++ b/tests/unit-tests/matrices/tnlSparseMatrixTester.h
@@ -206,7 +206,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
SparseTester__setElementFastTestCudaKernel< MatrixType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -215,7 +215,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -269,7 +269,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
SparseTester__setElementFast_DiagonalMatrixTestCudaKernel< MatrixType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -278,7 +278,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -353,7 +353,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
SparseTester__setElementFast_DenseTestCudaKernel1< MatrixType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -362,7 +362,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -388,7 +388,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
SparseTester__setElementFast_DenseTestCudaKernel2< MatrixType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -397,7 +397,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -467,7 +467,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
SparseTester__setElementFast_LowerTriangularMatrixTestCudaKernel1< MatrixType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -476,7 +476,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -502,7 +502,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
SparseTester__setElementFast_LowerTriangularMatrixTestCudaKernel2< MatrixType >
<<< cudaGridSize, cudaBlockSize >>>
@@ -511,7 +511,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -613,7 +613,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
SparseTester__setRowFast_DiagonalMatrixTestCudaKernel< MatrixType >
@@ -623,7 +623,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -720,7 +720,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
SparseTester__setRowFast_DenseTestCudaKernel1< MatrixType >
@@ -730,7 +730,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -760,7 +760,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
SparseTester__setRowFast_DenseTestCudaKernel2< MatrixType >
@@ -770,7 +770,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -861,7 +861,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
SparseTester__setRowFast_LowerTriangularMatrixTestCudaKernel< MatrixType >
@@ -871,7 +871,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
@@ -901,7 +901,7 @@ class SparseTester : public CppUnit :: TestCase
MatrixType* kernel_matrix = Devices::Cuda::passToDevice( m );
bool testResult( true );
bool* kernel_testResult = Devices::Cuda::passToDevice( testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
dim3 cudaBlockSize( 256 ), cudaGridSize( 1 );
int sharedMemory = 100 * ( sizeof( IndexType ) + sizeof( RealType ) );
SparseTester__setRowFast_LowerTriangularMatrixTestCudaKernel< MatrixType >
@@ -911,7 +911,7 @@ class SparseTester : public CppUnit :: TestCase
CPPUNIT_ASSERT( Devices::Cuda::passFromDevice( kernel_testResult ) );
Devices::Cuda::freeFromDevice( kernel_matrix );
Devices::Cuda::freeFromDevice( kernel_testResult );
- checkCudaDevice;
+ TNL_CHECK_CUDA_DEVICE;
#endif
}
diff --git a/tests/unit-tests/mesh/CMakeLists.txt b/tests/unit-tests/mesh/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/tests/unit-tests/mesh/tnlGrid2DTester.h b/tests/unit-tests/mesh/tnlGrid2DTester.h
index bb493c54d1cd51f9346bed34f5aa44d002fbec13..fcb9b47131f90684d0538d0d67a3842073ca1dd3 100644
--- a/tests/unit-tests/mesh/tnlGrid2DTester.h
+++ b/tests/unit-tests/mesh/tnlGrid2DTester.h
@@ -174,29 +174,29 @@ class GridTester< 2, RealType, Device, IndexType >: public CppUnit :: TestCase
{
const CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( -1, 0 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< -1, 0 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< -1, 0 >() ) );
}
if( cell.getCoordinates().x() < xSize - 1 )
{
const CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( 1, 0 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< 1, 0 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< 1, 0 >() ) );
}
if( cell.getCoordinates().y() > 0 )
{
const CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( 0, -1 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< 0, -1 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< 0, -1 >() ) );
}
if( cell.getCoordinates().y() < ySize - 1 )
{
const CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( 0, 1 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< 0, 1 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< 0, 1 >() ) );
}
}
}
@@ -224,35 +224,35 @@ class GridTester< 2, RealType, Device, IndexType >: public CppUnit :: TestCase
{
//const IndexType cellIndex = grid.getEntityIndex( cell );
cell.refresh(); //setIndex( cellIndex );
- auto neighbourEntities = cell.template getNeighbourEntities< GridType::Face::entityDimension >();
+ auto neighborEntities = cell.template getNeighborEntities< GridType::Face::entityDimension >();
FaceType face1( grid,
cell.getCoordinates(),
EntityOrientationType( -1, 0 ),
EntityBasisType( 0, 1 ) );
IndexType face1Index = grid.template getEntityIndex( face1 );
- CPPUNIT_ASSERT( ( face1Index == neighbourEntities.template getEntityIndex< -1, 0 >() ) );
+ CPPUNIT_ASSERT( ( face1Index == neighborEntities.template getEntityIndex< -1, 0 >() ) );
FaceType face2( grid,
cell.getCoordinates() + CoordinatesType( 1, 0 ),
EntityOrientationType( 1, 0 ),
EntityBasisType( 0, 1 ) );
IndexType face2Index = grid.template getEntityIndex( face2 );
- CPPUNIT_ASSERT( ( face2Index == neighbourEntities.template getEntityIndex< 1, 0 >() ) );
+ CPPUNIT_ASSERT( ( face2Index == neighborEntities.template getEntityIndex< 1, 0 >() ) );
FaceType face3( grid,
cell.getCoordinates(),
EntityOrientationType( 0, -1 ),
EntityBasisType( 1, 0 ) );
IndexType face3Index = grid.template getEntityIndex( face3 );
- CPPUNIT_ASSERT( ( face3Index == neighbourEntities.template getEntityIndex< 0, -1 >() ) );
+ CPPUNIT_ASSERT( ( face3Index == neighborEntities.template getEntityIndex< 0, -1 >() ) );
FaceType face4( grid,
cell.getCoordinates() + CoordinatesType( 0, 1 ),
EntityOrientationType( 0, 1 ),
EntityBasisType( 1, 0 ) );
IndexType face4Index = grid.template getEntityIndex( face4 );
- CPPUNIT_ASSERT( ( face4Index == neighbourEntities.template getEntityIndex< 0, 1 >() ) );
+ CPPUNIT_ASSERT( ( face4Index == neighborEntities.template getEntityIndex< 0, 1 >() ) );
}
}
@@ -281,20 +281,20 @@ class GridTester< 2, RealType, Device, IndexType >: public CppUnit :: TestCase
face.setOrientation( EntityOrientationType( 1, 0 ) );
//const IndexType faceIndex = grid.getEntityIndex( face );
face.refresh(); //setIndex( faceIndex );
- auto neighbourCells = face.template getNeighbourEntities< GridType::Cell::entityDimension >();
+ auto neighborCells = face.template getNeighborEntities< GridType::Cell::entityDimension >();
if( face.getCoordinates().x() > 0 )
{
CellType cell( grid, face.getCoordinates() + CoordinatesType( -1, 0 ) );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourCells.template getEntityIndex< -1, 0 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborCells.template getEntityIndex< -1, 0 >() ) );
}
if( face.getCoordinates().x() < xSize )
{
CellType cell( grid, face.getCoordinates() + CoordinatesType( 0, 0 ) );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourCells.template getEntityIndex< 1, 0 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborCells.template getEntityIndex< 1, 0 >() ) );
}
}
if( face.getCoordinates().x() < xSize )
@@ -302,19 +302,19 @@ class GridTester< 2, RealType, Device, IndexType >: public CppUnit :: TestCase
face.setOrientation( EntityOrientationType( 0, 1 ) );
//const IndexType faceIndex = grid.getEntityIndex( face );
face.refresh();//setIndex( faceIndex );
- auto neighbourCells = face.template getNeighbourEntities< GridType::Cell::entityDimension >();
+ auto neighborCells = face.template getNeighborEntities< GridType::Cell::entityDimension >();
if( face.getCoordinates().y() > 0 )
{
CellType cell( grid, face.getCoordinates() + CoordinatesType( 0, -1 ) );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourCells.template getEntityIndex< 0, -1 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborCells.template getEntityIndex< 0, -1 >() ) );
}
if( face.getCoordinates().y() < ySize )
{
CellType cell( grid, face.getCoordinates() + CoordinatesType( 0, 0 ) );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourCells.template getEntityIndex< 0, 1 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborCells.template getEntityIndex< 0, 1 >() ) );
}
}
}
diff --git a/tests/unit-tests/mesh/tnlGrid3DTester.h b/tests/unit-tests/mesh/tnlGrid3DTester.h
index 5890ff641302a5bf4bb75a580d718026d5d6b563..fed97012f5fcb2305e6ec0d3d582e7c6e8f98231 100644
--- a/tests/unit-tests/mesh/tnlGrid3DTester.h
+++ b/tests/unit-tests/mesh/tnlGrid3DTester.h
@@ -291,43 +291,43 @@ class GridTester< 3, RealType, Device, IndexType >: public CppUnit :: TestCase
{
CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( -1, 0, 0 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< -1, 0, 0 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< -1, 0, 0 >() ) );
}
if( cell.getCoordinates().x() < xSize - 1 )
{
CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( 1, 0, 0 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< 1, 0, 0 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< 1, 0, 0 >() ) );
}
if( cell.getCoordinates().y() > 0 )
{
CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( 0, -1, 0 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< 0, -1, 0 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< 0, -1, 0 >() ) );
}
if( cell.getCoordinates().y() < ySize - 1 )
{
CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( 0, 1, 0 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< 0, 1, 0 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< 0, 1, 0 >() ) );
}
if( cell.getCoordinates().z() > 0 )
{
CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( 0, 0, -1 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< 0, 0, -1 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< 0, 0, -1 >() ) );
}
if( cell.getCoordinates().z() < zSize - 1 )
{
CellType auxCell( grid, cell.getCoordinates() + CoordinatesType( 0, 0, 1 ) );
const IndexType auxCellIndex = grid.getEntityIndex( auxCell );
- auto neighbourEntities = cell.getNeighbourEntities();
- CPPUNIT_ASSERT( ( auxCellIndex == neighbourEntities.template getEntityIndex< 0, 0, 1 >() ) );
+ auto neighborEntities = cell.getNeighborEntities();
+ CPPUNIT_ASSERT( ( auxCellIndex == neighborEntities.template getEntityIndex< 0, 0, 1 >() ) );
}
}
}
@@ -357,44 +357,44 @@ class GridTester< 3, RealType, Device, IndexType >: public CppUnit :: TestCase
{
//const IndexType cellIndex = grid.getEntityIndex( cell );
cell.refresh();//setIndex( cellIndex );
- auto neighbourEntities = cell.template getNeighbourEntities< GridType::Face::entityDimension >();
+ auto neighborEntities = cell.template getNeighborEntities< GridType::Face::entityDimension >();
face.setCoordinates( cell.getCoordinates() );
face.setOrientation( EntityOrientationType( 1, 0, 0 ) );
//CoordinatesType faceCoordinates( i, j, k );
IndexType faceIndex = grid.getEntityIndex( face );
- CPPUNIT_ASSERT( ( faceIndex == neighbourEntities.template getEntityIndex< -1, 0, 0 >() ) );
+ CPPUNIT_ASSERT( ( faceIndex == neighborEntities.template getEntityIndex< -1, 0, 0 >() ) );
//faceCoordinates = CoordinatesType( i + 1, j, k );
face.setCoordinates( cell.getCoordinates() + CoordinatesType( 1, 0, 0 ) );
face.setOrientation( EntityOrientationType( 1, 0 , 0 ) );
faceIndex = grid.getEntityIndex( face );
- CPPUNIT_ASSERT( ( faceIndex == neighbourEntities.template getEntityIndex< 1, 0, 0 >() ) );
+ CPPUNIT_ASSERT( ( faceIndex == neighborEntities.template getEntityIndex< 1, 0, 0 >() ) );
//faceCoordinates = CoordinatesType( i, j, k );
face.setCoordinates( cell.getCoordinates() );
face.setOrientation( EntityOrientationType( 0, -1, 0 ) );
faceIndex = grid.getEntityIndex( face );
- CPPUNIT_ASSERT( ( faceIndex == neighbourEntities.template getEntityIndex< 0, -1, 0 >() ) );
+ CPPUNIT_ASSERT( ( faceIndex == neighborEntities.template getEntityIndex< 0, -1, 0 >() ) );
//faceCoordinates = CoordinatesType( i, j + 1, k );
face.setCoordinates( cell.getCoordinates() + CoordinatesType( 0, 1, 0 ) );
face.setOrientation( EntityOrientationType( 0, 1, 0 ) );
faceIndex = grid.getEntityIndex( face );
- CPPUNIT_ASSERT( ( faceIndex == neighbourEntities.template getEntityIndex< 0, 1, 0 >() ) );
+ CPPUNIT_ASSERT( ( faceIndex == neighborEntities.template getEntityIndex< 0, 1, 0 >() ) );
//faceCoordinates = CoordinatesType( i, j, k );
face.setCoordinates( cell.getCoordinates() );
face.setOrientation( EntityOrientationType( 0, 0, -1 ) );
faceIndex = grid.getEntityIndex( face );
- CPPUNIT_ASSERT( ( faceIndex == neighbourEntities.template getEntityIndex< 0, 0, -1 >() ) );
+ CPPUNIT_ASSERT( ( faceIndex == neighborEntities.template getEntityIndex< 0, 0, -1 >() ) );
//faceCoordinates = CoordinatesType( i, j, k + 1 );
face.setCoordinates( cell.getCoordinates() + CoordinatesType( 0, 0, 1 ) );
face.setOrientation( EntityOrientationType( 0, 0, 1 ) );
faceIndex = grid.getEntityIndex( face );
- CPPUNIT_ASSERT( ( faceIndex == neighbourEntities.template getEntityIndex< 0, 0, 1 >() ) );
+ CPPUNIT_ASSERT( ( faceIndex == neighborEntities.template getEntityIndex< 0, 0, 1 >() ) );
}
}
@@ -428,20 +428,20 @@ class GridTester< 3, RealType, Device, IndexType >: public CppUnit :: TestCase
face.setOrientation( EntityOrientationType( 1, 0, 0 ) );
//const IndexType faceIndex = grid.getEntityIndex( face );
face.refresh();//setIndex( faceIndex );
- auto neighbourEntities = face.template getNeighbourEntities< GridType::Cell::entityDimension >();
+ auto neighborEntities = face.template getNeighborEntities< GridType::Cell::entityDimension >();
if( face.getCoordinates().x() > 0 )
{
CellType cell( grid, face.getCoordinates() + CoordinatesType( -1, 0, 0 ) );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourEntities.template getEntityIndex< -1, 0, 0 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborEntities.template getEntityIndex< -1, 0, 0 >() ) );
}
if( face.getCoordinates().x() < xSize )
{
CellType cell( grid, face.getCoordinates() );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourEntities.template getEntityIndex< 1, 0, 0 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborEntities.template getEntityIndex< 1, 0, 0 >() ) );
}
}
if( face.getCoordinates().x() < xSize && face.getCoordinates().z() < zSize )
@@ -449,19 +449,19 @@ class GridTester< 3, RealType, Device, IndexType >: public CppUnit :: TestCase
face.setOrientation( EntityOrientationType( 0, 1, 0 ) );
//const IndexType faceIndex = grid.getEntityIndex( face );
face.refresh();//setIndex( faceIndex );
- auto neighbourEntities = face.template getNeighbourEntities< GridType::Cell::entityDimension >();
+ auto neighborEntities = face.template getNeighborEntities< GridType::Cell::entityDimension >();
if( face.getCoordinates().y() > 0 )
{
CellType cell( grid, face.getCoordinates() + CoordinatesType( 0, -1, 0 ) );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourEntities.template getEntityIndex< 0, -1, 0 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborEntities.template getEntityIndex< 0, -1, 0 >() ) );
}
if( face.getCoordinates().y() < ySize )
{
CellType cell( grid, face.getCoordinates() );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourEntities.template getEntityIndex< 0, 1, 0 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborEntities.template getEntityIndex< 0, 1, 0 >() ) );
}
}
if( face.getCoordinates().x() < xSize && face.getCoordinates().y() < ySize )
@@ -469,19 +469,19 @@ class GridTester< 3, RealType, Device, IndexType >: public CppUnit :: TestCase
face.setOrientation( EntityOrientationType( 0, 0, 1 ) );
//const IndexType faceIndex = grid.getEntityIndex( face );
face.refresh();//setIndex( faceIndex );
- auto neighbourEntities = face.template getNeighbourEntities< GridType::Cell::entityDimension >();
+ auto neighborEntities = face.template getNeighborEntities< GridType::Cell::entityDimension >();
if( face.getCoordinates().z() > 0 )
{
CellType cell( grid, face.getCoordinates() + CoordinatesType( 0, 0, -1 ) );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourEntities.template getEntityIndex< 0, 0, -1 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborEntities.template getEntityIndex< 0, 0, -1 >() ) );
}
if( face.getCoordinates().z() < zSize )
{
CellType cell( grid, face.getCoordinates() );
IndexType cellIndex = grid.getEntityIndex( cell );
- CPPUNIT_ASSERT( ( cellIndex == neighbourEntities.template getEntityIndex< 0, 0, 1 >() ) );
+ CPPUNIT_ASSERT( ( cellIndex == neighborEntities.template getEntityIndex< 0, 0, 1 >() ) );
}
}
}
diff --git a/tests/unit-tests/operators/CMakeLists.txt b/tests/unit-tests/operators/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/tests/unit-tests/operators/diffusion/CMakeLists.txt b/tests/unit-tests/operators/diffusion/CMakeLists.txt
index 5ad9a0af6f04b4798a5075bd6cd22b6191dd05a0..5d09ac10a1aeb9daca61a35f5075c8b60c1fad5d 100644
--- a/tests/unit-tests/operators/diffusion/CMakeLists.txt
+++ b/tests/unit-tests/operators/diffusion/CMakeLists.txt
@@ -10,8 +10,7 @@ if( BUILD_CUDA )
CUDA_ADD_EXECUTABLE( tnlLinearDiffusionTest-cuda${mpiExt}${debugExt} ${headers} tnlLinearDiffusionTest.cu )
TARGET_LINK_LIBRARIES( tnlLinearDiffusionTest-cuda${mpiExt}${debugExt} ${CPPUNIT_LIBRARIES}
tnl${mpiExt}${debugExt}-0.1 )
- CUDA_ADD_EXECUTABLE( tnlOneSidedMeanCurvatureTest-cuda${mpiExt}${debugExt} ${headers} tnlOneSidedMeanCurvatureTest.cu
- OPTIONS ${CUDA_ADD_EXECUTABLE_OPTIONS} )
+ CUDA_ADD_EXECUTABLE( tnlOneSidedMeanCurvatureTest-cuda${mpiExt}${debugExt} ${headers} tnlOneSidedMeanCurvatureTest.cu )
TARGET_LINK_LIBRARIES( tnlOneSidedMeanCurvatureTest-cuda${mpiExt}${debugExt} ${CPPUNIT_LIBRARIES}
tnl${mpiExt}${debugExt}-0.1 )
diff --git a/tests/unit-tests/operators/fdm/CMakeLists.txt b/tests/unit-tests/operators/fdm/CMakeLists.txt
old mode 100755
new mode 100644
index a85d068b485abd21f1c29e27746c722684d9929e..ad2395de3c40fc3c5c49221a1b023e744f74fceb
--- a/tests/unit-tests/operators/fdm/CMakeLists.txt
+++ b/tests/unit-tests/operators/fdm/CMakeLists.txt
@@ -3,8 +3,7 @@ TARGET_LINK_LIBRARIES( tnlFiniteDifferencesTest${mpiExt}${debugExt} ${CPPUNIT_LI
tnl${mpiExt}${debugExt}-0.1 )
if( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE( tnlFiniteDifferencesTest-cuda${mpiExt}${debugExt} ${headers} tnlFiniteDifferencesTest.cu
- OPTIONS ${CUDA_ADD_EXECUTABLE_OPTIONS} )
+ CUDA_ADD_EXECUTABLE( tnlFiniteDifferencesTest-cuda${mpiExt}${debugExt} ${headers} tnlFiniteDifferencesTest.cu )
TARGET_LINK_LIBRARIES( tnlFiniteDifferencesTest-cuda${mpiExt}${debugExt} ${CPPUNIT_LIBRARIES}
tnl${mpiExt}${debugExt}-0.1 )
endif()
diff --git a/tests/unit-tests/operators/geometric/CMakeLists.txt b/tests/unit-tests/operators/geometric/CMakeLists.txt
index 71d62e19415fc60329923ad0e8a8247c82286800..dac9802609b798141ae4c1ed7fbc8b5ce01f088e 100644
--- a/tests/unit-tests/operators/geometric/CMakeLists.txt
+++ b/tests/unit-tests/operators/geometric/CMakeLists.txt
@@ -11,20 +11,15 @@ TARGET_LINK_LIBRARIES( tnlCoFVMGradientNormTest${mpiExt}${debugExt} ${CPPUNIT_LI
tnl${mpiExt}${debugExt}-0.1 )
if( BUILD_CUDA )
- CUDA_ADD_EXECUTABLE( tnlFDMGradientNormTest-cuda${mpiExt}${debugExt} ${headers} tnlFDMGradientNormTest.cu
- OPTIONS ${CUDA_ADD_EXECUTABLE_OPTIONS} )
+ CUDA_ADD_EXECUTABLE( tnlFDMGradientNormTest-cuda${mpiExt}${debugExt} ${headers} tnlFDMGradientNormTest.cu )
TARGET_LINK_LIBRARIES( tnlFDMGradientNormTest-cuda${mpiExt}${debugExt} ${CPPUNIT_LIBRARIES}
tnl${mpiExt}${debugExt}-0.1 )
- CUDA_ADD_EXECUTABLE( tnlTwoSidedGradientNormTest-cuda${mpiExt}${debugExt} ${headers} tnlTwoSidedGradientNormTest.cu
- OPTIONS ${CUDA_ADD_EXECUTABLE_OPTIONS} )
+ CUDA_ADD_EXECUTABLE( tnlTwoSidedGradientNormTest-cuda${mpiExt}${debugExt} ${headers} tnlTwoSidedGradientNormTest.cu )
TARGET_LINK_LIBRARIES( tnlTwoSidedGradientNormTest-cuda${mpiExt}${debugExt} ${CPPUNIT_LIBRARIES}
tnl${mpiExt}${debugExt}-0.1 )
- CUDA_ADD_EXECUTABLE( tnlCoFVMGradientNormTest-cuda${mpiExt}${debugExt} ${headers} tnlCoFVMGradientNormTest.cu
- OPTIONS ${CUDA_ADD_EXECUTABLE_OPTIONS} )
+ CUDA_ADD_EXECUTABLE( tnlCoFVMGradientNormTest-cuda${mpiExt}${debugExt} ${headers} tnlCoFVMGradientNormTest.cu )
TARGET_LINK_LIBRARIES( tnlCoFVMGradientNormTest-cuda${mpiExt}${debugExt} ${CPPUNIT_LIBRARIES}
tnl${mpiExt}${debugExt}-0.1 )
-
endif()
-
diff --git a/tests/unit-tests/solver/CMakeLists.txt b/tests/unit-tests/solver/CMakeLists.txt
old mode 100755
new mode 100644
diff --git a/tests/unit-tests/solver/pde/CMakeLists.txt b/tests/unit-tests/solver/pde/CMakeLists.txt
old mode 100755
new mode 100644