Commit e1ad4cb0 authored by Tomáš Oberhuber's avatar Tomáš Oberhuber
Browse files

Fixing CUDA compilation.

parent d100e4cb

CMakeLists.txt

+6 −8
Original line number Diff line number Diff line
@@ -33,24 +33,22 @@ if( CMAKE_BUILD_TYPE STREQUAL "Debug") 
    set( LIBRARY_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Debug/lib )

    set( EXECUTABLE_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Debug/bin )

    set( debugExt -dbg )

    AddCompilerFlag( "-std=c++11 -g -rdynamic" )

else()

    set( PROJECT_BUILD_PATH ${PROJECT_SOURCE_DIR}/Release/src )

    set( PROJECT_TESTS_PATH ${PROJECT_SOURCE_DIR}/Release/tests )

    set( PROJECT_TOOLS_PATH ${PROJECT_SOURCE_DIR}/Release/tools )

    set( LIBRARY_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Release/lib)

    set( EXECUTABLE_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/Release/bin)

    AddCompilerFag( "-std=c++11 -O3 -march=native -DNDEBUG -g" )

endif()



# set Debug/Release options

#set( CMAKE_CXX_FLAGS "-std=c++11" )

#set( CMAKE_CXX_FLAGS_DEBUG "-g" )

#set( CMAKE_CXX_FLAGS_RELEASE  )

set( CMAKE_CXX_FLAGS "-std=c++11" )

set( CMAKE_CXX_FLAGS_DEBUG "-g" )

set( CMAKE_CXX_FLAGS_RELEASE "-O3 -march=native -DNDEBUG" )

# pass -rdynamic only in Debug mode

#set( CMAKE_SHARED_LIBRARY_LINK_C_FLAGS "" )

#set( CMAKE_SHARED_LIBRARY_LINK_C_FLAGS_DEBUG "" )

#set( CMAKE_SHARED_LIBRARY_LINK_C_FLAGS_RELEASE "" )

set( CMAKE_SHARED_LIBRARY_LINK_C_FLAGS "" )

set( CMAKE_SHARED_LIBRARY_LINK_C_FLAGS_DEBUG "-rdynamic" )

set( CMAKE_SHARED_LIBRARY_LINK_C_FLAGS_RELEASE "" )



get_filename_component( CXX_COMPILER_NAME ${CMAKE_CXX_COMPILER} NAME )

if( CXX_COMPILER_NAME MATCHES "icpc" )

tests/benchmarks/heat-equation-benchmark/BenchmarkLaplace.h

+3 −3
Original line number Diff line number Diff line
@@ -35,8 +35,8 @@ class BenchmarkLaplace< tnlGrid< 1,MeshReal, Device, MeshIndex >, Real, Index > 
                       const MeshEntity& entity,

                       const RealType& time = 0.0 ) const;



      __cuda_callable__

      template< typename MeshEntity >

      __cuda_callable__

      Index getLinearSystemRowLength( const MeshType& mesh,

                                      const IndexType& index,

                                      const MeshEntity& entity ) const;
@@ -77,8 +77,8 @@ class BenchmarkLaplace< tnlGrid< 2,MeshReal, Device, MeshIndex >, Real, Index > 
                       const MeshEntity& entity,

                       const RealType& time = 0.0 ) const;



      __cuda_callable__

      template< typename MeshEntity >

      __cuda_callable__

      Index getLinearSystemRowLength( const MeshType& mesh,

                                      const IndexType& index,

                                      const MeshEntity& entity ) const;
@@ -119,8 +119,8 @@ class BenchmarkLaplace< tnlGrid< 3,MeshReal, Device, MeshIndex >, Real, Index > 
                       const MeshEntity& entity,

                       const RealType& time = 0.0 ) const;



      __cuda_callable__

      template< typename MeshEntity >

      __cuda_callable__

      Index getLinearSystemRowLength( const MeshType& mesh,

                                      const IndexType& index,

                                      const MeshEntity& entity ) const;