Commit 9a01097d authored by Tomáš Oberhuber's avatar Tomáš Oberhuber
Browse files

Implementing CUDA parallel reduction.

parent a37a9195
Loading
Loading
Loading
Loading
+4 −3
Original line number Diff line number Diff line
TODO: implementovat tridu tnlFileName pro generovani jmen souboru

TODO: metodu pro tnlString pro nahrazeni napr. podretezce XXXXX indexem 00001 tj. uXXXXX.bin -> u00001.bin
      to by melo byt robustnejsi, nez doposavadni pristup 

TODO: implementovat tridu tnlParabolicSolver pro odvozovani resicu k casove promennym uloham

TODO: Nahradit mGrid2D, mGrid3D za mGrid obecne dimenze
@@ -8,6 +11,4 @@ TODO: zavets iteratory pres uzle site misto for cyklu

TODO: implementovat Mersona v CUDA

TODO: metoda Test do tnlObject

TODO: trida tnlTester pro rizeni testu   
 No newline at end of file
TODO: objekt pro osetreni chyb - zavedeni funkce tnlGetError   
 No newline at end of file
+22 −1
Original line number Diff line number Diff line
@@ -37,6 +37,12 @@ AC_ARG_WITH(cuda_libdir,
            AS_HELP_STRING([--with-cuda-libdir],
                           [says where the CUDA libraries can be found, default is /usr/local/cuda/lib]),
            CUDA_LIBS=$withval)
AC_ARG_WITH(cuda_arch,
            AS_HELP_STRING([--with-cuda-arch],
                           [specifies the CUDA architecture, can be 1.0, 1.1, 1.2 or 1.3 - default is 1.3]),
            CUDA_ARCH=$withval,
            CUDA_ARCH="1.3")

working_nvcc="no"  
if test x$with_cuda = xyes;
then
@@ -89,6 +95,21 @@ then
      CUDA_LDFLAGS="$CUDA_LDFLAGS -lcudart"
      CC="nvcc"
      CXX="nvcc"
      case "$CUDA_ARCH"  in
         1.0 )
            CUDA_CXXFLAGS="$CUDA_CXXFLAGS -arch=sm_10"
         ;;
         1.1 )
            CUDA_CXXFLAGS="$CUDA_CXXFLAGS -arch=sm_11"
         ;;
         1.2 )
            CUDA_CXXFLAGS="$CUDA_CXXFLAGS -arch=sm_12"
         ;;
         1.3 )
            CUDA_CXXFLAGS="$CUDA_CXXFLAGS -arch=sm_13"
         ;;
      esac  
      DBGCXXFLAGS="$DBGCXXFLAGS -deviceemu"    
   else
      CUDA_LDFLAGS=""
      CUDA_CXXFLAGS=""
@@ -121,7 +142,7 @@ dnl ----------- check for debug--------------
dnl -----------------------------------------
AC_ARG_ENABLE(debug,[  --enable-debug=[no/yes]	turn on debugging [default=no]] )
if test x"$enable_debug" = xyes; then
   DBGCXXFLAGS="-O0 -DDEBUG"
   DBGCXXFLAGS="$DBGCXXFLAGS -O0 -DDEBUG"
   if test x$CXX != xnvcc;
   then
      DBGCXXFLAGS="$DBGCXXFLAGS -g3 -Wall -W -ansi -Wno-unused"
+386 −110

File changed.

Preview size limit exceeded, changes collapsed.

+884 −167

File changed.

Preview size limit exceeded, changes collapsed.

+235 −59

File changed.

Preview size limit exceeded, changes collapsed.

Loading