diff --git a/buildAll b/buildAll index f44119ab3bd3a5061d5e67dde1697908720191ca..245a1d107464179bc9cecbc98685a28a6c6c1896 100755 --- a/buildAll +++ b/buildAll @@ -1,6 +1,9 @@ #!/bin/bash INSTALL_PREFIX=${HOME}/local +WITH_CUDA=yes +CUDA_ARCHITECTURE=2.0 +VERBOSE=0 CPUS=`grep -c processor /proc/cpuinfo` @@ -16,13 +19,13 @@ then fi cd Debug -cmake .. -DCMAKE_BUILD_TYPE=Debug -DCMAKE_INSTALL_PREFIX=${HOME}/local -DCUDA_ARCHITECTURE=1.2 -DWITH_CUDA=no -make -j${CPUS} #VERBOSE=1 +cmake .. -DCMAKE_BUILD_TYPE=Debug -DCMAKE_INSTALL_PREFIX=${HOME}/local -DCUDA_ARCHITECTURE=${CUDA_ARCHITECTURE} -DWITH_CUDA=${WITH_CUDA} +make -j${CPUS} #VERBOSE=${VERBOSE} make -j${CPUS} test make -j${CPUS} install cd ../Release -cmake .. -DCMAKE_INSTALL_PREFIX=${HOME}/local -DCUDA_ARCHITECTURE=1.2 -DWITH_CUDA=no -make -j${CPUS} #VERBOSE=1 +cmake .. -DCMAKE_INSTALL_PREFIX=${HOME}/local -DCUDA_ARCHITECTURE=${CUDA_ARCHITECTURE} -DWITH_CUDA=${WITH_CUDA} +make -j${CPUS} #VERBOSE=${VERBOSE} make -j${CPUS} test make -j${CPUS} install diff --git a/src/matrix/tnlRgCSRMatrix.h b/src/matrix/tnlRgCSRMatrix.h index f4cc933eb878674d0d02a49fc23c6b238a47e2fc..83636c2f44c59321a41080ec7780d180f54bebbf 100644 --- a/src/matrix/tnlRgCSRMatrix.h +++ b/src/matrix/tnlRgCSRMatrix.h @@ -1107,9 +1107,9 @@ __global__ void tnlRgCSRMatrixAdpativeGroupSizeVectorProductKernel( const Index for( Index i = threadIndexInRow; i < nonzeros; i += threadsPerRow ) { const Index column = columns[ pos ]; - if( column == -1 ) - printf( "* rowInMatrix = %d blockIdx. x = %d threadIdx. x = %d threadIndexInRow = %d i = %d \n", - rowInMatrix, blockIndex, threadIdx. x, threadIndexInRow, i ); + //if( column == -1 ) + // printf( "* rowInMatrix = %d blockIdx. x = %d threadIdx. x = %d threadIndexInRow = %d i = %d \n", + // rowInMatrix, blockIndex, threadIdx. x, threadIndexInRow, i ); if( column != -1 ) partialSums[ threadIdx. x ] += nonzeroElements[ pos ] * vec_x[ column ]; //if( rowInMatrix == 0 )