Loading src/Benchmarks/SpMV/spmv-legacy.h +33 −21 Original line number Diff line number Diff line Loading @@ -49,7 +49,16 @@ using SlicedEllpackAlias = Matrices::Legacy::SlicedEllpack< Real, Device, Index // Segments based sparse matrix aliases template< typename Real, typename Device, typename Index > using SparseMatrix_CSR = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRDefault >; using SparseMatrix_CSR_Scalar = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRScalar >; template< typename Real, typename Device, typename Index > using SparseMatrix_CSR_Vector = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRVector >; template< typename Real, typename Device, typename Index > using SparseMatrix_CSR_Hybrid = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRHybrid >; template< typename Real, typename Device, typename Index > using SparseMatrix_CSR_Adaptive = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRAdaptive >; template< typename Device, typename Index, typename IndexAllocator > using EllpackSegments = Algorithms::Segments::Ellpack< Device, Index, IndexAllocator >; Loading Loading @@ -320,7 +329,10 @@ benchmarkSpmvSynthetic( Benchmark& benchmark, benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Adaptive >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_MultiVector >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_LightWithoutAtomic >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Scalar >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Vector >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Hybrid >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Adaptive >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, Matrices::Legacy::Ellpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_Ellpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SlicedEllpackAlias >( benchmark, hostOutVector, inputFileName, verboseMR ); Loading src/TNL/Algorithms/Segments/details/CSRAdaptiveKernelBlockDescriptor.h +1 −1 Original line number Diff line number Diff line Loading @@ -185,7 +185,7 @@ struct CSRAdaptiveKernelBlockDescriptor //return ( twobytes[ sizeof( Index ) == 4 ? 3 : 5 ] & 0x3FFF ); } __cuda_callable__ const uint8_t getWarpIdx() const __cuda_callable__ uint8_t getWarpIdx() const { return this->warpIdx; } Loading Loading
src/Benchmarks/SpMV/spmv-legacy.h +33 −21 Original line number Diff line number Diff line Loading @@ -49,7 +49,16 @@ using SlicedEllpackAlias = Matrices::Legacy::SlicedEllpack< Real, Device, Index // Segments based sparse matrix aliases template< typename Real, typename Device, typename Index > using SparseMatrix_CSR = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRDefault >; using SparseMatrix_CSR_Scalar = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRScalar >; template< typename Real, typename Device, typename Index > using SparseMatrix_CSR_Vector = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRVector >; template< typename Real, typename Device, typename Index > using SparseMatrix_CSR_Hybrid = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRHybrid >; template< typename Real, typename Device, typename Index > using SparseMatrix_CSR_Adaptive = Matrices::SparseMatrix< Real, Device, Index, Matrices::GeneralMatrix, Algorithms::Segments::CSRAdaptive >; template< typename Device, typename Index, typename IndexAllocator > using EllpackSegments = Algorithms::Segments::Ellpack< Device, Index, IndexAllocator >; Loading Loading @@ -320,7 +329,10 @@ benchmarkSpmvSynthetic( Benchmark& benchmark, benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Adaptive >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_MultiVector >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_LightWithoutAtomic >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Scalar >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Vector >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Hybrid >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Adaptive >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, Matrices::Legacy::Ellpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_Ellpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SlicedEllpackAlias >( benchmark, hostOutVector, inputFileName, verboseMR ); Loading
src/TNL/Algorithms/Segments/details/CSRAdaptiveKernelBlockDescriptor.h +1 −1 Original line number Diff line number Diff line Loading @@ -185,7 +185,7 @@ struct CSRAdaptiveKernelBlockDescriptor //return ( twobytes[ sizeof( Index ) == 4 ? 3 : 5 ] & 0x3FFF ); } __cuda_callable__ const uint8_t getWarpIdx() const __cuda_callable__ uint8_t getWarpIdx() const { return this->warpIdx; } Loading