Skip to content
Snippets Groups Projects
Commit 3c9dbc5f authored by Tomáš Oberhuber's avatar Tomáš Oberhuber
Browse files

Added new CSR segments kernels to SpMV benchmark.

parent ac783cf2
No related branches found
No related tags found
1 merge request!83To/matrices
Hide whitespace changes
Inline Side-by-side
src/Benchmarks/SpMV/spmv-legacy.h
+ 33
− 21
View file @ 3c9dbc5f
...@@ -49,7 +49,16 @@ using SlicedEllpackAlias = Matrices::Legacy::SlicedEllpack< Real, Device, Index ...@@ -49,7 +49,16 @@ using SlicedEllpackAlias = Matrices::Legacy::SlicedEllpack< Real, Device, Index
// Segments based sparse matrix aliases // Segments based sparse matrix aliases
template< typename Real, typename Device, typename Index > 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 > template< typename Device, typename Index, typename IndexAllocator >
using EllpackSegments = Algorithms::Segments::Ellpack< Device, Index, IndexAllocator >; using EllpackSegments = Algorithms::Segments::Ellpack< Device, Index, IndexAllocator >;
...@@ -309,26 +318,29 @@ benchmarkSpmvSynthetic( Benchmark& benchmark, ...@@ -309,26 +318,29 @@ benchmarkSpmvSynthetic( Benchmark& benchmark,
benchmark.time< Devices::Cuda >( resetCusparseVectors, "GPU", spmvCusparse, cusparseBenchmarkResults ); benchmark.time< Devices::Cuda >( resetCusparseVectors, "GPU", spmvCusparse, cusparseBenchmarkResults );
#endif #endif
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Scalar >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Scalar >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Vector >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Vector >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light2 >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light2 >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light3 >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light3 >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light4 >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light4 >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light5 >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light5 >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light6 >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Light6 >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Adaptive >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_Adaptive >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_MultiVector>( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrixLegacy_CSR_MultiVector >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrixLegacy_CSR_LightWithoutAtomic>( 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, Matrices::Legacy::Ellpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Vector >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrix_Ellpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Hybrid >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SlicedEllpackAlias >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_CSR_Adaptive >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrix_SlicedEllpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, Matrices::Legacy::Ellpack >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, Matrices::Legacy::ChunkedEllpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_Ellpack >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrix_ChunkedEllpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SlicedEllpackAlias >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, Matrices::Legacy::BiEllpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, SparseMatrix_SlicedEllpack >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrix_BiEllpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, Matrices::Legacy::ChunkedEllpack >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrix_ChunkedEllpack >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, Matrices::Legacy::BiEllpack >( benchmark, hostOutVector, inputFileName, verboseMR );
benchmarkSpMV< Real, SparseMatrix_BiEllpack >( benchmark, hostOutVector, inputFileName, verboseMR );
/* AdEllpack is broken /* AdEllpack is broken
benchmarkSpMV< Real, Matrices::AdEllpack >( benchmark, hostOutVector, inputFileName, verboseMR ); benchmarkSpMV< Real, Matrices::AdEllpack >( benchmark, hostOutVector, inputFileName, verboseMR );
*/ */
......
src/TNL/Algorithms/Segments/details/CSRAdaptiveKernelBlockDescriptor.h
+ 1
− 1
View file @ 3c9dbc5f
...@@ -185,7 +185,7 @@ struct CSRAdaptiveKernelBlockDescriptor ...@@ -185,7 +185,7 @@ struct CSRAdaptiveKernelBlockDescriptor
//return ( twobytes[ sizeof( Index ) == 4 ? 3 : 5 ] & 0x3FFF ); //return ( twobytes[ sizeof( Index ) == 4 ? 3 : 5 ] & 0x3FFF );
} }
__cuda_callable__ const uint8_t getWarpIdx() const __cuda_callable__ uint8_t getWarpIdx() const
{ {
return this->warpIdx; return this->warpIdx;
} }
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment