Composable Kernel: ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem Struct Reference

Problem Struct Reference

Problem Struct Reference#

Composable Kernel: ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem Struct Reference

Public Member Functions | Public Attributes | List of all members

ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem Struct Reference

#include <gridwise_gemm_dpp.hpp>

Inheritance diagram for ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem:

Public Member Functions
__host__	Problem (index_t M_, index_t N_, index_t K_, index_t StrideA_, index_t StrideB_, index_t StrideC_)
__host__ void	Print () const

Public Attributes
index_t	M
index_t	N
index_t	K
index_t	StrideA
index_t	StrideB
index_t	StrideC
index_t	MPadded
index_t	NPadded
index_t	AK0
index_t	BK0

Constructor & Destructor Documentation

◆ Problem()

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

__host__ ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::Problem	(	index_t	M_,
		index_t	N_,
		index_t	K_,
		index_t	StrideA_,
		index_t	StrideB_,
		index_t	StrideC_ )

inline

Member Function Documentation

◆ Print()

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

__host__ void ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::Print ( ) const

inline

Member Data Documentation

◆ AK0

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::AK0

◆ BK0

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::BK0

◆ K

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::K

◆ M

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::M

◆ MPadded

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::MPadded

◆ N

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::N

◆ NPadded

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::NPadded

◆ StrideA

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::StrideA

◆ StrideB

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::StrideB

◆ StrideC

template<index_t BlockSize, typename ABDataType, typename AccDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename ALayout, typename BLayout, typename CLayout, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerDpp, index_t NPerDpp, index_t AK1Value, index_t BK1Value, index_t MDppPerWave, index_t NDppPerWave, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BBlockLdsExtraN, typename CThreadTransferSrcDstAccessOrder, index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, index_t NumGemmKPrefetchStage = 1, PipelineVersion PipelineVer = PipelineVersion::v1>

index_t ck::GridwiseGemm_ak0mak1_bk0nbk1_mn_dpp< BlockSize, ABDataType, AccDataType, CDataType, CGlobalMemoryDataOperation, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, MPerBlock, NPerBlock, KPerBlock, MPerDpp, NPerDpp, AK1Value, BK1Value, MDppPerWave, NDppPerWave, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, NumGemmKPrefetchStage, PipelineVer >::Problem::StrideC

The documentation for this struct was generated from the following file:

gridwise_gemm_dpp.hpp