Composable Kernel: ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument Struct Reference

Argument Struct Reference

Argument Struct Reference#

Composable Kernel: ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument Struct Reference

Public Member Functions | Public Attributes | List of all members

ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument Struct Reference

#include <gridwise_gemm_multiple_d_xdl_cshuffle_lds_direct_load.hpp>

Inheritance diagram for ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument:

Public Member Functions
	Argument (const void p_a_grid, const void p_b_grid, std::array< const void , NumDTensor > p_ds_grid, void p_e_grid, index_t MRaw, index_t NRaw, index_t KRaw, index_t StrideA, index_t StrideB, std::array< index_t, NumDTensor > StrideDs, index_t StrideE, AElementwiseOperation a_element_op, BElementwiseOperation b_element_op, CDEElementwiseOperation cde_element_op)
void	Print () const
Public Member Functions inherited from ck::tensor_operation::device::BaseArgument
	BaseArgument ()=default
	BaseArgument (const BaseArgument &)=default
BaseArgument &	operator= (const BaseArgument &)=default
virtual	~BaseArgument ()

Public Attributes
const ADataType *	p_a_grid_
const BDataType *	p_b_grid_
DsGridPointer	p_ds_grid_
EDataType *	p_e_grid_
AGridDesc_M_K	a_grid_desc_m_k_
BGridDesc_N_K	b_grid_desc_n_k_
DsGridDesc_M_N	ds_grid_desc_m_n_
EGridDesc_M_N	e_grid_desc_m_n_
AGridDesc_AK0_M_AK1	a_grid_desc_ak0_m_ak1_
BGridDesc_BK0_N_BK1	b_grid_desc_bk0_n_bk1_
DsGridDesc_MBlock_MPerBlock_NBlock_NPerBlock	ds_grid_desc_mblock_mperblock_nblock_nperblock_
EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock	e_grid_desc_mblock_mperblock_nblock_nperblock_
Block2ETileMap	block_2_etile_map_
AElementwiseOperation	a_element_op_
BElementwiseOperation	b_element_op_
CDEElementwiseOperation	cde_element_op_
index_t	MRaw_
index_t	NRaw_
index_t	KRaw_
Public Attributes inherited from ck::tensor_operation::device::BaseArgument
void *	p_workspace_ = nullptr

Constructor & Destructor Documentation

◆ Argument()

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::Argument	(	const void *	p_a_grid,
		const void *	p_b_grid,
		std::array< const void *, NumDTensor >	p_ds_grid,
		void *	p_e_grid,
		index_t	MRaw,
		index_t	NRaw,
		index_t	KRaw,
		index_t	StrideA,
		index_t	StrideB,
		std::array< index_t, NumDTensor >	StrideDs,
		index_t	StrideE,
		AElementwiseOperation	a_element_op,
		BElementwiseOperation	b_element_op,
		CDEElementwiseOperation	cde_element_op )

inline

Member Function Documentation

◆ Print()

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

void ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::Print ( ) const

inline

Member Data Documentation

◆ a_element_op_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

AElementwiseOperation ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::a_element_op_

◆ a_grid_desc_ak0_m_ak1_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

AGridDesc_AK0_M_AK1 ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::a_grid_desc_ak0_m_ak1_

◆ a_grid_desc_m_k_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

AGridDesc_M_K ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::a_grid_desc_m_k_

◆ b_element_op_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

BElementwiseOperation ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::b_element_op_

◆ b_grid_desc_bk0_n_bk1_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

BGridDesc_BK0_N_BK1 ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::b_grid_desc_bk0_n_bk1_

◆ b_grid_desc_n_k_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

BGridDesc_N_K ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::b_grid_desc_n_k_

◆ block_2_etile_map_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

Block2ETileMap ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::block_2_etile_map_

◆ cde_element_op_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

CDEElementwiseOperation ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::cde_element_op_

◆ ds_grid_desc_m_n_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

DsGridDesc_M_N ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::ds_grid_desc_m_n_

◆ ds_grid_desc_mblock_mperblock_nblock_nperblock_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

DsGridDesc_MBlock_MPerBlock_NBlock_NPerBlock ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::ds_grid_desc_mblock_mperblock_nblock_nperblock_

◆ e_grid_desc_m_n_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

EGridDesc_M_N ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::e_grid_desc_m_n_

◆ e_grid_desc_mblock_mperblock_nblock_nperblock_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::e_grid_desc_mblock_mperblock_nblock_nperblock_

◆ KRaw_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

index_t ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::KRaw_

◆ MRaw_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

index_t ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::MRaw_

◆ NRaw_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

index_t ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::NRaw_

◆ p_a_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

const ADataType* ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::p_a_grid_

◆ p_b_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

const BDataType* ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::p_b_grid_

◆ p_ds_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

DsGridPointer ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::p_ds_grid_

◆ p_e_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename ELayout, typename ADataType, typename BDataType, typename AComputeDataType_, typename AccDataType, typename CShuffleDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, InMemoryDataOperationEnum EGlobalMemoryDataOperation, tensor_operation::device::GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferScalarPerVector, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferScalarPerVector, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v4, typename BComputeDataType_ = AComputeDataType_>

EDataType* ck::GridwiseGemmMultipleD_Xdl_CShuffle_LdsDirectLoad< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AComputeDataType_, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, EGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferScalarPerVector, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferScalarPerVector, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, BComputeDataType_ >::Argument::p_e_grid_

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

gridwise_gemm_multiple_d_xdl_cshuffle_lds_direct_load.hpp