Argument Struct Reference

Argument Struct Reference#

Composable Kernel: ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument Struct Reference

Public Member Functions | Public Attributes | List of all members

ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument Struct Reference

#include <device_gemm_dl.hpp>

Inheritance diagram for ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument:

Public Member Functions
	Argument (const ADataType p_a_grid, const BDataType p_b_grid, CDataType *p_c_grid, index_t M, index_t N, index_t K, index_t StrideA, index_t StrideB, index_t StrideC, index_t M01, index_t N01, AElementwiseOperation a_element_op, BElementwiseOperation b_element_op, CElementwiseOperation c_element_op)

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_

CDataType *	p_c_grid_

AGridDesc_K0_M_K1	a_grid_desc_k0_m_k1_

BGridDesc_K0_N_K1	b_grid_desc_k0_n_k1_

CGridDesc_M_N	c_grid_desc_m_n_

AGridDesc_K0_M0_M1_K1	a_grid_desc_k0_m0_m1_k1_

BGridDesc_K0_N0_N1_K1	b_grid_desc_k0_n0_n1_k1_

CGridDesc_M0_M10_M11_N0_N10_N11	c_grid_desc_m0_m10_m11_n0_n10_n11_

DefaultBlock2CTileMap	block_2_ctile_map_

index_t	M01_

index_t	N01_

index_t	M_raw_

index_t	N_raw_

index_t	K_raw_

AElementwiseOperation	a_element_op_

BElementwiseOperation	b_element_op_

CElementwiseOperation	c_element_op_

Public Attributes inherited from ck::tensor_operation::device::BaseArgument
void *	p_workspace_ = nullptr

Constructor & Destructor Documentation

◆ Argument()

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::Argument	(	const ADataType *	p_a_grid,
		const BDataType *	p_b_grid,
		CDataType *	p_c_grid,
		index_t	M,
		index_t	N,
		index_t	K,
		index_t	StrideA,
		index_t	StrideB,
		index_t	StrideC,
		index_t	M01,
		index_t	N01,
		AElementwiseOperation	a_element_op,
		BElementwiseOperation	b_element_op,
		CElementwiseOperation	c_element_op
	)

inline

Member Data Documentation

◆ a_element_op_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

AElementwiseOperation ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::a_element_op_

◆ a_grid_desc_k0_m0_m1_k1_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

AGridDesc_K0_M0_M1_K1 ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::a_grid_desc_k0_m0_m1_k1_

◆ a_grid_desc_k0_m_k1_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

AGridDesc_K0_M_K1 ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::a_grid_desc_k0_m_k1_

◆ b_element_op_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

BElementwiseOperation ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::b_element_op_

◆ b_grid_desc_k0_n0_n1_k1_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

BGridDesc_K0_N0_N1_K1 ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::b_grid_desc_k0_n0_n1_k1_

◆ b_grid_desc_k0_n_k1_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

BGridDesc_K0_N_K1 ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::b_grid_desc_k0_n_k1_

◆ block_2_ctile_map_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

DefaultBlock2CTileMap ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::block_2_ctile_map_

◆ c_element_op_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

CElementwiseOperation ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::c_element_op_

◆ c_grid_desc_m0_m10_m11_n0_n10_n11_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

CGridDesc_M0_M10_M11_N0_N10_N11 ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::c_grid_desc_m0_m10_m11_n0_n10_n11_

◆ c_grid_desc_m_n_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

CGridDesc_M_N ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::c_grid_desc_m_n_

◆ K_raw_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

index_t ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::K_raw_

◆ M01_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

index_t ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::M01_

◆ M_raw_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

index_t ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::M_raw_

◆ N01_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

index_t ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::N01_

◆ N_raw_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

index_t ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::N_raw_

◆ p_a_grid_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

const ADataType* ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::p_a_grid_

◆ p_b_grid_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

const BDataType* ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::p_b_grid_

◆ p_c_grid_

template<typename ADataType , typename BDataType , typename CDataType , typename AccDataType , typename ALayout , typename BLayout , typename CLayout , typename AElementwiseOperation , typename BElementwiseOperation , typename CElementwiseOperation , GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t K0PerBlock, index_t K1, index_t M1PerThread, index_t N1PerThread, index_t KPerThread, typename M1N1ThreadClusterM1Xs , typename M1N1ThreadClusterN1Xs , typename ABlockTransferThreadSliceLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterLengths_K0_M0_M1_K1 , typename ABlockTransferThreadClusterArrangeOrder , typename ABlockTransferSrcAccessOrder , typename ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1 , typename ABlockTransferSrcVectorTensorContiguousDimOrder , typename ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1 , typename BBlockTransferThreadSliceLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterLengths_K0_N0_N1_K1 , typename BBlockTransferThreadClusterArrangeOrder , typename BBlockTransferSrcAccessOrder , typename BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1 , typename BBlockTransferSrcVectorTensorContiguousDimOrder , typename BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1 , typename CThreadTransferSrcDstAccessOrder , index_t CThreadTransferSrcDstVectorDim, index_t CThreadTransferDstScalarPerVector, enable_if_t< is_same_v< AElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< BElementwiseOperation, ck::tensor_operation::element_wise::PassThrough > &&is_same_v< CElementwiseOperation, ck::tensor_operation::element_wise::PassThrough >, bool > = false>

CDataType* ck::tensor_operation::device::DeviceGemmDl< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, K0PerBlock, K1, M1PerThread, N1PerThread, KPerThread, M1N1ThreadClusterM1Xs, M1N1ThreadClusterN1Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector, >::Argument::p_c_grid_

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

/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/docs-6.4.3/include/ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp