Composable Kernel: ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument Struct Reference

Argument Struct Reference

Argument Struct Reference#

Composable Kernel: ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument Struct Reference

Public Member Functions | Public Attributes | List of all members

ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument Struct Reference

#include <device_batched_gemm_softmax_gemm_permute_wmma_cshuffle.hpp>

Inheritance diagram for ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument:

Public Member Functions
	Argument (const ADataType p_a_grid, const B0DataType p_b0_grid, const B1DataType p_b1_grid, CDataType p_c_grid, const std::array< void , NumAcc0Bias > p_acc0_biases, const std::array< void , NumAcc1Bias > p_acc1_biases, const std::array< index_t, NumDimG+NumDimM+NumDimN > &a_gs_ms_ks_lengths, const std::array< index_t, NumDimG+NumDimM+NumDimN > &a_gs_ms_ks_strides, const std::array< index_t, NumDimG+NumDimM+NumDimN > &b0_gs_ls_ks_lengths, const std::array< index_t, NumDimG+NumDimM+NumDimN > &b0_gs_ls_ks_strides, const std::array< index_t, NumDimG+NumDimM+NumDimN > &b1_gs_ns_ls_lengths, const std::array< index_t, NumDimG+NumDimM+NumDimN > &b1_gs_ns_ls_strides, const std::array< index_t, NumDimG+NumDimM+NumDimN > &c_gs_ms_ns_lengths, const std::array< index_t, NumDimG+NumDimM+NumDimN > &c_gs_ms_ns_strides, const std::array< std::vector< ck::index_t >, NumAcc0Bias > acc0_biases_gs_ms_ls_lengths, const std::array< std::vector< ck::index_t >, NumAcc0Bias > acc0_biases_gs_ms_ls_strides, const std::array< std::vector< ck::index_t >, NumAcc1Bias > acc1_biases_gs_ms_ns_lengths, const std::array< std::vector< ck::index_t >, NumAcc1Bias > acc1_biases_gs_ms_ns_strides, const index_t M01, const index_t N01, AElementwiseOperation a_element_op, B0ElementwiseOperation b0_element_op, AccElementwiseOperation acc_element_op, B1ElementwiseOperation b1_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 B0DataType *	p_b0_grid_
const B1DataType *	p_b1_grid_
CDataType *	p_c_grid_
AGridDesc	a_grid_desc
B0GridDesc	b0_grid_desc
B1GridDesc	b1_grid_desc
CGridDesc_M_N	c_grid_desc_m_n_
AGridDesc_G_M_K	a_grid_desc_g_m_k_
B0GridDesc_G_L_K	b0_grid_desc_g_l_k_
B1GridDesc_G_N_L	b1_grid_desc_g_n_l_
CGridDesc_G_M_N	c_grid_desc_g_m_n_
GridwiseOp::CGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock	c_grid_desc_mblock_mperblock_nblock_nperblock_
GridwiseOp::DefaultBlock2CTileMap	block_2_ctile_map_
AElementwiseOperation	a_element_op_
B0ElementwiseOperation	b0_element_op_
AccElementwiseOperation	acc_element_op_
B1ElementwiseOperation	b1_element_op_
CElementwiseOperation	c_element_op_
C0MatrixMask	c0_matrix_mask_
std::array< index_t, NumDimG+NumDimM+NumDimN >	raw_lengths_mz_lz_kz_nz_
std::array< index_t, NumDimG+NumDimM+NumDimN >	a_mz_kz_strides_
std::array< index_t, NumDimG+NumDimM+NumDimN >	b0_lz_kz_strides_
std::array< index_t, NumDimG+NumDimM+NumDimN >	b1_nz_lz_strides_
std::array< index_t, NumDimG+NumDimM+NumDimN >	c_mz_nz_strides_
index_t	batch_count_
ComputeBasePtrOfStridedBatch	compute_ptr_offset_of_batch_
Public Attributes inherited from ck::tensor_operation::device::BaseArgument
void *	p_workspace_ = nullptr

Constructor & Destructor Documentation

◆ Argument()

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::Argument	(	const ADataType *	p_a_grid,
		const B0DataType *	p_b0_grid,
		const B1DataType *	p_b1_grid,
		CDataType *	p_c_grid,
		const std::array< void *, NumAcc0Bias >	p_acc0_biases,
		const std::array< void *, NumAcc1Bias >	p_acc1_biases,
		const std::array< index_t, NumDimG+NumDimM+NumDimN > &	a_gs_ms_ks_lengths,
		const std::array< index_t, NumDimG+NumDimM+NumDimN > &	a_gs_ms_ks_strides,
		const std::array< index_t, NumDimG+NumDimM+NumDimN > &	b0_gs_ls_ks_lengths,
		const std::array< index_t, NumDimG+NumDimM+NumDimN > &	b0_gs_ls_ks_strides,
		const std::array< index_t, NumDimG+NumDimM+NumDimN > &	b1_gs_ns_ls_lengths,
		const std::array< index_t, NumDimG+NumDimM+NumDimN > &	b1_gs_ns_ls_strides,
		const std::array< index_t, NumDimG+NumDimM+NumDimN > &	c_gs_ms_ns_lengths,
		const std::array< index_t, NumDimG+NumDimM+NumDimN > &	c_gs_ms_ns_strides,
		const std::array< std::vector< ck::index_t >, NumAcc0Bias >	acc0_biases_gs_ms_ls_lengths,
		const std::array< std::vector< ck::index_t >, NumAcc0Bias >	acc0_biases_gs_ms_ls_strides,
		const std::array< std::vector< ck::index_t >, NumAcc1Bias >	acc1_biases_gs_ms_ns_lengths,
		const std::array< std::vector< ck::index_t >, NumAcc1Bias >	acc1_biases_gs_ms_ns_strides,
		const index_t	M01,
		const index_t	N01,
		AElementwiseOperation	a_element_op,
		B0ElementwiseOperation	b0_element_op,
		AccElementwiseOperation	acc_element_op,
		B1ElementwiseOperation	b1_element_op,
		CElementwiseOperation	c_element_op )

inline

Member Data Documentation

◆ a_element_op_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

AElementwiseOperation ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::a_element_op_

◆ a_grid_desc

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

AGridDesc ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::a_grid_desc

◆ a_grid_desc_g_m_k_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

AGridDesc_G_M_K ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::a_grid_desc_g_m_k_

◆ a_mz_kz_strides_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

std::array<index_t, NumDimG + NumDimM + NumDimN> ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::a_mz_kz_strides_

◆ acc_element_op_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

AccElementwiseOperation ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::acc_element_op_

◆ b0_element_op_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

B0ElementwiseOperation ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::b0_element_op_

◆ b0_grid_desc

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

B0GridDesc ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::b0_grid_desc

◆ b0_grid_desc_g_l_k_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

B0GridDesc_G_L_K ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::b0_grid_desc_g_l_k_

◆ b0_lz_kz_strides_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

std::array<index_t, NumDimG + NumDimM + NumDimN> ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::b0_lz_kz_strides_

◆ b1_element_op_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

B1ElementwiseOperation ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::b1_element_op_

◆ b1_grid_desc

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

B1GridDesc ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::b1_grid_desc

◆ b1_grid_desc_g_n_l_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

B1GridDesc_G_N_L ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::b1_grid_desc_g_n_l_

◆ b1_nz_lz_strides_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

std::array<index_t, NumDimG + NumDimM + NumDimN> ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::b1_nz_lz_strides_

◆ batch_count_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

index_t ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::batch_count_

◆ block_2_ctile_map_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

GridwiseOp::DefaultBlock2CTileMap ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::block_2_ctile_map_

◆ c0_matrix_mask_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

C0MatrixMask ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::c0_matrix_mask_

◆ c_element_op_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

CElementwiseOperation ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::c_element_op_

◆ c_grid_desc_g_m_n_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

CGridDesc_G_M_N ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::c_grid_desc_g_m_n_

◆ c_grid_desc_m_n_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

CGridDesc_M_N ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::c_grid_desc_m_n_

◆ c_grid_desc_mblock_mperblock_nblock_nperblock_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

GridwiseOp::CGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::c_grid_desc_mblock_mperblock_nblock_nperblock_

◆ c_mz_nz_strides_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

std::array<index_t, NumDimG + NumDimM + NumDimN> ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::c_mz_nz_strides_

◆ compute_ptr_offset_of_batch_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

ComputeBasePtrOfStridedBatch ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::compute_ptr_offset_of_batch_

◆ p_a_grid_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

const ADataType* ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::p_a_grid_

◆ p_b0_grid_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

const B0DataType* ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::p_b0_grid_

◆ p_b1_grid_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

const B1DataType* ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::p_b1_grid_

◆ p_c_grid_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

CDataType* ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::p_c_grid_

◆ raw_lengths_mz_lz_kz_nz_

template<index_t NumDimG, index_t NumDimM, index_t NumDimL, index_t NumDimK, index_t NumDimN, typename ADataType, typename B0DataType, typename B1DataType, typename CDataType, typename Acc0BiasDataType, typename Acc0DataType, typename Acc1BiasDataType, typename Acc1DataType, typename CShuffleDataType, typename AElementwiseOperation, typename B0ElementwiseOperation, typename AccElementwiseOperation, typename B1ElementwiseOperation, typename CElementwiseOperation, GemmSpecialization GemmSpec, TensorSpecialization ASpec, TensorSpecialization B0Spec, TensorSpecialization B1Spec, TensorSpecialization CSpec, ck::index_t NumPrefetch, ck::index_t BlockSize, ck::index_t MPerBlock, ck::index_t LPerBlock, ck::index_t KPerBlock, ck::index_t AK1, ck::index_t BK1, ck::index_t NPerBlock, ck::index_t LTilePerBlock, ck::index_t L1, ck::index_t MPerWmma, ck::index_t LPerWmma, ck::index_t NPerWmma, ck::index_t MRepeat, ck::index_t LRepeat, ck::index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, ck::index_t ABlockTransferSrcVectorDim, ck::index_t ABlockTransferSrcScalarPerVector, ck::index_t ABlockTransferDstScalarPerVector_K1, bool ABlockLdsAddExtraM, typename B0BlockTransferThreadClusterLengths_K0_L_K1, typename B0BlockTransferThreadClusterArrangeOrder, typename B0BlockTransferSrcAccessOrder, ck::index_t B0BlockTransferSrcVectorDim, ck::index_t B0BlockTransferSrcScalarPerVector, ck::index_t B0BlockTransferDstScalarPerVector_K1, bool B0BlockLdsAddExtraL, typename B1BlockTransferThreadClusterLengths_L0_N_L1, typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferSrcAccessOrder, ck::index_t B1BlockTransferSrcVectorDim, ck::index_t B1BlockTransferSrcScalarPerVector, ck::index_t B1BlockTransferDstScalarPerVector_L1, bool B1BlockLdsAddExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpecialization MaskingSpec, ck::LoopScheduler LoopSched = make_default_loop_scheduler(), ck::PipelineVersion PipelineVer = ck::PipelineVersion::v1>

std::array<index_t, NumDimG + NumDimM + NumDimN> ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::Argument::raw_lengths_mz_lz_kz_nz_

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

device_batched_gemm_softmax_gemm_permute_wmma_cshuffle.hpp