Argument Struct Reference

Argument Struct Reference#

Composable Kernel: ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument Struct Reference
Public Member Functions | Public Attributes | List of all members
ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument Struct Reference

#include <device_gemm_multiple_d_layernorm_wmma_cshuffle_v3.hpp>

Inheritance diagram for ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument:
ck::tensor_operation::device::BaseArgument

Public Member Functions

 Argument (const void *p_a_grid, const void *p_b_grid, std::array< const void *, NumDTensor > p_ds_grid, const void *p_gamma_grid, const void *p_beta_grid, void *p_h_grid, index_t MRaw, index_t NRaw, index_t KRaw, index_t StrideA, index_t StrideB, std::array< index_t, NumDTensor > StrideDs, index_t StrideH, double epsilon, AElementwiseOperation a_element_op, BElementwiseOperation b_element_op, CDEElementwiseOperation cde_element_op, HElementwiseOperation h_element_op)
Public Member Functions inherited from ck::tensor_operation::device::BaseArgument
 BaseArgument ()=default
 BaseArgument (const BaseArgument &)=default
BaseArgument & operator= (const BaseArgument &)=default
virtual ~BaseArgument ()

Public Attributes

const ADataType * p_a_grid_
const BDataType * p_b_grid_
std::array< const void *, NumDTensor > p_ds_grid_
void * p_workspace_e_grid_
void * p_workspace_mean_
void * p_workspace_var_
void * p_workspace_count_
const GammaDataType * p_gamma_grid_
const BetaDataType * p_beta_grid_
HDataType * p_h_grid_
EHGridDesc_M_N layernorm_e_grid_desc_m_n_
LayernormMeanVarGridDesc_M_NBlock layernorm_mean_var_grid_desc_m_nblock_
LayernormCountGridDesc_M_NBlock layernorm_count_grid_desc_m_nblock_
GammaBetaGridDesc_N gamma_grid_desc_n_
GammaBetaGridDesc_N beta_grid_desc_n_
EHGridDesc_M_N h_grid_desc_m_n_
AElementwiseOperation a_element_op_
BElementwiseOperation b_element_op_
CDEElementwiseOperation cde_element_op_
HElementwiseOperation h_element_op_
index_t MRaw_
index_t NRaw_
index_t KRaw_
index_t StrideA_
index_t StrideB_
std::array< index_t, NumDTensor > StrideDs_
index_t StrideH_
index_t gemm_nblock_
AccDataType epsilon_
Public Attributes inherited from ck::tensor_operation::device::BaseArgument
void * p_workspace_ = nullptr

Constructor & Destructor Documentation

◆ Argument()

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::Argument ( const void * p_a_grid,
const void * p_b_grid,
std::array< const void *, NumDTensor > p_ds_grid,
const void * p_gamma_grid,
const void * p_beta_grid,
void * p_h_grid,
index_t MRaw,
index_t NRaw,
index_t KRaw,
index_t StrideA,
index_t StrideB,
std::array< index_t, NumDTensor > StrideDs,
index_t StrideH,
double epsilon,
AElementwiseOperation a_element_op,
BElementwiseOperation b_element_op,
CDEElementwiseOperation cde_element_op,
HElementwiseOperation h_element_op )
inline

Member Data Documentation

◆ a_element_op_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
AElementwiseOperation ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::a_element_op_

◆ b_element_op_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
BElementwiseOperation ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::b_element_op_

◆ beta_grid_desc_n_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
GammaBetaGridDesc_N ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::beta_grid_desc_n_

◆ cde_element_op_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
CDEElementwiseOperation ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::cde_element_op_

◆ epsilon_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
AccDataType ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::epsilon_

◆ gamma_grid_desc_n_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
GammaBetaGridDesc_N ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::gamma_grid_desc_n_

◆ gemm_nblock_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
index_t ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::gemm_nblock_

◆ h_element_op_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
HElementwiseOperation ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::h_element_op_

◆ h_grid_desc_m_n_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
EHGridDesc_M_N ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::h_grid_desc_m_n_

◆ KRaw_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
index_t ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::KRaw_

◆ layernorm_count_grid_desc_m_nblock_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
LayernormCountGridDesc_M_NBlock ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::layernorm_count_grid_desc_m_nblock_

◆ layernorm_e_grid_desc_m_n_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
EHGridDesc_M_N ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::layernorm_e_grid_desc_m_n_

◆ layernorm_mean_var_grid_desc_m_nblock_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
LayernormMeanVarGridDesc_M_NBlock ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::layernorm_mean_var_grid_desc_m_nblock_

◆ MRaw_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
index_t ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::MRaw_

◆ NRaw_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
index_t ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::NRaw_

◆ p_a_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
const ADataType* ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_a_grid_

◆ p_b_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
const BDataType* ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_b_grid_

◆ p_beta_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
const BetaDataType* ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_beta_grid_

◆ p_ds_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
std::array<const void*, NumDTensor> ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_ds_grid_

◆ p_gamma_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
const GammaDataType* ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_gamma_grid_

◆ p_h_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
HDataType* ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_h_grid_

◆ p_workspace_count_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
void* ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_workspace_count_

◆ p_workspace_e_grid_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
void* ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_workspace_e_grid_

◆ p_workspace_mean_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
void* ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_workspace_mean_

◆ p_workspace_var_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
void* ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::p_workspace_var_

◆ StrideA_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
index_t ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::StrideA_

◆ StrideB_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
index_t ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::StrideB_

◆ StrideDs_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
std::array<index_t, NumDTensor> ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::StrideDs_

◆ StrideH_

template<typename ALayout, typename BLayout, typename DsLayout, typename HLayout, typename ADataType, typename BDataType, typename DsDataType, typename HDataType, typename AccDataType, typename CShuffleDataType, typename EMeanVarDataType, typename GammaDataType, typename BetaDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CDEElementwiseOperation, typename HElementwiseOperation, GemmSpecialization GemmSpec, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, index_t MPerWmma, index_t NPerWmma, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CDEShuffleBlockTransferScalarPerVector, typename LayernormThreadClusterSize_M_N, index_t LayernormThreadSliceSize_M, BlockGemmPipelineScheduler BlkGemmPipeSched = BlockGemmPipelineScheduler::Intrawave, BlockGemmPipelineVersion BlkGemmPipelineVer = BlockGemmPipelineVersion::v1, typename ComputeTypeA = HDataType, typename ComputeTypeB = ComputeTypeA, bool PermuteA = false, bool PermuteB = false>
index_t ck::tensor_operation::device::DeviceGemmMultipleDLayernorm_Wmma_CShuffleV3< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, DsDataType, HDataType, AccDataType, CShuffleDataType, EMeanVarDataType, GammaDataType, BetaDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, HElementwiseOperation, GemmSpec, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CDEShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVector, LayernormThreadClusterSize_M_N, LayernormThreadSliceSize_M, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >::Argument::StrideH_
The documentation for this struct was generated from the following file: