GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer > Struct Template Reference

GridwiseFpAintBGemm_Wmma&lt; BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer &gt; Struct Template Reference#

Composable Kernel: ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer > Struct Template Reference
Classes | Public Types | Static Public Member Functions | Static Public Attributes | List of all members
ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer > Struct Template Reference

#include <gridwise_fpAintB_gemm_wmma.hpp>

Classes

struct  SharedMemTrait

Public Types

using ThisThreadBlock = ThisThreadBlock<BlockSize>
using GridwiseGemmPipe
using CGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
using DefaultBlock2CTileMap

Static Public Member Functions

__host__ static __device__ constexpr auto MakeABlockDescriptor ()
__host__ static __device__ constexpr auto MakeBBlockDescriptor ()
__host__ static __device__ constexpr auto MakeABlockSliceCopyStep ()
__host__ static __device__ constexpr auto MakeBBlockSliceCopyStep ()
template<typename ABlockDesc_>
__host__ static __device__ constexpr auto MakeAWaveDescriptor (const ABlockDesc_ &)
template<typename BBlockDesc_>
__host__ static __device__ constexpr auto MakeBWaveDescriptor (const BBlockDesc_ &)
__host__ static __device__ constexpr auto GetCShuffleBlockDescriptor_MShRepeat_MPerShRepeat_NShRepeat_NPerShRepeat ()
template<typename Block2CTileMap>
__host__ static __device__ constexpr bool CheckValidity (const AGridDesc &a_grid_desc, const BGridDesc &b_grid_desc, const CGridDesc_M_N &c_grid_desc_m_n, const Block2CTileMap &block_2_ctile_map)
__host__ static __device__ constexpr bool CalculateHasMainKBlockLoop (index_t K)
__host__ static __device__ constexpr auto MakeCGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock (const CGridDesc_M_N &c_grid_desc_m_n)
__host__ static __device__ constexpr auto MakeDefaultBlock2CTileMap (const CGridDesc_M_N &c_grid_desc_m_n, index_t, index_t)
template<bool HasMainKBlockLoop, typename Block2CTileMap = DefaultBlock2CTileMap>
static __device__ void Run (const ADataType *__restrict__ p_a_grid, const BDataType *__restrict__ p_b_grid, const ScaleDataType *__restrict__ p_scale_grid, CDataType *__restrict__ p_c_grid, void *__restrict__ p_shared, const AGridDesc &a_grid_desc, const BGridDesc &b_grid_desc, const ScaleGridDesc &scale_grid_desc, const CGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock &c_grid_desc_mblock_mperblock_nblock_nperblock, const AElementwiseOperation &a_element_op, const BElementwiseOperation &b_element_op, const CElementwiseOperation &c_element_op, const Block2CTileMap &block_2_ctile_map)

Static Public Attributes

static constexpr auto I0 = Number<0>{}
static constexpr auto I1 = Number<1>{}
static constexpr auto I2 = Number<2>{}
static constexpr auto I3 = Number<3>{}
static constexpr auto I4 = Number<4>{}
static constexpr auto I5 = Number<5>{}
static constexpr auto I6 = Number<6>{}
static constexpr auto I7 = Number<7>{}
static constexpr auto K1 = Number<K1Value>{}
static constexpr auto MWaves = MPerBlock / (MRepeat * MPerWmma)
static constexpr auto NWaves = NPerBlock / (NRepeat * NPerWmma)
static constexpr auto WmmaK = K1 == 16 ? 32 : 16

Member Typedef Documentation

◆ CGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
using ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::CGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
Initial value:
remove_cvref_t<decltype(MakeCGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
CGridDesc_M_N{}))>
ck::remove_cvref_t
remove_cv_t< remove_reference_t< T > > remove_cvref_t
Definition type.hpp:297
ck::GridwiseFpAintBGemm_Wmma::MakeCGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
__host__ static __device__ constexpr auto MakeCGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(const CGridDesc_M_N &c_grid_desc_m_n)
Definition gridwise_fpAintB_gemm_wmma.hpp:529

◆ DefaultBlock2CTileMap

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
using ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::DefaultBlock2CTileMap
Initial value:
remove_cvref_t<decltype(MakeDefaultBlock2CTileMap(CGridDesc_M_N{}, 1, 1))>
ck::GridwiseFpAintBGemm_Wmma::MakeDefaultBlock2CTileMap
__host__ static __device__ constexpr auto MakeDefaultBlock2CTileMap(const CGridDesc_M_N &c_grid_desc_m_n, index_t, index_t)
Definition gridwise_fpAintB_gemm_wmma.hpp:548

◆ GridwiseGemmPipe

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
using ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::GridwiseGemmPipe
Initial value:
remove_cvref_t<decltype(GridwiseGemmPipeline_Selector<PipelineVer,
NumGemmKPrefetchStage,
LoopSched,
AEnableLds,
BEnableLds>())>
ck::GridwiseGemmPipeline_Selector
constexpr auto GridwiseGemmPipeline_Selector()
Definition gridwise_gemm_pipeline_selector.hpp:31

◆ ThisThreadBlock

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
using ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::ThisThreadBlock = ThisThreadBlock<BlockSize>

Member Function Documentation

◆ CalculateHasMainKBlockLoop()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
__host__ static __device__ constexpr bool ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::CalculateHasMainKBlockLoop ( index_t K)
inlinestaticconstexpr

◆ CheckValidity()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
template<typename Block2CTileMap>
__host__ static __device__ constexpr bool ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::CheckValidity ( const AGridDesc & a_grid_desc,
const BGridDesc & b_grid_desc,
const CGridDesc_M_N & c_grid_desc_m_n,
const Block2CTileMap & block_2_ctile_map )
inlinestaticconstexpr

◆ GetCShuffleBlockDescriptor_MShRepeat_MPerShRepeat_NShRepeat_NPerShRepeat()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
__host__ static __device__ constexpr auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::GetCShuffleBlockDescriptor_MShRepeat_MPerShRepeat_NShRepeat_NPerShRepeat ( )
inlinestaticconstexpr

◆ MakeABlockDescriptor()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
__host__ static __device__ constexpr auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::MakeABlockDescriptor ( )
inlinestaticconstexpr

◆ MakeABlockSliceCopyStep()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
__host__ static __device__ constexpr auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::MakeABlockSliceCopyStep ( )
inlinestaticconstexpr

◆ MakeAWaveDescriptor()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
template<typename ABlockDesc_>
__host__ static __device__ constexpr auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::MakeAWaveDescriptor ( const ABlockDesc_ & )
inlinestaticconstexpr

◆ MakeBBlockDescriptor()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
__host__ static __device__ constexpr auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::MakeBBlockDescriptor ( )
inlinestaticconstexpr

◆ MakeBBlockSliceCopyStep()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
__host__ static __device__ constexpr auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::MakeBBlockSliceCopyStep ( )
inlinestaticconstexpr

◆ MakeBWaveDescriptor()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
template<typename BBlockDesc_>
__host__ static __device__ constexpr auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::MakeBWaveDescriptor ( const BBlockDesc_ & )
inlinestaticconstexpr

◆ MakeCGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
__host__ static __device__ constexpr auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::MakeCGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock ( const CGridDesc_M_N & c_grid_desc_m_n)
inlinestaticconstexpr

◆ MakeDefaultBlock2CTileMap()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
__host__ static __device__ constexpr auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::MakeDefaultBlock2CTileMap ( const CGridDesc_M_N & c_grid_desc_m_n,
index_t ,
index_t  )
inlinestaticconstexpr

◆ Run()

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
template<bool HasMainKBlockLoop, typename Block2CTileMap = DefaultBlock2CTileMap>
__device__ void ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::Run ( const ADataType *__restrict__ p_a_grid,
const BDataType *__restrict__ p_b_grid,
const ScaleDataType *__restrict__ p_scale_grid,
CDataType *__restrict__ p_c_grid,
void *__restrict__ p_shared,
const AGridDesc & a_grid_desc,
const BGridDesc & b_grid_desc,
const ScaleGridDesc & scale_grid_desc,
const CGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock & c_grid_desc_mblock_mperblock_nblock_nperblock,
const AElementwiseOperation & a_element_op,
const BElementwiseOperation & b_element_op,
const CElementwiseOperation & c_element_op,
const Block2CTileMap & block_2_ctile_map )
inlinestatic

Member Data Documentation

◆ I0

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::I0 = Number<0>{}
staticconstexpr

◆ I1

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::I1 = Number<1>{}
staticconstexpr

◆ I2

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::I2 = Number<2>{}
staticconstexpr

◆ I3

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::I3 = Number<3>{}
staticconstexpr

◆ I4

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::I4 = Number<4>{}
staticconstexpr

◆ I5

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::I5 = Number<5>{}
staticconstexpr

◆ I6

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::I6 = Number<6>{}
staticconstexpr

◆ I7

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::I7 = Number<7>{}
staticconstexpr

◆ K1

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::K1 = Number<K1Value>{}
staticconstexpr

◆ MWaves

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::MWaves = MPerBlock / (MRepeat * MPerWmma)
staticconstexpr

◆ NWaves

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::NWaves = NPerBlock / (NRepeat * NPerWmma)
staticconstexpr

◆ WmmaK

template<index_t BlockSize, typename ADataType, typename BDataType, typename ScaleDataType, typename AccDataType, typename CShuffleDataType, typename CDataType, InMemoryDataOperationEnum CGlobalMemoryDataOperation, typename AGridDesc, typename BGridDesc, typename ScaleGridDesc, typename CGridDesc_M_N, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t MPerWmma, index_t NPerWmma, index_t K1Value, index_t MRepeat, index_t NRepeat, typename ABlockTransferThreadClusterLengths_K0_M_K1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_K1, bool AThreadTransferSrcResetCoordinateAfterRun, bool AEnableLds, bool ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_K0_N_K1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_K1, bool BThreadTransferSrcResetCoordinateAfterRun, bool BEnableLds, bool BBlockLdsExtraN, index_t CShuffleMRepeatPerShuffle, index_t CShuffleNRepeatPerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, index_t NumGemmKPrefetchStage = 1, LoopScheduler LoopSched = make_default_loop_scheduler(), PipelineVersion PipelineVer = PipelineVersion::weight_only>
auto ck::GridwiseFpAintBGemm_Wmma< BlockSize, ADataType, BDataType, ScaleDataType, AccDataType, CShuffleDataType, CDataType, CGlobalMemoryDataOperation, AGridDesc, BGridDesc, ScaleGridDesc, CGridDesc_M_N, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, MPerBlock, NPerBlock, KPerBlock, MPerWmma, NPerWmma, K1Value, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, AThreadTransferSrcResetCoordinateAfterRun, AEnableLds, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_K0_N_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_K1, BThreadTransferSrcResetCoordinateAfterRun, BEnableLds, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, NumGemmKPrefetchStage, LoopSched, PipelineVer >::WmmaK = K1 == 16 ? 32 : 16
staticconstexpr
The documentation for this struct was generated from the following file: