Composable Kernel: ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB > Struct Template Reference

GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB > Struct Template Reference

GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB > Struct Template Reference#

Composable Kernel: ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB > Struct Template Reference

Classes | Public Types | Static Public Member Functions | Static Public Attributes | List of all members

ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB > Struct Template Reference

#include <gridwise_gemm_xdl_cshuffle_v1.hpp>

Classes
struct	Problem
struct	Argument

Public Types
using	ThisThreadBlock = ThisThreadBlock<BlockSize>
using	GridwiseGemmPipe
using	Block2CTileMap = BlockToCTileMap_M00_N0_M01Adapt<MPerBlock, NPerBlock>

Static Public Member Functions
static __host__ auto	CalculateGridSize (index_t M, index_t N)
static __host__ auto	CalculateMPadded (index_t M)
static __host__ auto	CalculateNPadded (index_t N)
static __host__ auto	CalculateKPadded (index_t K)
static __host__ auto	CalculateAK0 (index_t K)
static __host__ auto	CalculateBK0 (index_t K)
static __host__ auto	CalculateMBlock (index_t M)
static __host__ auto	CalculateNBlock (index_t N)
static __device__ auto	MakeAGridDescriptor_AK0_M_AK1 (index_t M, index_t MPad, index_t K, index_t KPad, index_t StrideA, index_t AK0)
static __device__ auto	MakeBGridDescriptor_BK0_N_BK1 (index_t K, index_t KPad, index_t N, index_t NPad, index_t StrideB, index_t BK0)
__host__ static __device__ auto	MakeCGridDescriptor_M_N (index_t M, index_t MPad, index_t N, index_t NPad, index_t StrideC)
static __device__ constexpr auto	GetABlockDescriptor_AK0PerBlock_MPerBlock_AK1 ()
static __device__ constexpr auto	GetBBlockDescriptor_BK0PerBlock_NPerBlock_BK1 ()
static __device__ constexpr auto	GetCShuffleBlockDescriptor_MBlock_MPerBlock_NBlock_NPerBlock ()
static __device__ constexpr index_t	GetSharedMemoryNumberOfByte ()
template<InMemoryDataOperationEnum CGlobalMemoryDataOperation_ = InMemoryDataOperationEnum::Set>
static __device__ bool constexpr	IsValidCompilationParameter ()
static __host__ constexpr bool	CheckValidity (const Problem &problem)
static __host__ constexpr bool	CalculateHasMainKBlockLoop (index_t K)
template<typename CGridDesc>
static __device__ constexpr auto	MakeCGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock (const CGridDesc &c_grid_desc_m_n, index_t MBlock, index_t NBlock)
template<bool HasMainKBlockLoop>
static __device__ void	Run (const FloatA __restrict__ p_a_grid, const FloatB __restrict__ p_b_grid, FloatC __restrict__ p_c_grid, void __restrict__ p_shared, const Problem &problem)

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	AK0Number = Number<KPerBlock / AK1Value>{}
static constexpr auto	BK0Number = Number<KPerBlock / BK1Value>{}
static constexpr auto	AK1Number = Number<AK1Value>{}
static constexpr auto	BK1Number = Number<BK1Value>{}

Member Typedef Documentation

◆ Block2CTileMap

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

using ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::Block2CTileMap = BlockToCTileMap_M00_N0_M01Adapt<MPerBlock, NPerBlock>

◆ GridwiseGemmPipe

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

using ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::GridwiseGemmPipe

Initial value:

remove_cvref_t<

decltype(GridwiseGemmPipeline_Selector<PipelineVer, NumGemmKPrefetchStage, LoopSched>())>

ck::GridwiseGemmPipeline_Selector

constexpr auto GridwiseGemmPipeline_Selector()

Definition gridwise_gemm_pipeline_selector.hpp:31

ck::remove_cvref_t

remove_cv_t< remove_reference_t< T > > remove_cvref_t

Definition type.hpp:297

◆ ThisThreadBlock

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

using ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::ThisThreadBlock = ThisThreadBlock<BlockSize>

Member Function Documentation

◆ CalculateAK0()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CalculateAK0 ( index_t K )

inlinestatic

◆ CalculateBK0()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CalculateBK0 ( index_t K )

inlinestatic

◆ CalculateGridSize()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CalculateGridSize	(	index_t	M,
		index_t	N )

inlinestatic

◆ CalculateHasMainKBlockLoop()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ constexpr bool ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CalculateHasMainKBlockLoop ( index_t K )

inlinestaticconstexpr

◆ CalculateKPadded()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CalculateKPadded ( index_t K )

inlinestatic

◆ CalculateMBlock()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CalculateMBlock ( index_t M )

inlinestatic

◆ CalculateMPadded()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CalculateMPadded ( index_t M )

inlinestatic

◆ CalculateNBlock()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CalculateNBlock ( index_t N )

inlinestatic

◆ CalculateNPadded()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CalculateNPadded ( index_t N )

inlinestatic

◆ CheckValidity()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ constexpr bool ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::CheckValidity ( const Problem & problem )

inlinestaticconstexpr

◆ GetABlockDescriptor_AK0PerBlock_MPerBlock_AK1()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__device__ constexpr auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::GetABlockDescriptor_AK0PerBlock_MPerBlock_AK1 ( )

inlinestaticconstexpr

◆ GetBBlockDescriptor_BK0PerBlock_NPerBlock_BK1()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__device__ constexpr auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::GetBBlockDescriptor_BK0PerBlock_NPerBlock_BK1 ( )

inlinestaticconstexpr

◆ GetCShuffleBlockDescriptor_MBlock_MPerBlock_NBlock_NPerBlock()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__device__ constexpr auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::GetCShuffleBlockDescriptor_MBlock_MPerBlock_NBlock_NPerBlock ( )

inlinestaticconstexpr

◆ GetSharedMemoryNumberOfByte()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__device__ constexpr index_t ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::GetSharedMemoryNumberOfByte ( )

inlinestaticconstexpr

◆ IsValidCompilationParameter()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

template<InMemoryDataOperationEnum CGlobalMemoryDataOperation_ = InMemoryDataOperationEnum::Set>

__device__ bool constexpr ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::IsValidCompilationParameter ( )

inlinestaticconstexpr

◆ MakeAGridDescriptor_AK0_M_AK1()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__device__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::MakeAGridDescriptor_AK0_M_AK1	(	index_t	M,
		index_t	MPad,
		index_t	K,
		index_t	KPad,
		index_t	StrideA,
		index_t	AK0 )

inlinestatic

◆ MakeBGridDescriptor_BK0_N_BK1()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__device__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::MakeBGridDescriptor_BK0_N_BK1	(	index_t	K,
		index_t	KPad,
		index_t	N,
		index_t	NPad,
		index_t	StrideB,
		index_t	BK0 )

inlinestatic

◆ MakeCGridDescriptor_M_N()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

__host__ static __device__ auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::MakeCGridDescriptor_M_N	(	index_t	M,
		index_t	MPad,
		index_t	N,
		index_t	NPad,
		index_t	StrideC )

inlinestatic

◆ MakeCGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

template<typename CGridDesc>

__device__ constexpr auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::MakeCGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock	(	const CGridDesc &	c_grid_desc_m_n,
		index_t	MBlock,
		index_t	NBlock )

inlinestaticconstexpr

◆ Run()

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

template<bool HasMainKBlockLoop>

__device__ void ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::Run	(	const FloatA *__restrict__	p_a_grid,
		const FloatB *__restrict__	p_b_grid,
		FloatC *__restrict__	p_c_grid,
		void *__restrict__	p_shared,
		const Problem &	problem )

inlinestatic

Member Data Documentation

◆ AK0Number

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::AK0Number = Number<KPerBlock / AK1Value>{}

staticconstexpr

◆ AK1Number

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::AK1Number = Number<AK1Value>{}

staticconstexpr

◆ BK0Number

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::BK0Number = Number<KPerBlock / BK1Value>{}

staticconstexpr

◆ BK1Number

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::BK1Number = Number<BK1Value>{}

staticconstexpr

◆ I0

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::I0 = Number<0>{}

staticconstexpr

◆ I1

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::I1 = Number<1>{}

staticconstexpr

◆ I2

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::I2 = Number<2>{}

staticconstexpr

◆ I3

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::I3 = Number<3>{}

staticconstexpr

◆ I4

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::I4 = Number<4>{}

staticconstexpr

◆ I5

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::I5 = Number<5>{}

staticconstexpr

◆ I6

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::I6 = Number<6>{}

staticconstexpr

◆ I7

template<typename ALayout, typename BLayout, typename CLayout, typename FloatA, typename FloatB, typename FloatGemmAcc, typename FloatCShuffle, typename FloatC, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, tensor_operation::device::GemmSpecialization GemmSpec, InMemoryDataOperationEnum CGlobalMemoryDataOperation, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1Value, index_t BK1Value, index_t MPerXdl, index_t NPerXdl, index_t MXdlPerWave, index_t NXdlPerWave, typename ABlockTransferThreadClusterLengths_AK0_M_AK1, typename ABlockTransferThreadClusterArrangeOrder, typename ABlockTransferSrcAccessOrder, index_t ABlockTransferSrcVectorDim, index_t ABlockTransferSrcScalarPerVector, index_t ABlockTransferDstScalarPerVector_AK1, bool AThreadTransferSrcResetCoordinateAfterRun, index_t ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BThreadTransferSrcResetCoordinateAfterRun, index_t BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, LoopScheduler LoopSched, PipelineVersion PipelineVer = PipelineVersion::v1, typename ComputeTypeA = FloatC, typename ComputeTypeB = ComputeTypeA>

auto ck::GridwiseGemm_k0mk1_k0nk1_mn_xdl_cshuffle_v1< ALayout, BLayout, CLayout, FloatA, FloatB, FloatGemmAcc, FloatCShuffle, FloatC, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, CGlobalMemoryDataOperation, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1Value, BK1Value, MPerXdl, NPerXdl, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, AThreadTransferSrcResetCoordinateAfterRun, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BThreadTransferSrcResetCoordinateAfterRun, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB >::I7 = Number<7>{}

staticconstexpr

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

gridwise_gemm_xdl_cshuffle_v1.hpp