Composable Kernel: ck::tensor_operation::device::DeviceGemmBiasAddReduce_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, BiasDataType, D0DataType, GemmAccDataType, CShuffleDataType, ReduceAccDataType, ReducePtrsGlobal, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, D0ElementwiseOperation, ReduceOperations, ReduceInElementwiseOperations, ReduceAccElementwiseOperations, ReduceGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, CReduceThreadClusterLengths_MPerBlock_NPerBlock, CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopSched >::Invoker Struct Reference

Invoker Struct Reference

Invoker Struct Reference#

Composable Kernel: ck::tensor_operation::device::DeviceGemmBiasAddReduce_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, BiasDataType, D0DataType, GemmAccDataType, CShuffleDataType, ReduceAccDataType, ReducePtrsGlobal, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, D0ElementwiseOperation, ReduceOperations, ReduceInElementwiseOperations, ReduceAccElementwiseOperations, ReduceGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, CReduceThreadClusterLengths_MPerBlock_NPerBlock, CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopSched >::Invoker Struct Reference

Public Types | Public Member Functions | List of all members

ck::tensor_operation::device::DeviceGemmBiasAddReduce_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, BiasDataType, D0DataType, GemmAccDataType, CShuffleDataType, ReduceAccDataType, ReducePtrsGlobal, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, D0ElementwiseOperation, ReduceOperations, ReduceInElementwiseOperations, ReduceAccElementwiseOperations, ReduceGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, CReduceThreadClusterLengths_MPerBlock_NPerBlock, CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopSched >::Invoker Struct Reference

#include <device_gemm_bias_add_reduce_xdl_cshuffle.hpp>

Inheritance diagram for ck::tensor_operation::device::DeviceGemmBiasAddReduce_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, BiasDataType, D0DataType, GemmAccDataType, CShuffleDataType, ReduceAccDataType, ReducePtrsGlobal, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, D0ElementwiseOperation, ReduceOperations, ReduceInElementwiseOperations, ReduceAccElementwiseOperations, ReduceGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, CReduceThreadClusterLengths_MPerBlock_NPerBlock, CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopSched >::Invoker:

Public Types
using	Argument = DeviceOp::Argument

Public Member Functions
template<typename GridwiseGemm>
float	RunImp (const Argument &arg, const StreamConfig &stream_config=StreamConfig{})
INVOKER_RUN_IMPL float	Run (const BaseArgument *p_arg, const StreamConfig &stream_config=StreamConfig{}) override
Public Member Functions inherited from ck::tensor_operation::device::BaseInvoker
	BaseInvoker ()=default
	BaseInvoker (const BaseInvoker &)=default
BaseInvoker &	operator= (const BaseInvoker &)=default
virtual	~BaseInvoker ()

Member Typedef Documentation

◆ Argument

template<typename ALayout, typename BLayout, typename CLayout, typename ADataType, typename BDataType, typename CDataType, typename BiasDataType, typename D0DataType, typename GemmAccDataType, typename CShuffleDataType, typename ReduceAccDataType, typename ReducePtrsGlobal, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, typename D0ElementwiseOperation, typename ReduceOperations, typename ReduceInElementwiseOperations, typename ReduceAccElementwiseOperations, typename ReduceGlobalMemoryDataOperation, GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, 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 ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, typename CReduceThreadClusterLengths_MPerBlock_NPerBlock, index_t CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, index_t CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopScheduler LoopSched = make_default_loop_scheduler()>

using ck::tensor_operation::device::DeviceGemmBiasAddReduce_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, BiasDataType, D0DataType, GemmAccDataType, CShuffleDataType, ReduceAccDataType, ReducePtrsGlobal, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, D0ElementwiseOperation, ReduceOperations, ReduceInElementwiseOperations, ReduceAccElementwiseOperations, ReduceGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, CReduceThreadClusterLengths_MPerBlock_NPerBlock, CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopSched >::Invoker::Argument = DeviceOp::Argument

Member Function Documentation

◆ Run()

template<typename ALayout, typename BLayout, typename CLayout, typename ADataType, typename BDataType, typename CDataType, typename BiasDataType, typename D0DataType, typename GemmAccDataType, typename CShuffleDataType, typename ReduceAccDataType, typename ReducePtrsGlobal, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, typename D0ElementwiseOperation, typename ReduceOperations, typename ReduceInElementwiseOperations, typename ReduceAccElementwiseOperations, typename ReduceGlobalMemoryDataOperation, GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, 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 ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, typename CReduceThreadClusterLengths_MPerBlock_NPerBlock, index_t CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, index_t CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopScheduler LoopSched = make_default_loop_scheduler()>

INVOKER_RUN_IMPL float ck::tensor_operation::device::DeviceGemmBiasAddReduce_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, BiasDataType, D0DataType, GemmAccDataType, CShuffleDataType, ReduceAccDataType, ReducePtrsGlobal, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, D0ElementwiseOperation, ReduceOperations, ReduceInElementwiseOperations, ReduceAccElementwiseOperations, ReduceGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, CReduceThreadClusterLengths_MPerBlock_NPerBlock, CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopSched >::Invoker::Run	(	const BaseArgument *	p_arg,
		const StreamConfig &	stream_config = StreamConfig{} )

inlineoverridevirtual

Reimplemented from ck::tensor_operation::device::BaseInvoker.

◆ RunImp()

template<typename ALayout, typename BLayout, typename CLayout, typename ADataType, typename BDataType, typename CDataType, typename BiasDataType, typename D0DataType, typename GemmAccDataType, typename CShuffleDataType, typename ReduceAccDataType, typename ReducePtrsGlobal, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation, typename D0ElementwiseOperation, typename ReduceOperations, typename ReduceInElementwiseOperations, typename ReduceAccElementwiseOperations, typename ReduceGlobalMemoryDataOperation, GemmSpecialization GemmSpec, index_t NumGemmKPrefetchStage, index_t BlockSize, index_t MPerBlock, index_t NPerBlock, index_t KPerBlock, index_t AK1, index_t BK1, 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 ABlockLdsExtraM, typename BBlockTransferThreadClusterLengths_BK0_N_BK1, typename BBlockTransferThreadClusterArrangeOrder, typename BBlockTransferSrcAccessOrder, index_t BBlockTransferSrcVectorDim, index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferDstScalarPerVector_BK1, bool BBlockLdsExtraN, index_t CShuffleMXdlPerWavePerShuffle, index_t CShuffleNXdlPerWavePerShuffle, typename CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, index_t CShuffleBlockTransferScalarPerVector_NPerBlock, typename CReduceThreadClusterLengths_MPerBlock_NPerBlock, index_t CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, index_t CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopScheduler LoopSched = make_default_loop_scheduler()>

template<typename GridwiseGemm>

float ck::tensor_operation::device::DeviceGemmBiasAddReduce_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, BiasDataType, D0DataType, GemmAccDataType, CShuffleDataType, ReduceAccDataType, ReducePtrsGlobal, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, D0ElementwiseOperation, ReduceOperations, ReduceInElementwiseOperations, ReduceAccElementwiseOperations, ReduceGlobalMemoryDataOperation, GemmSpec, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, CReduceThreadClusterLengths_MPerBlock_NPerBlock, CReduceThreadLds2VGprCopySrcDstScalarPerVector_NPerBlock, CReduceThreadVgpr2GlobalCopySrcDstScalarPerVector_MPerBlock, LoopSched >::Invoker::RunImp	(	const Argument &	arg,
		const StreamConfig &	stream_config = StreamConfig{} )

inline

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

device_gemm_bias_add_reduce_xdl_cshuffle.hpp