DeviceBatchedGemmV2BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleType, CDataType, ScaleBlockN, ScaleBlockK, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > Struct Template Reference

DeviceBatchedGemmV2BScale&lt; ALayout, BLayout, CLayout, ADataType, BDataType, BScaleType, CDataType, ScaleBlockN, ScaleBlockK, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation &gt; Struct Template Reference#

Composable Kernel: ck::tensor_operation::device::DeviceBatchedGemmV2BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleType, CDataType, ScaleBlockN, ScaleBlockK, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > Struct Template Reference
Public Member Functions | List of all members
ck::tensor_operation::device::DeviceBatchedGemmV2BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleType, CDataType, ScaleBlockN, ScaleBlockK, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > Struct Template Referenceabstract

#include <device_batched_gemm.hpp>

Inheritance diagram for ck::tensor_operation::device::DeviceBatchedGemmV2BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleType, CDataType, ScaleBlockN, ScaleBlockK, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >:
ck::tensor_operation::device::BaseOperator

Public Member Functions

virtual std::unique_ptr< BaseArgument > MakeArgumentPointer (const void *p_a, const void *p_b, void *p_c, ck::index_t M, ck::index_t N, ck::index_t K, ck::index_t StrideA, ck::index_t StrideB, ck::index_t StrideC, ck::index_t StrideScaleB, ck::index_t BatchStrideA, ck::index_t BatchStrideB, ck::index_t BatchStrideC, ck::index_t BatchStrideScaleB, const void *p_b_scale, ck::index_t Batch, ck::index_t KBatch, AElementwiseOperation a_element_op, BElementwiseOperation b_element_op, CElementwiseOperation c_element_op)=0
virtual std::unique_ptr< BaseInvoker > MakeInvokerPointer ()=0
virtual bool GetPermuteB ()=0
virtual ck::index_t GetKPerBlock ()=0
Public Member Functions inherited from ck::tensor_operation::device::BaseOperator
 BaseOperator ()=default
 BaseOperator (const BaseOperator &)=default
BaseOperator & operator= (const BaseOperator &)=default
virtual bool IsSupportedArgument (const BaseArgument *)
virtual std::string GetTypeString () const
virtual std::string GetInstanceString () const
virtual std::string GetTypeIdName () const
virtual std::optional< std::string > GetObjectName () const
virtual std::optional< std::string > GetTemplateInfo () const
virtual std::string GetTypeIdHashCode () const
virtual size_t GetWorkSpaceSize (const BaseArgument *) const
virtual void SetWorkSpacePointer (BaseArgument *p_arg, void *p_workspace, const StreamConfig &=StreamConfig{}) const
virtual ~BaseOperator ()

Member Function Documentation

◆ GetKPerBlock()

template<typename ALayout, typename BLayout, typename CLayout, typename ADataType, typename BDataType, typename BScaleType, typename CDataType, index_t ScaleBlockN, index_t ScaleBlockK, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual ck::index_t ck::tensor_operation::device::DeviceBatchedGemmV2BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleType, CDataType, ScaleBlockN, ScaleBlockK, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::GetKPerBlock ( )
pure virtual

Implemented in ck::tensor_operation::device::DeviceBatchedGemm_Wmma_CShuffleV3_BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleDataType, CDataType, AccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, ScaleBlockN, ScaleBlockK, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >, and ck::tensor_operation::device::DeviceBatchedGemm_Xdl_CShuffleV3_BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleDataType, CDataType, GemmAccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, ScaleBlockN, ScaleBlockK, 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, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >.

◆ GetPermuteB()

template<typename ALayout, typename BLayout, typename CLayout, typename ADataType, typename BDataType, typename BScaleType, typename CDataType, index_t ScaleBlockN, index_t ScaleBlockK, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual bool ck::tensor_operation::device::DeviceBatchedGemmV2BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleType, CDataType, ScaleBlockN, ScaleBlockK, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::GetPermuteB ( )
pure virtual

Implemented in ck::tensor_operation::device::DeviceBatchedGemm_Wmma_CShuffleV3_BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleDataType, CDataType, AccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, ScaleBlockN, ScaleBlockK, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >, and ck::tensor_operation::device::DeviceBatchedGemm_Xdl_CShuffleV3_BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleDataType, CDataType, GemmAccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, ScaleBlockN, ScaleBlockK, 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, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >.

◆ MakeArgumentPointer()

template<typename ALayout, typename BLayout, typename CLayout, typename ADataType, typename BDataType, typename BScaleType, typename CDataType, index_t ScaleBlockN, index_t ScaleBlockK, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual std::unique_ptr< BaseArgument > ck::tensor_operation::device::DeviceBatchedGemmV2BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleType, CDataType, ScaleBlockN, ScaleBlockK, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::MakeArgumentPointer ( const void * p_a,
const void * p_b,
void * p_c,
ck::index_t M,
ck::index_t N,
ck::index_t K,
ck::index_t StrideA,
ck::index_t StrideB,
ck::index_t StrideC,
ck::index_t StrideScaleB,
ck::index_t BatchStrideA,
ck::index_t BatchStrideB,
ck::index_t BatchStrideC,
ck::index_t BatchStrideScaleB,
const void * p_b_scale,
ck::index_t Batch,
ck::index_t KBatch,
AElementwiseOperation a_element_op,
BElementwiseOperation b_element_op,
CElementwiseOperation c_element_op )
pure virtual

Implemented in ck::tensor_operation::device::DeviceBatchedGemm_Wmma_CShuffleV3_BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleDataType, CDataType, AccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, ScaleBlockN, ScaleBlockK, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >, and ck::tensor_operation::device::DeviceBatchedGemm_Xdl_CShuffleV3_BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleDataType, CDataType, GemmAccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, ScaleBlockN, ScaleBlockK, 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, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >.

◆ MakeInvokerPointer()

template<typename ALayout, typename BLayout, typename CLayout, typename ADataType, typename BDataType, typename BScaleType, typename CDataType, index_t ScaleBlockN, index_t ScaleBlockK, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual std::unique_ptr< BaseInvoker > ck::tensor_operation::device::DeviceBatchedGemmV2BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleType, CDataType, ScaleBlockN, ScaleBlockK, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::MakeInvokerPointer ( )
pure virtual

Implemented in ck::tensor_operation::device::DeviceBatchedGemm_Wmma_CShuffleV3_BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleDataType, CDataType, AccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, ScaleBlockN, ScaleBlockK, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerWmma, NPerWmma, MRepeat, NRepeat, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >, and ck::tensor_operation::device::DeviceBatchedGemm_Xdl_CShuffleV3_BScale< ALayout, BLayout, CLayout, ADataType, BDataType, BScaleDataType, CDataType, GemmAccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, BlockSize, ScaleBlockN, ScaleBlockK, 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, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB, PermuteA, PermuteB >.

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