DeviceCGemm< AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > Struct Template Reference

DeviceCGemm&lt; AElementwiseOperation, BElementwiseOperation, CElementwiseOperation &gt; Struct Template Reference#

Composable Kernel: ck::tensor_operation::device::DeviceCGemm< AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > Struct Template Reference
Public Member Functions | List of all members
ck::tensor_operation::device::DeviceCGemm< AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > Struct Template Referenceabstract

#include <device_cgemm.hpp>

Inheritance diagram for ck::tensor_operation::device::DeviceCGemm< AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >:
ck::tensor_operation::device::BaseOperator ck::tensor_operation::device::DeviceCGemm_4Gemm_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, GemmAccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, 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, LoopSched, >

Public Member Functions

virtual std::unique_ptr< BaseArgument > MakeArgumentPointer (const void *p_a_real, const void *p_a_imag, const void *p_b_real, const void *p_b_imag, void *p_c_real, void *p_c_imag, void *p_workspace, ck::index_t M, ck::index_t N, ck::index_t K, ck::index_t StrideA, ck::index_t StrideB, ck::index_t StrideC, AElementwiseOperation a_element_op, BElementwiseOperation b_element_op, CElementwiseOperation c_element_op, ck::index_t KBatch=1)=0
virtual std::unique_ptr< BaseInvoker > MakeInvokerPointer ()=0
virtual std::size_t GetWorkspaceSize (index_t MRaw, index_t NRaw, index_t KRaw, index_t StrideA, index_t StrideB, index_t StrideC) const =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

◆ GetWorkspaceSize()

template<typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual std::size_t ck::tensor_operation::device::DeviceCGemm< AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::GetWorkspaceSize ( index_t MRaw,
index_t NRaw,
index_t KRaw,
index_t StrideA,
index_t StrideB,
index_t StrideC ) const
pure virtual

Implemented in ck::tensor_operation::device::DeviceCGemm_4Gemm_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, GemmAccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, 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, LoopSched, >.

◆ MakeArgumentPointer()

template<typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual std::unique_ptr< BaseArgument > ck::tensor_operation::device::DeviceCGemm< AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::MakeArgumentPointer ( const void * p_a_real,
const void * p_a_imag,
const void * p_b_real,
const void * p_b_imag,
void * p_c_real,
void * p_c_imag,
void * p_workspace,
ck::index_t M,
ck::index_t N,
ck::index_t K,
ck::index_t StrideA,
ck::index_t StrideB,
ck::index_t StrideC,
AElementwiseOperation a_element_op,
BElementwiseOperation b_element_op,
CElementwiseOperation c_element_op,
ck::index_t KBatch = 1 )
pure virtual

Implemented in ck::tensor_operation::device::DeviceCGemm_4Gemm_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, GemmAccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, 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, LoopSched, >.

◆ MakeInvokerPointer()

template<typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual std::unique_ptr< BaseInvoker > ck::tensor_operation::device::DeviceCGemm< AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::MakeInvokerPointer ( )
pure virtual

Implemented in ck::tensor_operation::device::DeviceCGemm_4Gemm_Xdl_CShuffle< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, GemmAccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, 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, LoopSched, >.

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