DeviceGroupedGemmMultiABDFixedNK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > Struct Template Reference

DeviceGroupedGemmMultiABDFixedNK&lt; AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation &gt; Struct Template Reference#

Composable Kernel: ck::tensor_operation::device::DeviceGroupedGemmMultiABDFixedNK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > Struct Template Reference
Public Member Functions | List of all members
ck::tensor_operation::device::DeviceGroupedGemmMultiABDFixedNK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > Struct Template Referenceabstract

#include <device_grouped_gemm_multi_abd_fixed_nk.hpp>

Inheritance diagram for ck::tensor_operation::device::DeviceGroupedGemmMultiABDFixedNK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >:
ck::tensor_operation::device::DeviceGroupedGemmMultiABD< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation > ck::tensor_operation::device::BaseOperator

Public Member Functions

virtual void SetDeviceKernelArgs (BaseArgument *p_arg, const void *kernel_args) const =0
virtual size_t GetDeviceKernelArgSize (const BaseArgument *p_arg) const =0
virtual void SetKBatch (BaseArgument *p_arg, index_t k_batch) const =0
Public Member Functions inherited from ck::tensor_operation::device::DeviceGroupedGemmMultiABD< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >
virtual std::unique_ptr< BaseArgument > MakeArgumentPointer (std::vector< std::array< const void *, NumATensor > > &p_as, std::vector< std::array< const void *, NumBTensor > > &p_bs, std::vector< std::array< const void *, NumDTensor > > &p_ds, std::vector< void * > &p_e, std::vector< GemmMultiABDDesc > &gemm_desc, AElementwiseOperation a_element_op=AElementwiseOperation{}, BElementwiseOperation b_element_op=BElementwiseOperation{}, CElementwiseOperation c_element_op=CElementwiseOperation{})=0
virtual std::unique_ptr< BaseInvoker > MakeInvokerPointer ()=0
virtual void SetElementwiseOps (BaseArgument *p_arg, AElementwiseOperation a_element_op, BElementwiseOperation b_element_op, CElementwiseOperation cde_element_op) 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 ()

Additional Inherited Members

Static Public Attributes inherited from ck::tensor_operation::device::DeviceGroupedGemmMultiABD< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >
static constexpr index_t NumATensor
static constexpr index_t NumBTensor
static constexpr index_t NumDTensor

Member Function Documentation

◆ GetDeviceKernelArgSize()

template<typename AsLayout, typename BsLayout, typename DsLayout, typename ELayout, typename AsDataType, typename BsDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual size_t ck::tensor_operation::device::DeviceGroupedGemmMultiABDFixedNK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::GetDeviceKernelArgSize ( const BaseArgument * p_arg) const
pure virtual

Implemented in ck::tensor_operation::device::DeviceGroupedGemm_Xdl_Multi_ABD_Fixed_NK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, GemmSpec, NumPrefetch, 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, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEBlockTransferScalarPerVector_NPerBlock, ComputeType, LoopSched >.

◆ SetDeviceKernelArgs()

template<typename AsLayout, typename BsLayout, typename DsLayout, typename ELayout, typename AsDataType, typename BsDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual void ck::tensor_operation::device::DeviceGroupedGemmMultiABDFixedNK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::SetDeviceKernelArgs ( BaseArgument * p_arg,
const void * kernel_args ) const
pure virtual

Implemented in ck::tensor_operation::device::DeviceGroupedGemm_Xdl_Multi_ABD_Fixed_NK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, GemmSpec, NumPrefetch, 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, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEBlockTransferScalarPerVector_NPerBlock, ComputeType, LoopSched >.

◆ SetKBatch()

template<typename AsLayout, typename BsLayout, typename DsLayout, typename ELayout, typename AsDataType, typename BsDataType, typename DsDataType, typename EDataType, typename AElementwiseOperation, typename BElementwiseOperation, typename CElementwiseOperation>
virtual void ck::tensor_operation::device::DeviceGroupedGemmMultiABDFixedNK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation >::SetKBatch ( BaseArgument * p_arg,
index_t k_batch ) const
pure virtual

Implemented in ck::tensor_operation::device::DeviceGroupedGemm_Xdl_Multi_ABD_Fixed_NK< AsLayout, BsLayout, DsLayout, ELayout, AsDataType, BsDataType, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, GemmSpec, NumPrefetch, 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, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEBlockTransferScalarPerVector_NPerBlock, ComputeType, LoopSched >.

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