Class Members - Typedefs

 

- o -

• OaccDataType : ck_tile::BlockFmhaFwdSplitKVCombinePipeline< Problem_, Policy_ > , ck_tile::BlockFmhaFwdSplitKVPipelineNWarpSShuffleQRKSVS< Problem_, Policy_ > , ck_tile::BlockFmhaFwdSplitKVPipelineProblem< QDataType_, KDataType_, VDataType_, SaccDataType_, SMPLComputeDataType_, BiasDataType_, LSEDataType_, PDataType_, OaccDataType_, ODataType_, BlockFmhaShape_, kIsGroupMode_, FmhaMask_, Traits_ > , ck_tile::BlockFmhaFwdSplitKVPipelineQRKSVS< Problem_, Policy_ > , ck_tile::BlockFmhaPipelineProblem< QDataType_, KDataType_, VDataType_, SaccDataType_, SMPLComputeDataType_, BiasDataType_, RandValOutputDataType_, LSEDataType_, PDataType_, OaccDataType_, ODataType_, BlockFmhaShape_, kIsGroupMode_, FmhaMask_, Traits_ > , ck_tile::BlockFmhaPipelineQRKSVS< Problem_, Policy_ > , ck_tile::BlockFmhaPipelineQRKSVSAsync< Problem_, Policy_ > , ck_tile::BlockFmhaPipelineQRKSVSFp8< Problem_, Policy_ > , ck_tile::BlockFmhaPipelineQSKSVS< Problem_, Policy_ > , ck_tile::BlockFmhaSplitKVCombinePipelineProblem< LSEDataType_, OaccDataType_, ODataType_, HeadDimV_, kIsGroupMode_, kN1_, Traits_ > , ck_tile::FmhaFwdSplitKVCombineKernel< FmhaPipeline_, EpiloguePipeline_ > , ck_tile::FmhaFwdSplitKVKernel< FmhaPipeline_, EpiloguePipeline_ >
• OAccType : ck_tile::naive_attention_fwd_kernel< QType, KType, VType, OType, AccType, KVScaleType, QLayout, KLayout, VLayout, OLayout, KScaleLayout, VScaleLayout, Traits >
• ODataType : ck_tile::BlockFmhaBwdDQDKDVPipelineKRKTRVR< Problem, Policy > , ck_tile::BlockFmhaBwdDQDKDVPipelineKRKTRVRIGLP< Problem, Policy > , ck_tile::BlockFmhaBwdOGradDotO< Problem, Policy > , ck_tile::BlockFmhaBwdOGradDotOPipelineProblem< ODataType_, OGradDataType_, DDataType_, kBlockSize_, kVHeaddim_, kIsGroupMode_, Traits_ > , ck_tile::BlockFmhaBwdPipelineProblem< QDataType_, KDataType_, VDataType_, GemmDataType_, LSEDataType_, AccDataType_, DDataType_, BiasDataType_, RandValOutputDataType_, ODataType_, OGradDataType_, QGradDataType_, KGradDataType_, VGradDataType_, BiasGradDataType_, BlockFmhaShape_, kIsGroupMode_, kIsDeterministic_, FmhaMask_, FmhaDropout_, Traits_ > , ck_tile::BlockFmhaFwdSplitKVCombinePipeline< Problem_, Policy_ > , ck_tile::BlockFmhaFwdSplitKVPipelineNWarpSShuffleQRKSVS< Problem_, Policy_ > , ck_tile::BlockFmhaFwdSplitKVPipelineProblem< QDataType_, KDataType_, VDataType_, SaccDataType_, SMPLComputeDataType_, BiasDataType_, LSEDataType_, PDataType_, OaccDataType_, ODataType_, BlockFmhaShape_, kIsGroupMode_, FmhaMask_, Traits_ > , ck_tile::BlockFmhaFwdSplitKVPipelineQRKSVS< Problem_, Policy_ > , ck_tile::BlockFmhaPipelineProblem< QDataType_, KDataType_, VDataType_, SaccDataType_, SMPLComputeDataType_, BiasDataType_, RandValOutputDataType_, LSEDataType_, PDataType_, OaccDataType_, ODataType_, BlockFmhaShape_, kIsGroupMode_, FmhaMask_, Traits_ > , ck_tile::BlockFmhaPipelineQRKSVS< Problem_, Policy_ > , ck_tile::BlockFmhaPipelineQRKSVSAsync< Problem_, Policy_ > , ck_tile::BlockFmhaPipelineQRKSVSFp8< Problem_, Policy_ > , ck_tile::BlockFmhaPipelineQSKSVS< Problem_, Policy_ > , ck_tile::BlockFmhaSplitKVCombinePipelineProblem< LSEDataType_, OaccDataType_, ODataType_, HeadDimV_, kIsGroupMode_, kN1_, Traits_ > , ck_tile::CShuffleEpilogue< Problem_, Policy_ > , ck_tile::CShuffleEpilogueProblem< AccDataType_, ODataType_, CLayout_, kBlockSize_, kM_, kN_, kMWave_, kNWave_, kMPerXdl_, kNPerXdl_, kKPerXdl_, isCTransposed_ > , ck_tile::Default2DEpilogue< Problem_, Policy_ > , ck_tile::Default2DEpilogueProblem< AccDataType_, ODataType_, kPadM_, kPadN_, UseRawStore_ > , ck_tile::DefaultGemm2DEpilogue< Problem_, Policy_ > , ck_tile::DynamicQuantEpilogue< Problem_, Policy_ > , ck_tile::DynamicQuantEpilogueProblem< AccDataType_, SmoothScaleDataType_, YScaleDataType_, ODataType_, BlockShape_, Traits_ > , ck_tile::FlatmmSn_32x128x512_1x4x1_16x16x32_BF16 , ck_tile::FlatmmSn_32x128x512_1x4x1_16x16x32_BF16_itl , ck_tile::FlatmmSn_32x128x512_1x4x1_16x16x32_FP16 , ck_tile::FlatmmSn_32x128x512_1x4x1_16x16x32_FP16_itl , ck_tile::FmhaBwdOGradDotOKernel< FmhaBwdOGradDotO_ > , ck_tile::FmhaFwdKernel< FmhaPipeline_, EpiloguePipeline_ > , ck_tile::FmhaFwdSplitKVCombineKernel< FmhaPipeline_, EpiloguePipeline_ > , ck_tile::FmhaFwdSplitKVKernel< FmhaPipeline_, EpiloguePipeline_ > , ck_tile::FusedMoeGemmKernel< Partitioner_, Pipeline_, Epilogue_ > , ck_tile::FusedMoeGemmPipeline_FlatmmEx< Problem_, Policy_ > , ck_tile::FusedMoeGemmPipeline_FlatmmUk< Problem_, Policy_ > , ck_tile::FusedMoeGemmPipelineProblem< ADataType_, GDataType_, DDataType_, AccDataType_, ODataType_, AScaleDataType_, GScaleDataType_, DScaleDataType_, YSmoothScaleDataType_, TopkWeightDataType_, IndexDataType_, GateActivation_, BlockShape_, Traits_ >
• OffsettedLocalBlock2ETileMap : ck::tensor_operation::device::DeviceGroupedGemmMultipleDXdlCShuffleTileLoop< ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, 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, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEShuffleBlockTransferScalarPerVectors, BlkGemmPipeSched, BlkGemmPipelineVer, ComputeTypeA, ComputeTypeB >
• OffsetTile1DPartitioner : ck_tile::GroupedGemmKernel< TilePartitioner_, GemmPipeline_, EpiloguePipeline_ >
• OGradDataType : ck_tile::BlockFmhaBwdDQDKDVPipelineKRKTRVR< Problem, Policy > , ck_tile::BlockFmhaBwdDQDKDVPipelineKRKTRVRIGLP< Problem, Policy > , ck_tile::BlockFmhaBwdOGradDotO< Problem, Policy > , ck_tile::BlockFmhaBwdOGradDotOPipelineProblem< ODataType_, OGradDataType_, DDataType_, kBlockSize_, kVHeaddim_, kIsGroupMode_, Traits_ > , ck_tile::BlockFmhaBwdPipelineProblem< QDataType_, KDataType_, VDataType_, GemmDataType_, LSEDataType_, AccDataType_, DDataType_, BiasDataType_, RandValOutputDataType_, ODataType_, OGradDataType_, QGradDataType_, KGradDataType_, VGradDataType_, BiasGradDataType_, BlockFmhaShape_, kIsGroupMode_, kIsDeterministic_, FmhaMask_, FmhaDropout_, Traits_ > , ck_tile::FmhaBwdDQDKDVKernel< FmhaPipeline_, KGradEpiloguePipeline_, VGradEpiloguePipeline_ > , ck_tile::FmhaBwdOGradDotOKernel< FmhaBwdOGradDotO_ >
• old_scan : ck::sequence_reverse_inclusive_scan< Sequence< I, Is... >, Reduce, Init > , ck_tile::impl::reverse_slice_sequence_impl< sequence< x, xs... >, sequence< m, ms... >, sequence< id, ids... >, SliceSize > , ck_tile::sequence_exclusive_scan< sequence< Xs... >, sequence< Y, Ys... >, Reduce > , ck_tile::sequence_reverse_inclusive_scan< sequence< I, Is... >, Reduce, Init >
• Op : ck::ThreadwiseReduction< AccDataType, SrcThreadDesc_M_K, DstThreadDesc_M, OpReduce, PropagateNan, Accumulation >
• opType : ck::reduce_binary_operator< ReduceTensorOp::ADD > , ck::reduce_binary_operator< ReduceTensorOp::AMAX > , ck::reduce_binary_operator< ReduceTensorOp::AVG > , ck::reduce_binary_operator< ReduceTensorOp::MAX > , ck::reduce_binary_operator< ReduceTensorOp::MIN > , ck::reduce_binary_operator< ReduceTensorOp::MUL > , ck::reduce_binary_operator< ReduceTensorOp::NORM1 > , ck::reduce_binary_operator< ReduceTensorOp::NORM2 >
• OutDataType : ck_tile::BlockImageToColumnProblem< InDataType_, OutDataType_, BlockShape_, NDimSpatial_, AligmentIn_, AligmentOut_ > , ck_tile::ImageToColumn< Problem_ >
• OutDataTypePointerTuple : ck::tensor_operation::device::DeviceElementwiseImpl< InDataTypeTuple, OutDataTypeTuple, ElementwiseOperation, NumDim, BlockSize, M0PerBlock, M1PerBlock, M0PerThread, M1PerThread, ThreadClusterArrangeOrder, InScalarPerVectorSeq, OutScalarPerVectorSeq > , ck::tensor_operation::device::DeviceMultipleReduceMultiBlock< NumReduction, InDataType, AccDataType, OutDataTypeTuple, Rank, NumReduceDim, ReduceOperation, InElementwiseOperationTuple, AccElementwiseOperationTuple, OutMemoryDataOperation, PropagateNan, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSizeSeq > , ck::tensor_operation::device::DeviceMultipleReduceThreadWise< NumReduction, InDataType, AccDataType, OutDataTypeTuple, Rank, NumReduceDim, ReduceOperation, InElementwiseOperationTuple, AccElementwiseOperationTuple, PropagateNan, BlockSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSizeSeq >
• OutElementwiseOperation : ck::tensor_operation::device::DeviceGemm_Xdl_CShuffleV3R1< ALayout, BLayout, DsLayout, CLayout, ADataType, BDataType, DsDataType, CDataType, GemmAccDataType, CShuffleDataType, AElementwiseOperation, BElementwiseOperation, CElementwiseOperation, GemmSpec, 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, BlkGemmPipeSched, BlkGemmPipelineVer, ReduceDataType, ComputeTypeA, ComputeTypeB >
• OutGrid1dDescTuple : ck::tensor_operation::device::DeviceElementwiseImpl< InDataTypeTuple, OutDataTypeTuple, ElementwiseOperation, NumDim, BlockSize, M0PerBlock, M1PerBlock, M0PerThread, M1PerThread, ThreadClusterArrangeOrder, InScalarPerVectorSeq, OutScalarPerVectorSeq >
• OutGridDesc : ck::tensor_operation::device::DevicePermuteImpl< NumDim, InDataType, OutDataType, ElementwiseOperation, BlockSize, NPerBlock, HPerBlock, WPerBlock, InBlockLdsExtraW, InBlockTransferThreadClusterLengths, InBlockTransferThreadClusterArrangeOrder, SrcVectorDim, DstVectorDim, SrcScalarPerVector, DstScalarPerVector >
• OutGridDesc_M : ck::tensor_operation::device::DeviceReduceThreadWiseMultiD< InDataType, DsDataType, AccDataType, OutDataType, Rank, NumReduceDim, ReduceOperation, InElementwiseOperation, OutElementwiseOperation, BlockSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, DsVectorSizeSequence >
• OutGridDesc_M_Tuple : ck::tensor_operation::device::DeviceMultipleReduceMultiBlock< NumReduction, InDataType, AccDataType, OutDataTypeTuple, Rank, NumReduceDim, ReduceOperation, InElementwiseOperationTuple, AccElementwiseOperationTuple, OutMemoryDataOperation, PropagateNan, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSizeSeq > , ck::tensor_operation::device::DeviceMultipleReduceThreadWise< NumReduction, InDataType, AccDataType, OutDataTypeTuple, Rank, NumReduceDim, ReduceOperation, InElementwiseOperationTuple, AccElementwiseOperationTuple, PropagateNan, BlockSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSizeSeq >
• OutGridDesc_M_Tuple_2 : ck::tensor_operation::device::DeviceMultipleReduceMultiBlock< NumReduction, InDataType, AccDataType, OutDataTypeTuple, Rank, NumReduceDim, ReduceOperation, InElementwiseOperationTuple, AccElementwiseOperationTuple, OutMemoryDataOperation, PropagateNan, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSizeSeq >
• OutGridDescTuple : ck::tensor_operation::device::DeviceElementwiseImpl< InDataTypeTuple, OutDataTypeTuple, ElementwiseOperation, NumDim, BlockSize, M0PerBlock, M1PerBlock, M0PerThread, M1PerThread, ThreadClusterArrangeOrder, InScalarPerVectorSeq, OutScalarPerVectorSeq >
• OutputArray : ck::tensor_operation::element_wise::FastNumericArrayConverter< uint8_t, half_t, 4 > , ck::tensor_operation::element_wise::FastNumericArrayConverter< uint8_t, half_t, N >
• OutputGridDesc : ck::tensor_operation::device::DeviceColumnToImageImpl< NDimSpatial, ImageLayout, InputDataType, OutputDataType, BlockSize, MPerBlock, KPerBlock, ThreadClusterLengths, ScalarPerVector,, bool, type > , ck::tensor_operation::device::DeviceImageToColumnImpl< NDimSpatial, ImageLayout, InputDataType, OutputDataType, BlockSize, MPerBlock, KPerBlock, ThreadClusterLengths, ScalarPerVector,, bool, type >