Skip to main content

Ctrl+K

ROCm™ Software Debian Build 7.2.4

Version List

GitHub
Community
Blogs
ROCm Developer Hub
ROCm Toolkits
Systems and Infra Docs
Infinity Hub
Support

ROCm documentation

Composable Kernel 1.2.0 Documentation

Install

Composable Kernel prerequisites
Build and install Composable Kernel
Composable Kernel Docker images

Conceptual

Composable Kernel structure
Composable Kernel mathematical basis

Tutorial

Composable Kernel examples

Reference

Composable Kernel scalar types
Composable Kernel custom types
Composable Kernel vector utilities
Composable Kernel wrapper
CK API reference
CK Tile API reference
Full API class list

About

Contributing to Composable Kernel
License

Composable Kernel: Class Members - Variables

Class Members - Variables

Class Members - Variables#

Composable Kernel: Class Members - Variables

Here is a list of all variables with links to the classes they belong to:

- q -

q_layout : ck_tile::naive_attention_fwd_traits
q_ptr : ck_tile::FmhaBatchPrefillWithPagedKVCacheKernel< FmhaPipeline_, EpiloguePipeline_ >::FmhaFwdCommonKargs, ck_tile::FmhaBwdDQDKDVKernel< FmhaPipeline_, KGradEpiloguePipeline_, VGradEpiloguePipeline_, QGradEpiloguePipeline_ >::FmhaBwdCommonKargs, ck_tile::FmhaFwdAppendKVKernel< FmhaPipeline_ >::BasicKargs, ck_tile::FmhaFwdKernel< FmhaPipeline_, EpiloguePipeline_ >::FmhaFwdCommonKargs, ck_tile::FmhaFwdPagedKVKernel< FmhaPipeline_, EpiloguePipeline_ >::FmhaFwdCommonKargs, ck_tile::FmhaFwdSplitKVKernel< FmhaPipeline_, EpiloguePipeline_ >::CommonKargs, ck_tile::FmhaFwdV3Kernel< FmhaPipeline_, EpiloguePipeline_ >::FmhaFwdCommonKargs, ck_tile::naive_attention_fwd_args
q_sequence_length_ : ck::tensor_operation::device::DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle< NumDimG, NumDimM, NumDimL, NumDimK, NumDimN, ADataType, B0DataType, B1DataType, CDataType, Acc0BiasDataType, Acc0DataType, Acc1BiasDataType, Acc1DataType, CShuffleDataType, AElementwiseOperation, B0ElementwiseOperation, AccElementwiseOperation, B1ElementwiseOperation, CElementwiseOperation, GemmSpec, ASpec, B0Spec, B1Spec, CSpec, NumPrefetch, BlockSize, MPerBlock, LPerBlock, KPerBlock, AK1, BK1, NPerBlock, LTilePerBlock, L1, MPerWmma, LPerWmma, NPerWmma, MRepeat, LRepeat, NRepeat, ABlockTransferThreadClusterLengths_K0_M_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_K1, ABlockLdsAddExtraM, B0BlockTransferThreadClusterLengths_K0_L_K1, B0BlockTransferThreadClusterArrangeOrder, B0BlockTransferSrcAccessOrder, B0BlockTransferSrcVectorDim, B0BlockTransferSrcScalarPerVector, B0BlockTransferDstScalarPerVector_K1, B0BlockLdsAddExtraL, B1BlockTransferThreadClusterLengths_L0_N_L1, B1BlockTransferThreadClusterArrangeOrder, B1BlockTransferSrcAccessOrder, B1BlockTransferSrcVectorDim, B1BlockTransferSrcScalarPerVector, B1BlockTransferDstScalarPerVector_L1, B1BlockLdsAddExtraN, CShuffleMRepeatPerShuffle, CShuffleNRepeatPerShuffle, CShuffleBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CShuffleBlockTransferScalarPerVector_NPerBlock, MaskingSpec, LoopSched, PipelineVer >::CrossAttnArg
q_type : ck_tile::naive_attention_fwd_traits
QK_A : ck_tile::QuantGemmKernelArgs, ck_tile::QuantGemmProblem, ck_tile::QuantGroupedGemmHostArgs
QK_B : ck_tile::QuantGemmKernelArgs, ck_tile::QuantGemmProblem, ck_tile::QuantGroupedGemmHostArgs
QLoadOnce : ck_tile::BlockFmhaPipelineQXCustomPolicy< false >, ck_tile::BlockFmhaPipelineQXCustomPolicy< true >
qo_head_idx : ck_tile::FmhaBatchPrefillWithPagedKVCacheKernel< FmhaPipeline_, EpiloguePipeline_ >::BlockIndices, ck_tile::FmhaFwdKernel< FmhaPipeline_, EpiloguePipeline_ >::BlockIndices, ck_tile::FmhaFwdPagedKVKernel< FmhaPipeline_, EpiloguePipeline_ >::BlockIndices, ck_tile::FmhaFwdSplitKVKernel< FmhaPipeline_, EpiloguePipeline_ >::BlockIndices
qs_element_op_ : ck::tensor_operation::device::DeviceGemmMultipleDMultipleR_Xdl_CShuffle< ALayout, BLayout, DELayout, ADataType, BDataType, GemmAccDataType, CShuffleDataType, DsDataType, EDataType, ReduceAccDataType, RsDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, QsElementwiseOperation, RsElementwiseOperation, ThreadReduceOperations, RsGlobalMemoryDataOperation, 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, CDRThreadTransferClusterLengths_MPerBlock_NPerBlock, CDEReduceThreadTransferScalarPerVector_NPerBlock, RThreadTransferDstScalarPerVector_MPerBlock, LoopSched >::Argument, ck::tensor_operation::device::DeviceGroupedConvFwdMultipleDMultipleR_Xdl_CShuffle< NDimSpatial, ALayout, BLayout, DELayout, RLayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType, ReduceAccDataType, RsDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, QsElementwiseOperation, RsElementwiseOperation, ThreadReduceOperations, RsGlobalMemoryDataOperation, ConvForwardSpecialization, 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, CDRThreadTransferClusterLengths_MPerBlock_NPerBlock, CDEBlockTransferScalarPerVector_NPerBlock, RThreadTransferDstScalarPerVector_MPerBlock, LoopSched >::Argument
QScalesPerBlockRow : ck_tile::BlockGemmWeightPreshuffleBQuantARegBRegCReg< Problem_, BlockPolicy_ >, ck_tile::WPQuantBPipelineAgBgCrV2< Problem, PipelinePolicy >
QScalesPerWarpGemmRow : ck_tile::BlockGemmWeightPreshuffleBQuantARegBRegCReg< Problem_, BlockPolicy_ >
quad_hs : ck_tile::DefaultTranspose< DataType >::ValidationTraitsImpl< InDstrEncode, ReverseDirection, LaneGroupSize >
quad_idx_offset : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_input_hs_lengthss : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_input_ps_to_rhss_major0 : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_input_ps_to_rhss_minor0 : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_output_hs_lengthss : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_output_ps_minor_offset : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_output_ps_to_rhss_major0 : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_output_ps_to_rhss_minor0 : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_output_ys_minor_offset : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_output_ys_to_rhs_major : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_output_ys_to_rhs_minor : ck_tile::TransposeTileDistributionTraits< TileDistributionEncoding_, DataType_, Policy, ReverseDirection >
quad_ps_major0 : ck_tile::DefaultTranspose< DataType >::ValidationTraitsImpl< InDstrEncode, ReverseDirection, LaneGroupSize >
quad_ps_minor0 : ck_tile::DefaultTranspose< DataType >::ValidationTraitsImpl< InDstrEncode, ReverseDirection, LaneGroupSize >
quad_ys_major : ck_tile::DefaultTranspose< DataType >::ValidationTraitsImpl< InDstrEncode, ReverseDirection, LaneGroupSize >
quad_ys_minor : ck_tile::DefaultTranspose< DataType >::ValidationTraitsImpl< InDstrEncode, ReverseDirection, LaneGroupSize >
quant_algo : ck_tile::naive_attention_fwd_kernel_traits< variation_, quant_algo_ >, ck_tile::naive_attention_fwd_traits
QuantPackedSize : ck_tile::F16xMXF4FlatmmKernel< TilePartitioner_, FlatmmPipeline_, EpiloguePipeline_ >
quot : imaxdiv_t

Terms and Conditions
ROCm Licenses and Disclaimers
Privacy
Trademarks
Supply Chain Transparency
Fair and Open Competition
UK Tax Strategy
Cookie Policy
Cookie Settings

© 2026 Advanced Micro Devices, Inc