GridwiseSoftmax_mk_to_mk< InDataType, OutDataType, AccDataType, GridDesc_M_K, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, SweepOnce > Struct Template Reference#
ck::GridwiseSoftmax_mk_to_mk< InDataType, OutDataType, AccDataType, GridDesc_M_K, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, SweepOnce > Struct Template Reference
#include <gridwise_softmax.hpp>
Public Types | |
| using | ThreadClusterLengths_M_K = Sequence<MThreadClusterSize, KThreadClusterSize> |
| using | ThreadBufferDimAccessOrder |
| using | ThreadClusterArrangeOrder |
| using | ThreadReduceSrcDesc_M_K |
| using | ThreadReduceDstDesc_M |
| using | PassThroughOp = tensor_operation::element_wise::PassThrough |
Static Public Member Functions | |
| static __device__ void | Run (const GridDesc_M_K &in_grid_desc_m_k, const GridDesc_M_K &out_grid_desc_m_k, index_t block_group_size, index_t num_k_block_tile_iteration, AccDataType alpha, const InDataType *const __restrict__ p_in_value_global, AccDataType beta, OutDataType *const __restrict__ p_out_value_global) |
Static Public Attributes | |
| static constexpr bool | reorder_thread_cluster = (InSrcVectorDim == 0) |
| static constexpr auto | thread_cluster_desc |
| static constexpr auto | I0 = Number<0>{} |
| static constexpr auto | I1 = Number<1>{} |
| static constexpr index_t | M_BlockTileSize = MThreadClusterSize * MThreadSliceSize |
| static constexpr index_t | K_BlockTileSize = KThreadClusterSize * KThreadSliceSize |
Member Typedef Documentation
◆ PassThroughOp
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
| using ck::GridwiseSoftmax_mk_to_mk< InDataType, OutDataType, AccDataType, GridDesc_M_K, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, SweepOnce >::PassThroughOp = tensor_operation::element_wise::PassThrough |
◆ ThreadBufferDimAccessOrder
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
| using ck::GridwiseSoftmax_mk_to_mk< InDataType, OutDataType, AccDataType, GridDesc_M_K, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, SweepOnce >::ThreadBufferDimAccessOrder |
Initial value:
Definition utility/sequence.hpp:43
Definition utility/functional.hpp:100
◆ ThreadClusterArrangeOrder
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
| using ck::GridwiseSoftmax_mk_to_mk< InDataType, OutDataType, AccDataType, GridDesc_M_K, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, SweepOnce >::ThreadClusterArrangeOrder |
Initial value:
◆ ThreadClusterLengths_M_K
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
| using ck::GridwiseSoftmax_mk_to_mk< InDataType, OutDataType, AccDataType, GridDesc_M_K, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, SweepOnce >::ThreadClusterLengths_M_K = Sequence<MThreadClusterSize, KThreadClusterSize> |
◆ ThreadReduceDstDesc_M
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
| using ck::GridwiseSoftmax_mk_to_mk< InDataType, OutDataType, AccDataType, GridDesc_M_K, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, SweepOnce >::ThreadReduceDstDesc_M |
Initial value:
__host__ __device__ constexpr auto make_naive_tensor_descriptor_packed(const Tuple< Lengths... > &lengths)
Definition tensor_descriptor_helper.hpp:101
__host__ __device__ constexpr auto make_tuple(Xs &&... xs)
Definition utility/tuple.hpp:211
◆ ThreadReduceSrcDesc_M_K
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
| using ck::GridwiseSoftmax_mk_to_mk< InDataType, OutDataType, AccDataType, GridDesc_M_K, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, SweepOnce >::ThreadReduceSrcDesc_M_K |
Initial value:
Member Function Documentation
◆ Run()
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
|
inlinestatic |
max(x)
sum(exp(x - max(x)))
softmax
Member Data Documentation
◆ I0
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
|
staticconstexpr |
◆ I1
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
|
staticconstexpr |
◆ K_BlockTileSize
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
|
staticconstexpr |
◆ M_BlockTileSize
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
|
staticconstexpr |
◆ reorder_thread_cluster
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
|
staticconstexpr |
◆ thread_cluster_desc
template<typename InDataType, typename OutDataType, typename AccDataType, typename GridDesc_M_K, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t InSrcVectorDim, index_t InSrcVectorSize, index_t OutDstVectorSize, bool SweepOnce>
|
staticconstexpr |
Initial value:
=
__host__ __device__ constexpr auto make_cluster_descriptor(const Lengths &lengths, ArrangeOrder order=typename arithmetic_sequence_gen< 0, Lengths::Size(), 1 >::type{})
Definition tensor_description/cluster_descriptor.hpp:13
Sequence< MThreadClusterSize, KThreadClusterSize > ThreadClusterLengths_M_K
Definition gridwise_softmax.hpp:63
typename conditional< reorder_thread_cluster, Sequence< 1, 0 >, Sequence< 0, 1 > >::type ThreadClusterArrangeOrder
Definition gridwise_softmax.hpp:68
The documentation for this struct was generated from the following file: