/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/develop/include/ck/tensor_operation/gpu/grid/gridwise_softmax.hpp File Reference

• develop
• include
• ck
• tensor_operation
• gpu
• grid

Classes | Namespaces | Functions

gridwise_softmax.hpp File Reference

#include "ck/utility/data_type.hpp"
#include "ck/utility/reduction_common.hpp"
#include "ck/utility/reduction_operator.hpp"
#include "ck/utility/reduction_functions_accumulate.hpp"
#include "ck/tensor_operation/gpu/block/reduction_functions_blockwise.hpp"
#include "ck/tensor_operation/gpu/thread/reduction_functions_threadwise.hpp"
#include "ck/tensor_operation/gpu/thread/threadwise_tensor_slice_transfer.hpp"
#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"

Go to the source code of this file.

Classes
struct	ck::GridwiseSoftmax_mk_to_mk< InDataType, OutDataType, AccDataType, GridDesc_M_K, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, InSrcVectorDim, InSrcVectorSize, OutDstVectorSize, SweepOnce >

Namespaces
	ck

Functions
template<typename GridwiseReduction , typename InDataType , typename OutDataType , typename AccDataType , typename GridDesc_M_K >
__global__ void	ck::kernel_softmax (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)