GridwiseGemmPipeline_v4< 2 > Struct Reference

GridwiseGemmPipeline_v4< 2 > Struct Reference#

Composable Kernel: ck::GridwiseGemmPipeline_v4< 2 > Struct Reference

#include <gridwise_gemm_pipeline_v4_direct_load.hpp>

Static Public Member Functions
__host__ static __device__ constexpr bool	IsSupported (index_t num_loop)
__host__ static __device__ constexpr bool	CalculateHasMainLoop (index_t num_loop)
template<bool HasMainLoop, typename AGridDesc, typename ABlockDesc, typename ABlockTransfer, typename AGridBuffer, typename ABlockBuffers, typename ABlockTransferStep, typename BGridDesc, typename BBlockDesc, typename BBlockTransfer, typename BGridBuffer, typename BBlockBuffers, typename BBlockTransferStep, typename BlockwiseGemm, typename CThreadBuffer>
static __device__ void	Run (const AGridDesc &a_grid_desc, const ABlockDesc &a_block_desc, ABlockTransfer &a_blockwise_copy, const AGridBuffer &a_grid_buf, ABlockBuffers &a_block_bufs, const ABlockTransferStep &a_block_copy_step, const BGridDesc &b_grid_desc, const BBlockDesc &b_block_desc, BBlockTransfer &b_blockwise_copy, const BGridBuffer &b_grid_buf, BBlockBuffers &b_block_bufs, const BBlockTransferStep &b_block_copy_step, const BlockwiseGemm &blockwise_gemm, CThreadBuffer &c_thread_buf, index_t num_loop)

Static Public Attributes
static constexpr auto	I0 = Number<0>{}
static constexpr auto	I1 = Number<1>{}

Member Function Documentation

◆ CalculateHasMainLoop()

__host__ static __device__ constexpr bool ck::GridwiseGemmPipeline_v4< 2 >::CalculateHasMainLoop ( index_t num_loop )

inlinestaticconstexpr

◆ IsSupported()

__host__ static __device__ constexpr bool ck::GridwiseGemmPipeline_v4< 2 >::IsSupported ( index_t num_loop )

inlinestaticconstexpr

◆ Run()

template<bool HasMainLoop, typename AGridDesc, typename ABlockDesc, typename ABlockTransfer, typename AGridBuffer, typename ABlockBuffers, typename ABlockTransferStep, typename BGridDesc, typename BBlockDesc, typename BBlockTransfer, typename BGridBuffer, typename BBlockBuffers, typename BBlockTransferStep, typename BlockwiseGemm, typename CThreadBuffer>

__device__ void ck::GridwiseGemmPipeline_v4< 2 >::Run	(	const AGridDesc &	a_grid_desc,
		const ABlockDesc &	a_block_desc,
		ABlockTransfer &	a_blockwise_copy,
		const AGridBuffer &	a_grid_buf,
		ABlockBuffers &	a_block_bufs,
		const ABlockTransferStep &	a_block_copy_step,
		const BGridDesc &	b_grid_desc,
		const BBlockDesc &	b_block_desc,
		BBlockTransfer &	b_blockwise_copy,
		const BGridBuffer &	b_grid_buf,
		BBlockBuffers &	b_block_bufs,
		const BBlockTransferStep &	b_block_copy_step,
		const BlockwiseGemm &	blockwise_gemm,
		CThreadBuffer &	c_thread_buf,
		index_t	num_loop )