gridwise_gemm_dl_v1r3.hpp File Reference

gridwise_gemm_dl_v1r3.hpp File Reference#

Composable Kernel: gridwise_gemm_dl_v1r3.hpp File Reference
Classes | Namespaces | Functions
gridwise_gemm_dl_v1r3.hpp File Reference
#include "ck/utility/common_header.hpp"
#include "ck/tensor_description/multi_index_transform_helper.hpp"
#include "ck/tensor_description/tensor_descriptor.hpp"
#include "ck/tensor_description/tensor_descriptor_helper.hpp"
#include "ck/tensor_operation/gpu/grid/block_to_ctile_map.hpp"
#include "ck/tensor_operation/gpu/grid/gridwise_gemm_pipeline_v1.hpp"
#include "ck/tensor_operation/gpu/block/blockwise_gemm_dl_v2r3.hpp"
#include "ck/tensor_operation/gpu/block/blockwise_tensor_slice_transfer_v5r1.hpp"
#include "ck/tensor_operation/gpu/thread/threadwise_tensor_slice_transfer.hpp"
#include "ck/tensor_operation/gpu/thread/threadwise_tensor_slice_set.hpp"
#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"

Go to the source code of this file.

Classes

struct  ck::GridwiseGemmDl_km_kn_mn_v1r3< BlockSize, FloatAB, FloatAcc, FloatC, CGlobalMemoryDataOperation, AGridDesc_K0_M_K1, BGridDesc_K0_N_K1, CGridDesc_M_N, MPerBlock, NPerBlock, K0PerBlock, K1Value, M1PerThreadM111, N1PerThreadN111, KPerThread, M11N11ThreadClusterM110Xs, M11N11ThreadClusterN110Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector >
struct  ck::GridwiseGemmDl_bkm_bkn_mn_v1r3< BlockSize, FloatAB, FloatAcc, FloatC, CGlobalMemoryDataOperation, AGridDesc_B_K0_M_K1, BGridDesc_B_K0_N_K1, CGridDesc_M_N, MPerBlock, NPerBlock, K0PerBlock, K1Value, M1PerThreadM111, N1PerThreadN111, KPerThread, M11N11ThreadClusterM110Xs, M11N11ThreadClusterN110Xs, ABlockTransferThreadSliceLengths_K0_M0_M1_K1, ABlockTransferThreadClusterLengths_K0_M0_M1_K1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1, ABlockTransferSrcVectorTensorContiguousDimOrder, ABlockTransferDstVectorTensorLengths_K0_M0_M1_K1, BBlockTransferThreadSliceLengths_K0_N0_N1_K1, BBlockTransferThreadClusterLengths_K0_N0_N1_K1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1, BBlockTransferSrcVectorTensorContiguousDimOrder, BBlockTransferDstVectorTensorLengths_K0_N0_N1_K1, CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstVectorDim, CThreadTransferDstScalarPerVector >

Namespaces

namespace  ck

Functions

template<typename GridwiseGemm, typename FloatAB, typename FloatC, typename AGridDesc_K0_M0_M1_K1, typename BGridDesc_K0_N0_N1_K1, typename CGridDesc_M0_M10_M11_N0_N10_N11, typename Block2CTileMap, bool HasMainKBlockLoop, bool HasDoubleTailKBlockLoop>
__global__ void ck::kernel_gemm_dl_v1r3 (const FloatAB *__restrict__ p_a_grid, const FloatAB *__restrict__ p_b_grid, FloatC *__restrict__ p_c_grid, const AGridDesc_K0_M0_M1_K1 a_grid_desc_k0_m0_m1_k1, const BGridDesc_K0_N0_N1_K1 b_grid_desc_k0_n0_n1_k1, const CGridDesc_M0_M10_M11_N0_N10_N11 c_grid_desc_m0_m10_m11_n0_n10_n11, const Block2CTileMap block_2_ctile_map)