UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor > Struct Template Reference#
The GEMM kernel device arguments. More...
#include <universal_gemm_kernel.hpp>
Public Attributes | |
| const std::array< const void *, NumATensor > | as_ptr |
| The As input tensor's pointer to device memory. | |
| const std::array< const void *, NumBTensor > | bs_ptr |
| The Bs input tensor's pointer to device memory. | |
| const std::array< const void *, NumDTensor > | ds_ptr |
| The Ds input tensor's pointer to device memory. | |
| void * | e_ptr |
| The E output tensor's pointer to device memory. | |
| index_t | M |
| GEMM's M dimension size. | |
| index_t | N |
| GEMM's N dimension size. | |
| index_t | K |
| GEMM's K dimension size. | |
| std::array< index_t, NumATensor > | stride_As |
| The distance between consecutive elements of non-contiguous dimension (in memory) of As tensor. | |
| std::array< index_t, NumBTensor > | stride_Bs |
| The distance between consecutive elements of non-contiguous dimension (in memory) of Bs tensor. | |
| std::array< index_t, NumDTensor > | stride_Ds |
| The distance between consecutive elements of non-contiguous dimension (in memory) of Ds tensor. | |
| index_t | stride_E |
| The distance between consecutive elements of non-contiguous dimension (in memory) of E tensor. | |
| index_t | k_batch |
Detailed Description
struct ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >
The GEMM kernel device arguments.
Member Data Documentation
◆ as_ptr
| const std::array<const void*, NumATensor> ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::as_ptr |
The As input tensor's pointer to device memory.
◆ bs_ptr
| const std::array<const void*, NumBTensor> ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::bs_ptr |
The Bs input tensor's pointer to device memory.
◆ ds_ptr
| const std::array<const void*, NumDTensor> ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::ds_ptr |
The Ds input tensor's pointer to device memory.
◆ e_ptr
| void* ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::e_ptr |
The E output tensor's pointer to device memory.
◆ K
| index_t ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::K |
GEMM's K dimension size.
◆ k_batch
| index_t ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::k_batch |
◆ M
| index_t ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::M |
GEMM's M dimension size.
◆ N
| index_t ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::N |
GEMM's N dimension size.
◆ stride_As
| std::array<index_t, NumATensor> ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::stride_As |
The distance between consecutive elements of non-contiguous dimension (in memory) of As tensor.
◆ stride_Bs
| std::array<index_t, NumBTensor> ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::stride_Bs |
The distance between consecutive elements of non-contiguous dimension (in memory) of Bs tensor.
◆ stride_Ds
| std::array<index_t, NumDTensor> ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::stride_Ds |
The distance between consecutive elements of non-contiguous dimension (in memory) of Ds tensor.
◆ stride_E
| index_t ck_tile::UniversalGemmKernelArgs< NumATensor, NumBTensor, NumDTensor >::stride_E |
The distance between consecutive elements of non-contiguous dimension (in memory) of E tensor.
The documentation for this struct was generated from the following file: