GemmKernelArgs< NumDTensor > Struct Template Reference

GemmKernelArgs&lt; NumDTensor &gt; Struct Template Reference#

Composable Kernel: ck_tile::GemmKernelArgs< NumDTensor > Struct Template Reference
ck_tile::GemmKernelArgs< NumDTensor > Struct Template Reference

The GEMM kernel device arguments. More...

#include <gemm_kernel.hpp>

Inheritance diagram for ck_tile::GemmKernelArgs< NumDTensor >:
ck_tile::BatchedGemmKernel< TilePartitioner_, GemmPipeline_, EpiloguePipeline_ >::BatchedGemmKernelArgs

Public Attributes

const void * a_ptr
 The A input tensor's pointer to device memory. More...
 
const void * b_ptr
 The B input tensor's pointer to device memory. More...
 
const std::array< const void *, NumDTensor > ds_ptr
 The Ds input tensor's pointer to device memory. More...
 
void * e_ptr
 The E output tensor's pointer to device memory. More...
 
index_t M
 GEMM's M dimension size. More...
 
index_t N
 GEMM's N dimension size. More...
 
index_t K
 GEMM's K dimension size. More...
 
index_t stride_A
 The distance between consecutive elements of non-contiguous dimension (in memory) of A tensor. More...
 
index_t stride_B
 The distance between consecutive elements of non-contiguous dimension (in memory) of B tensor. More...
 
std::array< index_t, NumDTensor > stride_Ds
 The distance between consecutive elements of non-contiguous dimension (in memory) of Ds tensor. More...
 
index_t stride_E
 The distance between consecutive elements of non-contiguous dimension (in memory) of E tensor. More...
 
index_t k_batch
 

Detailed Description

template<index_t NumDTensor = 0>
struct ck_tile::GemmKernelArgs< NumDTensor >

The GEMM kernel device arguments.

Member Data Documentation

◆ a_ptr

template<index_t NumDTensor = 0>
const void* ck_tile::GemmKernelArgs< NumDTensor >::a_ptr

The A input tensor's pointer to device memory.

◆ b_ptr

template<index_t NumDTensor = 0>
const void* ck_tile::GemmKernelArgs< NumDTensor >::b_ptr

The B input tensor's pointer to device memory.

◆ ds_ptr

template<index_t NumDTensor = 0>
const std::array<const void*, NumDTensor> ck_tile::GemmKernelArgs< NumDTensor >::ds_ptr

The Ds input tensor's pointer to device memory.

◆ e_ptr

template<index_t NumDTensor = 0>
void* ck_tile::GemmKernelArgs< NumDTensor >::e_ptr

The E output tensor's pointer to device memory.

◆ K

template<index_t NumDTensor = 0>
index_t ck_tile::GemmKernelArgs< NumDTensor >::K

GEMM's K dimension size.

◆ k_batch

template<index_t NumDTensor = 0>
index_t ck_tile::GemmKernelArgs< NumDTensor >::k_batch

◆ M

template<index_t NumDTensor = 0>
index_t ck_tile::GemmKernelArgs< NumDTensor >::M

GEMM's M dimension size.

◆ N

template<index_t NumDTensor = 0>
index_t ck_tile::GemmKernelArgs< NumDTensor >::N

GEMM's N dimension size.

◆ stride_A

template<index_t NumDTensor = 0>
index_t ck_tile::GemmKernelArgs< NumDTensor >::stride_A

The distance between consecutive elements of non-contiguous dimension (in memory) of A tensor.

◆ stride_B

template<index_t NumDTensor = 0>
index_t ck_tile::GemmKernelArgs< NumDTensor >::stride_B

The distance between consecutive elements of non-contiguous dimension (in memory) of B tensor.

◆ stride_Ds

template<index_t NumDTensor = 0>
std::array<index_t, NumDTensor> ck_tile::GemmKernelArgs< NumDTensor >::stride_Ds

The distance between consecutive elements of non-contiguous dimension (in memory) of Ds tensor.

◆ stride_E

template<index_t NumDTensor = 0>
index_t ck_tile::GemmKernelArgs< 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:
  • /home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/docs-7.0.0/include/ck_tile/ops/gemm/kernel/gemm_kernel.hpp