device_grouped_gemm_tile_loop.hpp Source File

device_grouped_gemm_tile_loop.hpp Source File#

Composable Kernel: device_grouped_gemm_tile_loop.hpp Source File
device_grouped_gemm_tile_loop.hpp
Go to the documentation of this file.
1// SPDX-License-Identifier: MIT
2// Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
3
4#pragma once
5
6#include "device_grouped_gemm.hpp"
7
8namespace ck {
9namespace tensor_operation {
10namespace device {
11
20
21template <typename ALayout,
22 typename BLayout,
23 typename DsLayout,
24 typename ELayout,
25 typename ADataType,
26 typename BDataType,
27 typename DsDataType,
28 typename EDataType,
29 typename AElementwiseOperation,
30 typename BElementwiseOperation,
31 typename CDEElementwiseOperation>
32struct DeviceGroupedGemmTileLoop : public DeviceGroupedGemm<ALayout,
33 BLayout,
34 DsLayout,
35 ELayout,
36 ADataType,
37 BDataType,
38 DsDataType,
39 EDataType,
40 AElementwiseOperation,
41 BElementwiseOperation,
42 CDEElementwiseOperation>
43{
44};
45
46} // namespace device
47} // namespace tensor_operation
48} // namespace ck
device_grouped_gemm.hpp
ck::tensor_operation::device
Definition convolution_backward_data_specialization.hpp:8
ck::tensor_operation
Definition convolution_backward_data_specialization.hpp:7
ck
Definition ck.hpp:268
ck::tensor_operation::device::DeviceGroupedGemm
Definition device_grouped_gemm.hpp:99
ck::tensor_operation::device::DeviceGroupedGemmTileLoop
Grouped GEMM kernel using output Tile Looping algorithm.
Definition device_grouped_gemm_tile_loop.hpp:43