ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-06-28 18:56:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add call to DeviceGemmXdl<> instances in library (#1174)

Browse Source
This commit is contained in:
Po Yen Chen
2024-02-23 20:18:16 +08:00
committed by GitHub
parent 4d96aae39f
commit 62db01c2cc
1 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
library/include/ck/library/tensor_operation_instance/gpu/gemm.hpp
View File

@@ -454,7 +454,7 @@ struct DeviceOperationInstanceFactory<
if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Row> &&
is_same_v<CLayout, Row>)
{
/// add_device_gemm_xdl_f16_f16_f16_mk_kn_mn_instances(op_ptrs);
add_device_gemm_xdl_f16_f16_f16_mk_kn_mn_instances(op_ptrs);
#ifdef DL_KERNELS
add_device_gemm_dl_f16_f16_f16_mk_kn_mn_instances(op_ptrs);
add_device_gemm_dl_f16_f16_f16_mk_kn_mn_irregular_instances(op_ptrs);
@@ -466,7 +466,7 @@ struct DeviceOperationInstanceFactory<
else if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Col> &&
is_same_v<CLayout, Row>)
{
/// add_device_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(op_ptrs);
add_device_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(op_ptrs);
#ifdef DL_KERNELS
add_device_gemm_dl_f16_f16_f16_mk_nk_mn_instances(op_ptrs);
add_device_gemm_dl_f16_f16_f16_mk_nk_mn_irregular_instances(op_ptrs);
@@ -481,7 +481,7 @@ struct DeviceOperationInstanceFactory<
else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Row> &&
is_same_v<CLayout, Row>)
{
/// add_device_gemm_xdl_f16_f16_f16_km_kn_mn_instances(op_ptrs);
add_device_gemm_xdl_f16_f16_f16_km_kn_mn_instances(op_ptrs);
#ifdef DL_KERNELS
add_device_gemm_dl_f16_f16_f16_km_kn_mn_instances(op_ptrs);
add_device_gemm_dl_f16_f16_f16_km_kn_mn_irregular_instances(op_ptrs);
@@ -493,7 +493,7 @@ struct DeviceOperationInstanceFactory<
else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Col> &&
is_same_v<CLayout, Row>)
{
/// add_device_gemm_xdl_f16_f16_f16_km_nk_mn_instances(op_ptrs);
add_device_gemm_xdl_f16_f16_f16_km_nk_mn_instances(op_ptrs);
#ifdef DL_KERNELS
add_device_gemm_dl_f16_f16_f16_km_nk_mn_instances(op_ptrs);
add_device_gemm_dl_f16_f16_f16_km_nk_mn_irregular_instances(op_ptrs);