ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-20 04:49:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add instances for conv_scale with fp8@bf8->fp8 (#1220)

Browse Source 
* Update device op api to support BComputeType

* Add example

* Add instances

* Add profiler mode

* Add client example

* Update copyright year

* Add BComputeType check

* Fix compute types

[ROCm/composable_kernel commit: a61e73bc56]
This commit is contained in:
Rostyslav Geyyer
2024-04-03 09:08:08 -05:00
committed by GitHub
parent 0adb068ce8
commit 0b8e766e55
17 changed files with 441 additions and 103 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/CMakeLists.txt
View File

@@ -41,4 +41,9 @@ if(DTYPES MATCHES "bf8" OR NOT DEFINED DTYPES)
xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_bf8_instance.cpp)
endif()
if((DTYPES MATCHES "fp8" AND DTYPES MATCHES "bf8") OR NOT DEFINED DTYPES)
list(APPEND GROUPED_CONV3D_FWD
xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_fp8_bf8_instance.cpp)
endif()
add_instance_library(device_grouped_conv3d_fwd_instance ${GROUPED_CONV3D_FWD})

54
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_fp8_bf8_instance.cpp Normal file
View File

@@ -0,0 +1,54 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
#include "ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_instance.hpp"
#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
namespace ck {
namespace tensor_operation {
namespace device {
namespace instance {
void add_device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f8_bf8_instances(
std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleABD<3,
NDHWGC,
GKZYXC,
Empty_Tuple,
NDHWGK,
F8,
BF8,
Empty_Tuple,
F8,
PassThrough,
PassThrough,
PassThrough,
F8,
BF8>>>& instances)
{
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f8_bf8_instances<3,
NDHWGC,
GKZYXC,
Empty_Tuple,
NDHWGK,
ConvFwdDefault>{});
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f8_bf8_instances<3,
NDHWGC,
GKZYXC,
Empty_Tuple,
NDHWGK,
ConvFwd1x1P0>{});
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f8_bf8_instances<3,
NDHWGC,
GKZYXC,
Empty_Tuple,
NDHWGK,
ConvFwd1x1S1P0>{});
}
} // namespace instance
} // namespace device
} // namespace tensor_operation
} // namespace ck