ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-23 22:34:36 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix conv2d bwd data bug when filter is 1x1 and stride = 2 (#132)

Browse Source 
* fix bwd data filter1strid2 bug

* fichangeshort to ck::bhalf_t

* reset input to zero

Co-authored-by: ltqin <letaoqin@amd.com>

[ROCm/composable_kernel commit: b51808d7a5]
This commit is contained in:
ltqin
2022-03-21 23:53:23 +08:00
committed by GitHub
parent 900ea4ae3e
commit bb66a0b0d6
5 changed files with 24 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
View File

@@ -9,7 +9,7 @@ namespace tensor_operation {
namespace device {
namespace device_conv2d_bwd_data_instance {
using BF16 = ushort;
using BF16 = ck::bhalf_t;
using F32 = float;
template <ck::index_t... Is>