ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-12 09:16:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

behavior has changed (better and worse), figuring out why

Browse Source
This commit is contained in:
Chao Liu
2019-05-21 16:43:56 -05:00
parent acd7082fe1
commit 2a48812edb
5 changed files with 16 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
driver/device_convolution_implicit_gemm_v1_nchw_cyxk_nkhw.hpp
View File

@@ -57,7 +57,7 @@ void device_convolution_implicit_gemm_v1_nchw_cyxk_nkhw(InDesc,
wei_cyxk_device_buf.ToDevice(wei_cyxk.mData.data());
out_nkhw_device_buf.ToDevice(out_nkhw.mData.data());
#if 1
#if 0
// for 3x3, 34x34, v1r3, Pascal
constexpr index_t BlockSize = 128;
@@ -162,7 +162,7 @@ void device_convolution_implicit_gemm_v1_nchw_cyxk_nkhw(InDesc,
constexpr index_t WeiBlockCopyDataPerRead_K = 4;
constexpr index_t OutThreadCopyDataPerWrite_W = 2;
#elif 0
#elif 1
// for 3x3, 34x34, v1r3, Vega 20, WoPerBlock = 8
constexpr index_t BlockSize = 256;