ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-16 19:09:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

adding int8 direct that reads pre-vectorized data

Browse Source 
[ROCm/composable_kernel commit: 050a1a6890]
This commit is contained in:
Chao Liu
2019-03-19 00:05:41 -05:00
parent 14ad082d6c
commit 80f2ab2e09
6 changed files with 247 additions and 95 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/include/threadwise_direct_convolution.hip.hpp
View File

@@ -51,10 +51,8 @@ __device__ void threadwise_direct_convolution_1(InDesc,
const unsigned out_index = out_desc.Get1dIndex(n, k, ho, wo);
fused_multiply_add(p_out[out_index],
p_wei[wei_index],
p_in[in_index],
p_out[out_index]);
fused_multiply_accumulate(
p_out[out_index], p_wei[wei_index], p_in[in_index]);
}
}
}