ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-24 23:05:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

added tuple

Browse Source 
[ROCm/composable_kernel commit: 625838def0]
This commit is contained in:
Chao Liu
2019-09-06 18:07:56 -05:00
parent d0a36e9a40
commit b0f3708397
10 changed files with 369 additions and 106 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
driver/include/host_conv.hpp
View File

@@ -65,9 +65,6 @@ void host_direct_convolution(const Tensor<TIn>& in_nchw,
index_t h_pad_low = LowerPads{}.Get(Number<0>{});
index_t w_pad_low = LowerPads{}.Get(Number<1>{});
index_t h_pad_up = UpperPads{}.Get(Number<0>{});
index_t w_pad_up = UpperPads{}.Get(Number<1>{});
auto f = [&](auto n, auto k, auto ho, auto wo) {
double v = 0;
for(int c = 0; c < wei_kcyx.mDesc.GetLengths()[1]; ++c)
@@ -125,9 +122,6 @@ void host_winograd_3x3_convolution(const Tensor<TIn>& in_nchw,
index_t h_pad_low = LowerPads{}.Get(Number<0>{});
index_t w_pad_low = LowerPads{}.Get(Number<1>{});
index_t h_pad_up = UpperPads{}.Get(Number<0>{});
index_t w_pad_up = UpperPads{}.Get(Number<1>{});
std::size_t HiPerTile = HoPerTile + Y - 1;
std::size_t WiPerTile = WoPerTile + X - 1;