ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-03 05:01:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fixing conv bug (#258)

Browse Source 
* debugging conv

* fix oversight where ctile map is constructed before initializing c desc

* example program should returns error code

* clean up

* changed Block2CTileMap in conv2d and convnd

* clean up

* clean up

* cleanup

Co-authored-by: Anthony Chang <ac.chang@outlook.com>
This commit is contained in:
Chao Liu
2022-05-27 09:29:37 -05:00
committed by GitHub
parent 3e6c2610ae
commit 91d8b7d67a
8 changed files with 32 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v3r1.hpp
View File

@@ -314,7 +314,7 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v3r1
using DefaultBlock2CTileMap =
remove_cvref_t<decltype(MakeDefaultBlock2CTileMap(CGridDesc_M_N{}, 1, 1))>;
template <bool HasMainK0BlockLoop, typename Block2CTileMap = DefaultBlock2CTileMap>
template <bool HasMainK0BlockLoop, typename Block2CTileMap>
__device__ static void
Run(const FloatAB* __restrict__ p_a_grid,
const FloatAB* __restrict__ p_b_grid,