ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-05 14:11:29 +00:00
Code Issues Packages Projects Releases Wiki Activity

use old ctile to avoid conv2d fwd bias relu add compute error (#271)

Browse Source
This commit is contained in:
Shaojie WANG
2022-06-03 03:06:42 +08:00
committed by GitHub
parent 86185bd7ce
commit 1c5d06f270
3 changed files with 13 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v3r3.hpp
View File

@@ -340,7 +340,7 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v3r3
using DefaultBlock2CTileMap =
remove_cvref_t<decltype(MakeDefaultBlock2CTileMap(CGridDesc_M_N{}, 1, 1))>;
template <bool HasMainKBlockLoop, typename Block2CTileMap = DefaultBlock2CTileMap>
template <bool HasMainKBlockLoop, typename Block2CTileMap>
__device__ static void
Run(const FloatAB* __restrict__ p_a_grid,
const FloatAB* __restrict__ p_b_grid,