ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-19 14:29:05 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix a typo in intrin_wmma_bf16_16x16x16_bf16_w32 (#2727)

Browse Source 
__builtin_amdgcn_wmma_bf16_16x16x16_bf16_w32 is only available in gfx11.
This commit is contained in:
linqunAMD
2025-09-03 08:07:09 +08:00
committed by GitHub
parent 8d43155bce
commit 00fd72b2d4
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
include/ck/utility/amd_wmma.hpp
View File

@@ -104,7 +104,7 @@ struct intrin_wmma_bf16_16x16x16_bf16_w32<16, 16, Opsel>
// opsel usage // opsel usage
// false: D0.[0:15] = result // false: D0.[0:15] = result
// true : D0.[16:31]= result // true : D0.[16:31]= result
#if defined(__gfx11__) || defined(__gfx12__) #if defined(__gfx11__)
reg_c.template AsType<bhalf16_t>()(Number<0>{}) = reg_c.template AsType<bhalf16_t>()(Number<0>{}) =
__builtin_amdgcn_wmma_bf16_16x16x16_bf16_w32( __builtin_amdgcn_wmma_bf16_16x16x16_bf16_w32(
reg_a, reg_b, reg_c.template AsType<bhalf16_t>()[Number<0>{}], Opsel); reg_a, reg_b, reg_c.template AsType<bhalf16_t>()[Number<0>{}], Opsel);