ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-20 06:49:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

Revert "Optimize fmha fwd decode & prefill for gfx950 (#2641)" (#2670)

Browse Source 
This reverts commit b7322a521a.
This commit is contained in:
asleepzzz
2025-08-12 20:27:10 +08:00
committed by GitHub
parent b7322a521a
commit 5b39de4bb6
31 changed files with 639 additions and 3545 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
include/ck_tile/ops/elementwise/unary_element_wise_operation.hpp
View File

@@ -330,6 +330,13 @@ struct PassThrough
y = type_convert<float>(x);
}
template <>
CK_TILE_HOST_DEVICE void
operator()<ck_tile::bf16_t, ck_tile::fp16_t>(ck_tile::bf16_t& y, const ck_tile::fp16_t& x) const
{
y = type_convert<ck_tile::bf16_t>(x);
}
template <>
CK_TILE_HOST_DEVICE void operator()<float, ck_tile::fp16_t>(float& y,
const ck_tile::fp16_t& x) const