ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-19 22:39:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

Enable storelse for fmha_fwd_trload kernel (#3023)

Browse Source
This commit is contained in:
Haocong WANG
2025-10-16 13:51:23 +08:00
committed by GitHub
parent 0dbd173500
commit 013ba3c737
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
example/ck_tile/01_fmha/codegen/ops/fmha_fwd.py
View File

@@ -608,7 +608,7 @@ class KernelComponentFactory:
else:
pipelines.append(FmhaFwdPipeline('qr_async', 'row', 't', 'f', 't', 't', logits, bias, lse, dropout, squant, mask, skip, 'f'))
pipelines.append(FmhaFwdPipeline('qr_async', 'row', 't', 't', 't', 't', logits, bias, lse, dropout, squant, mask, skip, 'f'))
if (hdim, hdim_v) in [(64, 64), (128, 128)] and logits == "f" and bias == "no" and dropout == "f" and lse == "f" and skip == "f":
if (hdim, hdim_v) in [(64, 64), (128, 128)] and logits == "f" and bias == "no" and dropout == "f" and skip == "f":
pipelines.append(FmhaFwdPipeline('qr_async_trload', 'row', 'f', 'f', 'f', 'f', logits, bias, lse, dropout, squant, mask, skip, 't'))
pipelines.append(FmhaFwdPipeline('qr_async_trload', 'row', 'f', 'f', 't', 't', logits, bias, lse, dropout, squant, mask, skip, 't'))
if receipt == 1 and bias != "bias":