onnxruntime
FlashAttention: use subgroup size 32 on Intel Xe2 and Xe3
#33808
Merged
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Overview
Commits
1
Changes
View On
GitHub
FlashAttention: use subgroup size 32 on Intel Xe2 and Xe3
#33808
qjia7
merged 1 commit into
microsoft:main
from
jchen10:fa_simd32
jchen10
requested a review
from
copilot-pull-request-reviewer
9 days ago
copilot-pull-request-reviewer
commented on 2026-11-02
qjia7
commented on 2026-11-03
jchen10
force pushed
from
ad053b03
to
0a05eb57
7 days ago
jchen10
changed the title
FlashAttention: use subgroup size 32 on Intel
FlashAttention: use subgroup size 16 on Intel
7 days ago
qjia7
dismissed these changes on 2026-11-04
jchen10
dismissed their stale review via
ed92d9f6
7 days ago
jchen10
force pushed
from
0a05eb57
to
ed92d9f6
7 days ago
daijh
commented on 2026-11-05
FlashAttention: use subgroup size 32 on Intel Xe2 and Xe3
ba9e2715
jchen10
force pushed
from
ed92d9f6
to
ba9e2715
6 days ago
jchen10
changed the title
FlashAttention: use subgroup size 16 on Intel
FlashAttention: use subgroup size 32 on Intel Xe2 and Xe3
6 days ago
qjia7
approved these changes on 2026-11-06
qjia7
enabled auto-merge (squash)
5 days ago
jchen10
closed this
2 days ago
disabled auto-merge
2 days ago
Pull request was closed
jchen10
reopened this
2 days ago
qjia7
approved these changes on 2026-11-11
qjia7
merged
47716d36
into main
1 day ago
Login to write a write a comment.
Login via GitHub
Reviewers
qjia7
daijh
Jiawei-Shao
copilot-pull-request-reviewer
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub