onnxruntime
[WebGPU] Initial PagedAttention implementation (1/n)
#31611
Merged
Overview
Commits
22
Changes
View On GitHub

[WebGPU] Initial PagedAttention implementation (1/n) #31611

hariharans29 merged 22 commits into main from hari/webgpu_paged_attention
hariharans29
hariharans29 [WebGPU] Add PagedAttention kernel skeleton (Phase 0)
be19980d
hariharans29 [PagedAttention] Move paged_attention_helper.h from cuda/ to cpu/
58deec9d
hariharans29 [WebGPU] PagedAttention: input validation + dispatch skeleton (Phase 1a)
48ed6b1a
hariharans29 [WebGPU] PagedAttention: scatter K/V into paged cache program + tests…
8ad5c732
hariharans29 [WebGPU] PagedAttention: fused rotary embedding for non-packed layout…
e56790a1
hariharans29 [WebGPU] PagedAttention: packed QKV + rotary (Phase 1b.2b)
d24235ce
hariharans29 [WebGPU] PagedAttention: v1 fallback attention (gather-then-flash) + …
1d956641
hariharans29 Merge remote-tracking branch 'origin' into hari/webgpu_paged_attention
37764222
hariharans29 hariharans29 requested a review from copilot-pull-request-reviewer copilot-pull-request-reviewer 18 days ago
copilot-pull-request-reviewer
copilot-pull-request-reviewer commented on 2026-08-03
hariharans29 [WebGPU] PagedAttention: address Copilot review comments
44ab1b66
hariharans29 hariharans29 requested a review from copilot-pull-request-reviewer copilot-pull-request-reviewer 18 days ago
copilot-pull-request-reviewer
copilot-pull-request-reviewer commented on 2026-08-03
hariharans29 [WebGPU] PagedAttention: clarify phased-plan labels in design doc
e1b5a73e
hariharans29 hariharans29 added ep:WebGPU
tianleiwu
tianleiwu commented on 2026-08-04
hariharans29
hariharans29 PR feedback 1
221cf642
github-actions
github-actions commented on 2026-08-05
github-advanced-security
github-advanced-security commented on 2026-08-05
hariharans29 Lint changes
d15f7811
hariharans29 PR feedback
4036b86b
hariharans29 PR feedback
0f0203b2
tianleiwu
tianleiwu dismissed these changes on 2026-08-05
qjia7
qjia7 commented on 2026-08-05
hariharans29 hariharans29 requested a review from copilot-pull-request-reviewer copilot-pull-request-reviewer 16 days ago
copilot-pull-request-reviewer
copilot-pull-request-reviewer commented on 2026-08-05
hariharans29 Merge remote-tracking branch 'origin/main' into hari/webgpu_paged_att…
0bb9324f
hariharans29 hariharans29 dismissed their stale review via 0bb9324f 16 days ago
hariharans29 WebGPU PagedAttention: adapt to merged schema surface
4b58b5cb
github-actions
github-actions commented on 2026-08-05
hariharans29 Update onnxruntime/contrib_ops/webgpu/bert/paged_attention.h
610194d1
hariharans29 Move PagedAttention correctness tests to contrib ops
37b4d51d
hariharans29 Merge remote-tracking branch 'origin/hari/webgpu_paged_attention' int…
9868a63b
hariharans29 test: avoid CUDA aliasing failure in shared paged attention OpTester
354e04a4
qjia7
qjia7 approved these changes on 2026-08-06
hariharans29 Merge remote-tracking branch 'origin' into hari/webgpu_paged_attention
674451f9
hariharans29 hariharans29 enabled auto-merge (squash) 15 days ago
disabled auto-merge 15 days ago
Manually disabled by user
hariharans29 hariharans29 closed this 15 days ago
hariharans29 hariharans29 reopened this 15 days ago
hariharans29 Merge remote-tracking branch 'origin' into hari/webgpu_paged_attention
203e5a1c
hariharans29 hariharans29 enabled auto-merge (squash) 15 days ago
tianleiwu
tianleiwu dismissed these changes on 2026-08-07
tianleiwu tianleiwu dismissed their stale review 15 days ago
It's fine that webgpu requires the output.
tianleiwu
tianleiwu approved these changes on 2026-08-07
hariharans29 hariharans29 merged 62aced3b into main 14 days ago
hariharans29 hariharans29 deleted the hari/webgpu_paged_attention branch 14 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
ep:WebGPU
Milestone
No milestone
Terms Of Use Privacy Policy Refund Policy Impressum