Skip to content

Avoid unused quantized SDPA scratch allocation - #22417

Draft
JacobSzwejbka wants to merge 1 commit into
perf/cpu-ring-attention-mask-skippingfrom
perf/cpu-sdpa-quant-scratch-allocation
Draft

Avoid unused quantized SDPA scratch allocation#22417
JacobSzwejbka wants to merge 1 commit into
perf/cpu-ring-attention-mask-skippingfrom
perf/cpu-sdpa-quant-scratch-allocation

Conversation

@JacobSzwejbka

Copy link
Copy Markdown
Contributor

Summary

  • allocate QDQ scratch space only for quantized SDPA
  • leave floating-point SDPA without the unused per-thread kv_split_size * head_size buffer
  • preserve the existing quantized execution path

Stack

Testing

  • existing CPU SDPA correctness coverage in the parent stack
  • verified the stacked diff is limited to op_sdpa_impl.h

@pytorch-bot

pytorch-bot Bot commented Sep 1, 2026

Copy link
Copy Markdown

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/22417

Note: Links to docs will display an error until the docs builds have been completed.

❌ 3 New Failures, 6 Unrelated Failures

As of commit 64cb0bc with merge base 6755ea3 (image):

NEW FAILURES - The following jobs have failed:

FLAKY - The following jobs failed but were likely due to flakiness present on trunk:

BROKEN TRUNK - The following jobs failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@meta-cla meta-cla Bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Sep 1, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant