Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[CPU] Improve vector size selection in vectorization pass #14403

Closed
wants to merge 1 commit into from

Conversation

dcaballe
Copy link
Contributor

@dcaballe dcaballe commented Jul 14, 2023

This PR should mostly be mostly NFC for existing working cases. It simplifies the retrieval of the vector sizes by removing the getCanonicalVectorShape and only looking at the lowering config of the operation to be vectorized and no longer looking at the lowering config of another (root) operation. It also gives priority to the lowering config found in the operation to be vectorize and fall back to retrieving it from the IR when the lowering config is not found.

This PR should be mostly RFC for existing working cases. It simplifies
the retrieval of the vector sizes by removing the `getCanonicalVectorShape`
and only looking at the lowering config of the operation to be
vectorized and no longer looking at the lowering config of another (root)
operation. It also gives priority to the lowering config found in the operation
to be vectorize and fall back to retrieving it from the IR when the lowering
config is not found.
@dcaballe dcaballe requested a review from hanhanW July 14, 2023 05:06
@hanhanW hanhanW added benchmarks:x86_64 Run default x86_64 benchmarks benchmarks:comp-stats Run default compilation statistics benchmarks benchmarks:android-cpu Run default Android CPU benchmarks labels Jul 14, 2023
@benvanik
Copy link
Collaborator

Closing stale PR.

@benvanik benvanik closed this May 14, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
benchmarks:android-cpu Run default Android CPU benchmarks benchmarks:comp-stats Run default compilation statistics benchmarks benchmarks:x86_64 Run default x86_64 benchmarks
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants