Skip to content

Use RustFFT more efficiently in benchmarks for a fair comparison#81

Draft
Shnatsel wants to merge 1 commit intomainfrom
rustfft-optimization
Draft

Use RustFFT more efficiently in benchmarks for a fair comparison#81
Shnatsel wants to merge 1 commit intomainfrom
rustfft-optimization

Conversation

@Shnatsel
Copy link
Collaborator

No description provided.

@smu160
Copy link
Member

smu160 commented Feb 22, 2026

LGTM

@codecov-commenter
Copy link

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.79%. Comparing base (d8b83d6) to head (a4dd686).

Additional details and impacted files
@@           Coverage Diff           @@
##             main      #81   +/-   ##
=======================================
  Coverage   99.79%   99.79%           
=======================================
  Files          13       13           
  Lines        2872     2872           
=======================================
  Hits         2866     2866           
  Misses          6        6           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

@Shnatsel
Copy link
Collaborator Author

Shnatsel commented Feb 22, 2026

I am quite confused by the benchmarks: on x86, everything up to and including Forward f32/RustFFT/2097152 regresses from this change, while larger sizes (Forward f32/RustFFT/4194304 and above) improve dramatically. And also f64 small size performance improves? What?!

@Shnatsel Shnatsel marked this pull request as draft February 22, 2026 21:07
@Shnatsel
Copy link
Collaborator Author

Zen4 benchmark results: https://gist.github.com/Shnatsel/a0b00f7e2a3b4eab74293b8996c94821

The f32 regressions are consistent between runs, it's not a transient measurement artifact

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants