Fs benchmarks take 10+ hours #49689

Activity

1. anonrig

   added

   fsIssues and PRs related to file-system APIs and the fs module.Issues and PRs related to file-system APIs and the fs module.

   benchmarkIssues and PRs related to Node.js benchmarks and benchmarking infrastructure.Issues and PRs related to Node.js benchmarks and benchmarking infrastructure.

   on Sep 17, 2023
2. joyeecheung commented on Sep 17, 2023

   

   joyeecheung

   on Sep 17, 2023

   Last edited by joyeecheung

   Member

   More actions

   I think the fs benchmarks are having the same kind of issue that happen to many other benchmarks - they have a bit too many combinations of the parameters, and some of them are unnecessary. For example there are usually too many size / len variations. For example in this.

   node/benchmark/fs/readfile-promises.js

   Lines 17 to 24 in cdcb01a

   	len: [
   	1024,
   	512 * 1024,
   	4 * 1024 ** 2,
   	8 * 1024 ** 2,
   	16 * 1024 ** 2,
   	32 * 1024 ** 2,
   	],

   We probably just need 2-3 combinations by default. Or at least only pick 2-3 combinations when we are running the entire fs benchmark. There could be something OS-specific about them to make certain options not representative enough to be in that default set but I think we can always expand the variations when there's actually any doubt.

   👍React with 👍2Reacted by Yagiz Nizipli and Vinicius Lourenço
3. mscdex commented on Sep 18, 2023

   

   mscdex

   on Sep 18, 2023

   Contributor

   More actions

   I think the fs benchmarks are having the same kind of issue that happen to many other benchmarks - they have a bit too many combinations of the parameters, and some of them are unnecessary.

   This would probably be solved by #39285
4. RafaelGSS commented on Sep 18, 2023

   

   RafaelGSS

   on Sep 18, 2023

   Member

   More actions

   Also related nodejs/build#3483.
5. tniessen commented on Sep 18, 2023

   

   tniessen

   on Sep 18, 2023

   Member

   More actions

   We could also make benchmark iterations scalable. Running a few benchmarks for 1e5 iterations is usually not a problem, but running all benchmarks for 1e5 iterations might be.

   👍React with 👍2Reacted by Yagiz Nizipli and Erick Wendel
6. Uzlopak commented on Sep 19, 2023

   

   Uzlopak

   on Sep 19, 2023

   Contributor

   More actions

   I usually start with 1e7 iterations, and if it takes more time than I can bear, I reduce it by a factor of 10 and check if the benched result is not significantly different. Rinse and repeat.
7. H4ad commented on Sep 28, 2023

   

   H4ad

   on Sep 28, 2023

   Member

   More actions

   Is possible to create some kind of repo/issue to track slow benchmarks? Like we have for flaky tests?
8. github-actions commented on May 28, 2026

   

   github-actionsbot

   on May 28, 2026 – with GitHub Actions

   Contributor

   More actions

   This issue has been marked as stale due to 210 days of inactivity.
   It will be automatically closed in 30 days if no further activity occurs. If this is still relevant, please leave a comment or update it to keep it open.
9. github-actions

   added

   staleIssues and PRs marked stale due to inactivity and scheduled for automatic closure.Issues and PRs marked stale due to inactivity and scheduled for automatic closure.

   on May 28, 2026
10. github-actions commented on Jun 28, 2026

    

    github-actionsbot

    on Jun 28, 2026 – with GitHub Actions

    Contributor

    More actions

    This issue has been automatically closed after 30 days of inactivity following its stale status (no activity for a total of 240 days).
    If this is still relevant, feel free to reopen it or leave a comment with additional details so we can continue the discussion.
11. github-actions

    closed this as not plannedon Jun 28, 2026