SEZ9 commented on issue #12063: URL: https://github.com/apache/seatunnel/issues/12063#issuecomment-5650668386
@Rangsh thanks for moving the Score / Error / CV evidence to the normal **Benchmarks** workflow (`benchmarks.yml`) and for flagging the CPU caveat yourself. With `AMD EPYC 9V74` before vs `AMD EPYC 7763` after, I agree the cross-run Score deltas in the Java 11 table can't be read as before/after, and the CV movement is only weakly informative. Treating the fork runs as observational and Diagnostics as causal-only is the right split. A few follow-ups on this point: 1. **Same-worker run.** Since you're blocked by the 403, @nzw921rx could you (or another committer) dispatch **Benchmarks** on `apache/seatunnel` with `pr_number=12173` (`baseline -> PR -> PR -> baseline`)? @Rangsh, once that finishes, please post the Score / Error / CV table for `IMapJobStorageBenchmark.(runningJobGrowth|completedJobHistoryGrowth)$` at `initialStoredJobCount=0` and `1000` from that run, so the comparison comes from one host. 2. **Commit clarification.** Your earlier comment lists the after run https://github.com/Rangsh/seatunnel/actions/runs/34354315647 as PR head `909f878`, while the table says PR @ `c0d59da`. Could you confirm which commit that run actually built, and make sure the head used for the same-worker run matches what's currently on the PR? 3. **`runningJobGrowth` at `initialStoredJobCount=1000`.** 19.13% CV on the after host is still high. If it stays in that band on the same-worker run, please note whether you plan more forks/iterations or whether that fixture is inherently noisy at that count, so we know how to read it. <!-- streview-comment:1011 --> -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
