adriangb commented on PR #24547: URL: https://github.com/apache/datafusion/pull/24547#issuecomment-5378360164
> Any chance we could note why these numbers were picked? Not critical, but might help future readers. LGTM either way! > > spilled_bytes > logical_bytes / 4, spilled_bytes < logical_bytes * 3, Thanks for asking. They were arbitrary, and measuring showed the 3x bound could never actually fire (with gc_view_arrays disabled the sort spill files only inflate to 1.73x at batch_size=1024). I retuned the sort test to batch_size=256 / 8 MB pool where compacted measures 1.0152x against 9.3384x regressed, tightened the bound to 2x, and recorded those measurements inline. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
