neilconway commented on PR #21240: URL: https://github.com/apache/datafusion/pull/21240#issuecomment-4164078933
Looking at the remaining performance regressions: * TPC-DS query 9 was slow because we weren't evaluating subqueries in parallel. That is fixed. * TPC-DS query 24 is slow (1.9x slower) because we don't overlap evaluating the main query with evaluating the subqueries. I prototyped a quick hack locally that overlaps the evaluation and fixes the performance issue (we get back to parity with main), but I need to understand how parallelism works inside DF better before I can figure out the right way to do this for real. * TPC-DS query 23 perf regression is also fixed by overlapping compute, same as in Q24. Still need to look into TPC-H Q22 at SF1. -- 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]
