gianm commented on issue #9321: Performance degradation in topN queries when SQL-compatible null handling is enabled URL: https://github.com/apache/druid/issues/9321#issuecomment-591671621 It could be that something about calling groupBy defeats de-virtualization or inlining as described on https://shipilev.net/blog/2015/black-magic-method-dispatch/. Maybe something that was monomorphic or bimorphic with only topNs in the mix becomes megamorphic once groupBys are in the mix as well (due to different usage patterns of some shared component). I don't think I can speculate about what component that might be, since there are a lot of shared components and some of them are pretty complex. It might not even have anything to do with groupBy — it's possible that your groupBy queries are using some features (specific aggregators, filters, etc) that your topNs aren't using. But if this is what's going on, then vectorization should fix it, since it makes function call overhead less of an issue. So we should continue with our program of vectorizing everything (topN isn't yet).
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
