gf2121 commented on PR #14365: URL: https://github.com/apache/lucene/pull/14365#issuecomment-2735314110
Thanks for running benchmark, the speed up is great! > Skipping these doc IDs looks like it hurts vectorization, I played with disabling these if statements locally and get a good speedup on queries sorted by numeric field. I try it after seeing your comment and see similar speed up. I think there might be more reason contributing to the speedup: * We are collecting more docs into the `DocIdSetBuilder` and the competitiveIterator has more chance to be a bitset iterator, which reduces the overhead of `LSBRadixSorter` and speed up `competitiveIterator#intoBitset`. This seems more like a performance-memory trade off for me. * The DayOfYear field has medium cardinality (365) so some blocks stored as bitset, bulk adding `DocBaseBitsetIterator` might also help the performance. -- 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: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
