msokolov commented on PR #14178: URL: https://github.com/apache/lucene/pull/14178#issuecomment-2855715698
>> but then multi-segment is only a bit slower (not proportionally so) > I think this is because Lucene can take advantage of a shared queue of results, and stop searching in segments where results are low scoring, because it has access to the collector during graph traversal -- as opposed to Faiss where we call the native library to give us K results per-segment, irrespective I think as of yesterday we are no longer using `MultiLeafKnnCollector` in the default case, but have switched to using "optimistic" collection based on pro-rating. Not sure which version Mike is testing there. -- 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
