mikemccand commented on PR #13190: URL: https://github.com/apache/lucene/pull/13190#issuecomment-2011765808
> Thanks for pushing on this change, I like it. The fact that the extra merge concurrency may not starve merging from threads is good. I'm curious how the nightly benchmarks will react to it, given the high number of CPU cores of beast3. Nightly benchy is still forcing its own thread pool into the HNSW indexing Codec component (`Lucene99Hnsw*VectorsFormat`) -- once this change is merged, I'll remove that so we switch to this awesome default approach. But maybe we won't see too much change in the nightly indexing throughput since it's already doing intra-merge concurrency "itself". > One question on my mind is whether this change should make us update the default number of merging threads that `ConcurrentMergeScheduler` configures (in a follow-up issue/PR). +1 to explore this. Nightly benchy hardwires the maxMergeCount=16, maxThreadCount=12. Maybe they should be higher :) But also note that the nightly benchy does not wait for merges on `IndexWriter.close`, so it's not really a fair test of merge 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