jpountz commented on issue #14281: URL: https://github.com/apache/lucene/issues/14281#issuecomment-2678895153
This is interesting, the linked change is expected to replace some potentially expensive calls to madvise (since they need to iterate all pages) with cheaper checks, so I wouldn't have expected a regression, unless the change somehow causes the read advice to not be applied correctly in some cases. I'm also curious why Elasticsearch would exhibit such a big regression when Lucene vector indexing benchmarks didn't observe a slowdown. https://benchmarks.mikemccandless.com/indexing.html#MedVectorsIndexTime Does Elasticsearch override the way how read advices get applied, by tweaking `org.apache.lucene.store.defaultReadAdvice` or overridding `Directory#openInput` in its custom directories? -- 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
