rmuir commented on PR #11987: URL: https://github.com/apache/lucene/pull/11987#issuecomment-1332307323
Thanks for running the stored fields benchmark: are you able to report the retrieval time as well? That's my first concern. Maybe, the StoredFieldsBenchmark.java needs to be run standalone to report it, here is the relevant code: https://github.com/mikemccand/luceneutil/blob/master/src/main/perf/StoredFieldsBenchmark.java#L89-L101 My other concern would be if we create too much pressure on GC for unoptimized merges. The StoredFieldsBenchmark uses geonames and does not delete/update documents, so it would never exercise this path much. We could run the StoredFieldsBenchmark before and after the change with `-Dorg.apache.lucene.codecs.lucene90.compressing.Lucene90CompressingStoredFieldsWriter.enableBulkMerge=false` to force the slow merge path. In all cases when running the benchmark, we may want to explicitly supply smaller heap (-Xmx), since the dataset is not very big and otherwise jvm may allocate a huge heap, dodging any GC impacts that we want to see. Thank you again for benchmarking, if you run into trouble I can try to help run these benchmarks too. -- 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
