Thanks Kevin. I'll move the discussion to dev mailing list On Wed, Apr 2, 2025, 4:18 PM Kevin Liang (BLOOMBERG/ 919 3RD A) < klian...@bloomberg.net> wrote:
> I don't have enough knowledge to give the greenlight on this, but on a > high level it seems reasonable. This question is probably more appropriate > to surface on the dev mailing list rather than the users mailing list. > Hopefully you get more thoughts on your patch suggestion there > > -Kevin > > From: users@solr.apache.org At: 03/25/25 20:24:32 UTC-4:00To: > users@solr.apache.org > Cc: chirayu.sama...@bloomreach.com > Subject: Performance issue: distributed grouping + dense vector search > > Hi All, > > When running vector search with grouping in multi-shard setting, the > KnnFloatVectorQuery is executed (2+rows) times. > > The culprit is this function call in > > https://github.com/apache/solr/blob/main/solr/core/src/java/org/apache/solr/sear > ch/grouping/distributed/command/TopGroupsFieldCommand.java#L202 > <https://github.com/apache/solr/blob/main/solr/core/src/java/org/apache/solr/search/grouping/distributed/command/TopGroupsFieldCommand.java#L202> > > if (needScores) { > for (GroupDocs<?> group : topGroups.groups) { > TopFieldCollector.populateScores(group.scoreDocs, searcher, query); > } > } > > Where > > https://github.com/apache/lucene/blob/main/lucene/core/src/java/org/apache/lucen > e/search/TopFieldCollector.java#L403 > <https://github.com/apache/lucene/blob/main/lucene/core/src/java/org/apache/lucene/search/TopFieldCollector.java#L403> > does > > final Weight weight = searcher.createWeight(searcher.rewrite(query), > ScoreMode.COMPLETE, 1); > > Here, if the query is KnnFloatVectorQuery, *searcher.rewrite(query) *will > execute the same vector search for each topGroups.groups > > a simple fix could be moving *searcher.rewrite(query) *out of the > topGroups.groups loop. Thoughts? > > Best, > > Yue > > >
