mikemccand commented on issue #854: Shared PQ Based Early Termination for Concurrent Search URL: https://github.com/apache/lucene-solr/pull/854#issuecomment-530549495 I agree the "maximum minimum" score (for all local PQs that are full) is compelling as it'd require minimum synchronization. Maybe it performs best overall for most use cases. It's downside is that it's collecting N * M (N = requested topN, M = number of thread work units). When N is smallish (common case) maybe the tradeoff is worthwhile? The "minimum minimum" score (once N hits have been collected sum'd across all thread work units) is closer to optimal, since it only collects N before skipping/pruning, but requires more work to update the "minimum minimum" any time any thread collects a newly competitive hit. If it's a simple array that's an O(M) scan each time; if it's a PQ then O(log(M)).
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
