You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@lucene.apache.org by "msokolov (via GitHub)" <gi...@apache.org> on 2025/05/16 20:35:28 UTC
[PR] Use per-segment K in filtered KNN fallback logic (fixes 14671) [lucene]
msokolov opened a new pull request, #14680:
URL: https://github.com/apache/lucene/pull/14680
Originally (before optimistic KNN query):
```
recall latency(ms) nDoc topK fanout maxConn beamWidth quantized visited selectivity filterType vec_disk(MB) vec_RAM(MB) indexType
0.693 7.374 1000000 100 50 16 50 no 3753 0.05 pre-filter 0.000 0.000 HNSW
0.705 7.752 1000000 100 100 16 50 no 4680 0.05 pre-filter 0.000 0.000 HNSW
0.764 13.086 1000000 100 250 16 50 no 6307 0.05 pre-filter 0.000 0.000 HNSW
```
Mainline (filtering broken):
```
recall latency(ms) nDoc topK fanout maxConn beamWidth quantized index(s) index_docs/s num_segments index_size(MB) vec_disk(MB) vec_RAM(MB) indexType
0.894 31.510 1000000 100 50 16 50 no 0.00 Infinity 0 0.00 0.000 0.000 HNSW
1.000 46.757 1000000 100 100 16 50 no 0.00 Infinity 0 0.00 0.000 0.000 HNSW
1.000 52.029 1000000 100 250 16 50 no 0.00 Infinity 0 0.00 0.000 0.000 HNSW
```
With this fix:
```
recall latency(ms) nDoc topK fanout maxConn beamWidth quantized visited selectivity filterType vec_disk(MB) vec_RAM(MB) indexType
0.734 9.018 1000000 100 50 16 50 no 8951 0.05 pre-filter 0.000 0.000 HNSW
0.742 10.259 1000000 100 100 16 50 no 9857 0.05 pre-filter 0.000 0.000 HNSW
0.771 10.981 1000000 100 250 16 50 no 12117 0.05 pre-filter 0.000 0.000 HNSW
```
--
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-unsubscribe@lucene.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For additional commands, e-mail: issues-help@lucene.apache.org
Re: [PR] Use per-segment K in filtered KNN fallback logic (fixes 14671) [lucene]
Posted by "msokolov (via GitHub)" <gi...@apache.org>.
msokolov merged PR #14680:
URL: https://github.com/apache/lucene/pull/14680
--
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-unsubscribe@lucene.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For additional commands, e-mail: issues-help@lucene.apache.org
Re: [PR] Use per-segment K in filtered KNN fallback logic (fixes 14671) [lucene]
Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] commented on PR #14680:
URL: https://github.com/apache/lucene/pull/14680#issuecomment-2887632883
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one. If the PR doesn't need a changelog entry, then add the skip-changelog-check label to it and you will stop receiving this reminder on future updates to the PR.
--
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-unsubscribe@lucene.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For additional commands, e-mail: issues-help@lucene.apache.org