[
https://issues.apache.org/jira/browse/LUCENE-10404?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17569237#comment-17569237
]
Julie Tibshirani commented on LUCENE-10404:
-------------------------------------------
As a note, LUCENE-10592 changes the index strategy so that we build the graph
as each document is added, instead of waiting until 'flush'. In the PR, graph
building still shares a single FixedBitSet to track the 'visited' set, but it's
continuously resized since we don't know the full number of docs up-front. So
maybe switching to a hash set could help even more after that change is merged.
> Use hash set for visited nodes in HNSW search?
> ----------------------------------------------
>
> Key: LUCENE-10404
> URL: https://issues.apache.org/jira/browse/LUCENE-10404
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Julie Tibshirani
> Priority: Minor
>
> While searching each layer, HNSW tracks the nodes it has already visited
> using a BitSet. We could look into using something like IntHashSet instead. I
> tried out the idea quickly by switching to IntIntHashMap (which has already
> been copied from hppc) and saw an improvement in index performance.
> *Baseline:* 760896 msec to write vectors
> *Using IntIntHashMap:* 733017 msec to write vectors
> I noticed search performance actually got a little bit worse with the change
> -- that is something to look into.
> For background, it's good to be aware that HNSW can visit a lot of nodes. For
> example, on the glove-100-angular dataset with ~1.2 million docs, HNSW search
> visits ~1000 - 15,000 docs depending on the recall. This number can increase
> when searching with deleted docs, especially if you hit a "pathological" case
> where the deleted docs happen to be closest to the query vector.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
