[
https://issues.apache.org/jira/browse/LUCENE-9845?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17302169#comment-17302169
]
Robert Muir commented on LUCENE-9845:
-------------------------------------
ok I see. Maybe we can take a step back and first detect "dense" case similar
to what docvalues does, and reduce memory/cpu overheads for that case, worrying
about sparse later? I imagine in a typical use-case the user would have values
for every doc, so it is good to optimize for?
> Improve encoding of HNSW graph offsets
> --------------------------------------
>
> Key: LUCENE-9845
> URL: https://issues.apache.org/jira/browse/LUCENE-9845
> Project: Lucene - Core
> Issue Type: Wish
> Reporter: Michael Sokolov
> Priority: Major
>
> Today we use a simple {{long[]}} to encode the offset of each document's
> array of neighbors in the HNSW graph. Instead we should use a data structure
> that is optimized for encoding an increasing numeric array, like monotonic
> {{PackedInts}}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
