Lucene: Status=Open

[Save this search]

Author relation

Updated ago

Comment count

Reaction count

Review Requested

User

Last comment user

Type

Filters: Status (Open), Issue type, Author relation, Created, Updated, Updated ago, Comment count, Reaction count, Review Requested, Mentioned, Reviewed, Commented, User, Last comment user, Draft, Component, Type, Labels, Commits?, Reporter, Assignee

#16693 PR: Prevent unintended multi-segment flushes in TestLucene104ScalarQuantizedVectorFormat (#16692)

1.9 hours ago 0 comments 0 votes 0 watches CH-Abhinav, github-actions[bot]

Resolves 16692 Description Follow-up to 16030: fixes test flakiness in TestLucene104ScalarQuanti ... Under randomized test runs (e.g., seed D9588E29B239F98E), IndexWriter could flush multiple segments ... Changes * Added w.forceMerge(1) in testDataBlindNoRawFloatVectors. * Set DISABLE_AUTO_FLUSH and ...

#16692: Flaky test: TestLucene104ScalarQuantizedVectorsFormat.testDataBlindNoRawFloatVectors fails when ...

1.9 hours ago 0 comments 0 votes 0 watches CH-Abhinav

Description TestLucene104ScalarQuantizedVectorsFormat.testDataBlindNoRawFloatVectors occasionally ... ContextIntroduced in 16030 ("Implement data-blind scalar quantization") where testDataBlindNoRawFloa ... `text java.lang.IllegalArgumentException: StandardDirectoryReader(segments_1:6 _0(11.0.0):c22 _1( ...

#16691 PR: Optimize DocValues bulk decoding using Panama widening casts (#16677)

2.7 hours ago 0 comments 0 votes 0 watches CH-Abhinav, github-actions[bot]

Description This PR introduces SIMD bulk decoding for DocValues (8-bit, 16-bit, and 32-bit ... Currently, default scalar loops handle these widening. This PR tries to use Panama Vector API's shape-widening operations (convertShape with ...

#16674: Can we speed up centroid calculation in the OSQ vector format?

3.7 hours ago 3 comments 0 votes 0 watches CH-Abhinav, kaivalnp, msokolov

Description In the OSQ based vector format, we recalculate centroids on merge when there are ... If there aren't any deletes, the merged centroid is simply a weighted average of segment-level ... I wonder if we can re-use the information of the previous centroid, even when there are deletes?

CH-Abhinav 3.9 hours ago:	@msokolov I think we should go with multiple unbiased samples, calculate their centroids and merge ... IDK if using one large sample is better than using multiple small samples.
CH-Abhinav 4.1 hours ago:	do you think doing unbiased samples will always work ? maybe a 1-5% sample would be good enough to calculate centroid for OSQ but that is completely ... I would say this approach will be more optimal :- 1. if delete's are N<1% for small segments and N ...

#16690 PR: Add a hybrid strategy for stored fields merge

7.5 hours ago 0 comments 0 votes 0 watches github-actions[bot], wwwa1ter

Closes 16045 # Description The stored fields merge currently falls back to a doc-by-doc copy ... This PR adds a hybrid merge strategy: contiguous ranges of live docs are collected, and - whole ... The hybrid strategy is enabled by default and can be disabled with the org.apache.lucene.codecs.luc ...

#16045: Optimize StoredFields merge strategy based on chunk deletion ratio

7.7 hours ago 1 comments 0 votes 0 watches kkewwei, wwwa1ter

Description For StoredField merging, the merge strategy has to switch from the BULK mode to the ...

wwwa1ter 7.7 hours ago:

Hi @kkewwei, PR #16690 implements the hybrid merge strategy from 16045. If you have a moment to review, that would be super helpful.Happy to make changes based on your ...

#16622 PR: Propagate getMergeInstance() through the quantized vector readers

10.3 hours ago 0 comments 0 votes 0 watches github-actions[bot], john-mlika

MergeState asks every reader for a merge instance when a merge starts, and Lucene99HnswVectorsReader ... For quantized fields the flat reader is Lucene104ScalarQuantizedVectorsReader (or, for older ... Same for finishMerge().

«
1
2
3
4
5
»

2,660 issues

[15.7 msec search, 16.7 msec total]