Re: [PR] Implement off-heap quantized scoring [lucene]

2025-07-15 Thread via GitHub
kaivalnp commented on PR #14863: URL: https://github.com/apache/lucene/pull/14863#issuecomment-3074451138 After [this conversation](https://github.com/apache/lucene/pull/14874#issuecomment-3071054127) I re-ran some benchmarks with `-XX:CompileCommand=inline,*PanamaVectorUtilSupport.*` to f

Re: [PR] Implement off-heap quantized scoring [lucene]

2025-07-14 Thread via GitHub
github-actions[bot] commented on PR #14863: URL: https://github.com/apache/lucene/pull/14863#issuecomment-3071448271 This PR has not had activity in the past 2 weeks, labeling it as stale. If the PR is waiting for review, notify the d...@lucene.apache.org list. Thank you for your contributi

Re: [PR] Implement off-heap quantized scoring [lucene]

2025-06-30 Thread via GitHub
kaivalnp commented on PR #14863: URL: https://github.com/apache/lucene/pull/14863#issuecomment-3019782866 Thanks @ChrisHegarty! I saw that we [use](https://github.com/apache/lucene/blob/f4339ee2aea65bdf1efb2a3c196c3e9a4adf9d67/lucene/core/src/java24/org/apache/lucene/internal/vectorization/L

Re: [PR] Implement off-heap quantized scoring [lucene]

2025-06-30 Thread via GitHub
ChrisHegarty commented on PR #14863: URL: https://github.com/apache/lucene/pull/14863#issuecomment-3019297586 > ..then we see a performance regression: > ... > Maybe I'm missing something obvious, but I haven't found the root cause yet.. yeah. I've seen similar before. You might

Re: [PR] Implement off-heap quantized scoring [lucene]

2025-06-29 Thread via GitHub
kaivalnp commented on PR #14863: URL: https://github.com/apache/lucene/pull/14863#issuecomment-3016581312 FYI I observed a strange phenomenon where if the query vector is on heap like: ```java this.query = MemorySegment.ofArray(targetBytes); ``` instead of the current off-hea

Re: [PR] Implement off-heap quantized scoring [lucene]

2025-06-29 Thread via GitHub
kaivalnp commented on PR #14863: URL: https://github.com/apache/lucene/pull/14863#issuecomment-3016573861 I ran some benchmarks on Cohere vectors (768d) for 7-bit and 4-bit (compressed) quantization.. `main` without `jdk.incubator.vector`: ``` recall latency(ms) netCPU avgCpu

Re: [PR] Implement off-heap quantized scoring [lucene]

2025-06-29 Thread via GitHub
github-actions[bot] commented on PR #14863: URL: https://github.com/apache/lucene/pull/14863#issuecomment-3016556401 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 label to it and you will stop

[PR] Implement off-heap quantized scoring [lucene]

2025-06-29 Thread via GitHub
kaivalnp opened a new pull request, #14863: URL: https://github.com/apache/lucene/pull/14863 ### Description Off-heap scoring for quantized vectors! Related to #13515 This scorer is in-line with [`Lucene99MemorySegmentFlatVectorsScorer`](https://github.com/apache/lucene/blob/7