vsop-479 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2774626745
@gf2121 , I measured it on a linux server (`uses preferredBitSize=512; FMA
enabled`), there is still a massive slowndown. I will dig more ...
```
Benchmark
vsop-479 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2768597044
Adjust `ENABLE_ADVANCE_WITHIN_BLOCK_VECTOR_OPTO` to 16 (at least 16 lanes,
such as: AVX, AVX-512).
--
This is an automated message from the Apache Git Service.
To respond to the messa
vsop-479 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2763043900
@gf2121
I also implemented `advanceExact` with vector, there is still a slowdown. I
will try to measure it on other laptop (with more vector lanes).
Benchmark
vsop-479 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2760720322
@gf2121
I implemented `VectorMask` approach. There is still a slowdown. I think the
reason is my laptop (Mac M2).
Benchmark Mode Cnt
vsop-479 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2753155562
> a bench in jmh will be great.
I measured it with `AdvanceSparseDISIBenchmark`:
Benchmark Mode CntScore
Error
gf2121 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2747159521
Thanks for running benchmark @vsop-479 !
> Maybe I should measure it with DVBench in luceneutil, or add a bench in
jmh.
Yes, you are right, a bench in jmh will be great. We h
vsop-479 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2747066080
Maybe I should measure it with `DVBench` in luceneutil, or add a bench in
jmh.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
vsop-479 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2747056694
@gf2121
For what it's worth, I implemented this patch, and measured with luceneutil
on `wikimedium10m`.
TaskQPS baseline StdDevQPS my_modified_version
vsop-479 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2740387593
Thanks for your feedback @gf2121. This patch is still in process, and have
not been measured.
> If you look at newest code in Lucene101PostingsReader, you may find we are
using
gf2121 commented on PR #14371:
URL: https://github.com/apache/lucene/pull/14371#issuecomment-2739828325
Thanks @vsop-479 , have you been able to measure the performance of your
patch?
I had similar idea recently. If you look at newest code in
`Lucene101PostingsReader`, you may find w
vsop-479 opened a new pull request, #14371:
URL: https://github.com/apache/lucene/pull/14371
### Description
Similar to https://github.com/apache/lucene/pull/13692.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub an
11 matches
Mail list logo
