LantaoJin commented on PR #16807: URL: https://github.com/apache/lucene/pull/16807#issuecomment-6121261627
The log scanning generates too many noise, the two cheap ways to get there: a luceneutil config that searches two fields with different quantization encodings in one JVM (as far as I can tell, each config runs in its own JVM with a single encoding today), and a `pollute` option for the byte kernels in `VectorUtilBenchmark`, like the one `VectorScorerFloat32Benchmark` already has, that warms the other variant during setup. The intrinsic and inlining logs are still very useful for working out *why* once a number looks off. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
