LantaoJin opened a new pull request, #16831: URL: https://github.com/apache/lucene/pull/16831
### Description JMH forks a new JVM for every benchmark, so VectorUtilBenchmark measures each byte kernel in a JVM where no other variant has run, and it did not show #16807. VectorUtilMixingBenchmark first runs every signed and unsigned dot product and square distance long enough for C2 to compile it, then measures each kernel on its own (pollute=all), and compares that with a fresh JVM (pollute=none). A combined method calls each kernel through its own DONT_INLINE helper, so that each kernel has its own compilation as behind the vector scorers, and the result does not depend on how many Vector API kernels C2 can inline into one benchmark method. -- 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]
