On Wed, 28 Oct 2020 21:59:53 GMT, Jason Tatton <github.com+70893615+jasontatton-...@openjdk.org> wrote:
> Please review the improvements which I have made to the > java/lang/StringIndexOfChar.java jmh benchmark. Please let me know if any > further improvements are required. > > Thanks, > Jason Looks OK. FWIW the current looping scheme look like what the [SafeLooping.measureWrong_2](https://github.com/openjdk/jmh/blob/620036ec27db3503fa2dee476036b3219c9a38bc/jmh-samples/src/main/java/org/openjdk/jmh/samples/JMHSample_34_SafeLooping.java#L115) warns us about, and could possibly be improved by taking cues from [SafeLooping.measureRight_1](https://github.com/openjdk/jmh/blob/620036ec27db3503fa2dee476036b3219c9a38bc/jmh-samples/src/main/java/org/openjdk/jmh/samples/JMHSample_34_SafeLooping.java#L136) and sink each `what.indexOf(..)` into a `BlackHole`. This might matter here since each unit of work is quite small. I'll approve the patch as-is, but please consider making those changes. test/micro/org/openjdk/bench/java/lang/StringIndexOfChar.java line 63: > 61: private String [] utf16_sse4; > 62: private String [] utf16_avx2; > 63: private String [] utf16_mixedLength; I'd be a little bit happier with `String[]` rather than `String []` ------------- Marked as reviewed by redestad (Reviewer). PR: https://git.openjdk.java.net/jdk/pull/918
