RFR: 8322732: ForkJoinPool may underutilize cores in async mode [v3]

Viktor Klang vklang at openjdk.org
Wed May 22 15:59:07 UTC 2024


On Wed, 22 May 2024 15:32:42 GMT, Doug Lea <dl at openjdk.org> wrote:

>> This set of changes address causes of poor utilization with small numbers of cores due to overly aggressive contention avoidance. A number of further adjustments were needed to still avoid most contention effects in deployments with large numbers of cores
>
> Doug Lea has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains 36 additional commits since the last revision:
> 
>  - Merge branch 'openjdk:master' into JDK-8322732
>  - More performance tradoffs
>  - Address review comments
>  - Merge branch 'openjdk:master' into JDK-8322732
>  - Repack some fields; adjust control flow
>  - Merge branch 'openjdk:master' into JDK-8322732
>  - Next version
>  - Merge branch 'openjdk:master' into JDK-8322732
>  - Reduce unneeded signals
>  - Merge branch 'openjdk:master' into JDK-8322732
>  - ... and 26 more: https://git.openjdk.org/jdk/compare/bdb1c0d3...f1fc4f3e

src/java.base/share/classes/java/util/concurrent/ForkJoinPool.java line 2053:

> 2051:                 WorkQueue[] qs;
> 2052:                 int spins = ((short)(qc >>> TC_SHIFT) << 1) + SPIN_WAITS + 1;
> 2053:                 while (((p = w.phase) & IDLE) != 0 && --spins > 0)

Nice catch on the spins :)

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/19131#discussion_r1610252517


More information about the core-libs-dev mailing list