RFR: 8305670: Performance regression in LockSupport.unpark with lots of idle threads [v3]

Daniel D. Daugherty dcubed at openjdk.org
Thu Apr 27 16:34:53 UTC 2023


On Fri, 21 Apr 2023 04:15:06 GMT, David Holmes <dholmes at openjdk.org> wrote:

>> Daniel D. Daugherty has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Use optimized logic in Unsafe_Unpark() instead of cv_internal_thread_to_JavaThread().
>
> Unpark changes look good. :) Thanks.

@dholmes-ora - Thanks for the re-review. I have more changes to make and
I need to checkout the performance testing that Eric Caspole has done with
his new JMH. His JMH reproduces the degradation from JDK8u -> JDK21
with specific linux-aarch64 machines and shows that this patch restores the
performance in JDK21. I need to double check what the testing on other
platforms shows.

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

PR Comment: https://git.openjdk.org/jdk/pull/13519#issuecomment-1525992980


More information about the hotspot-runtime-dev mailing list