On Thu, 8 Feb 2024 11:12:29 GMT, Alexey Ivanov <aiva...@openjdk.org> wrote:
> I ran the test in our CI and I didn't notice an improvement. On the other > hand, having separate test ids is more flexible. I have retested by the complete jdk_desktop group of tests: Base: real 6m57.901s user 123m45.204s sys 9m16.735s Fix: real 4m23.600s user 126m35.078s sys 8m50.104s So it might depend on the system and used concurrency. ------------- PR Comment: https://git.openjdk.org/jdk/pull/17719#issuecomment-1935234631