RFR: 8315936: Parallelize gc/stress/TestStressG1Humongous.java test

[Aleksey Shipilev]

On Wed, 27 Sep 2023 11:30:22 GMT, Ian Myers <duke at openjdk.org> wrote:

> TestStressG1Humongous runs in hotspot:tier4 and takes about 900 seconds to run. It is one of the slowest tests in tier4, and it runs when concurrency is generally low due to exclusive adjacent tests. Thus, it limits the effective parallelism of tier4.
> 
> In test main method, there are a few independent test cases, which can be expressed in jtreg to gain parallel execution.
> 
> TestStressG1Humongous current runtime: **4917.51s user 410.97s system 549% cpu 16:10.07 total**
> TestStressG1Humongous parallelized runtime: **4672.49s user 510.17s system 849% cpu 10:10.05 total**
> 
> This change takes the previous `run()` methods out of the main method and instead passes the parameters as part of the `@run` jtreg test annotation, allowing them to be run in parallel.
> 
> The last test runs the longest on its own, and this is why the runtime can't be reduced under the 10 minute mark.

All right, I think only one nit is left. Maybe @tschatzl would like to take a look as well.

test/hotspot/jtreg/gc/stress/TestStressG1Humongous.java line 26:

> 24: package gc.stress;
> 25: 
> 26:  /*

Here and later, this is misaligned. All stars should align, like before.

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

Marked as reviewed by shade (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/15943#pullrequestreview-1646496208
PR Review Comment: https://git.openjdk.org/jdk/pull/15943#discussion_r1338596192