On Thu, 31 Aug 2023 08:40:58 GMT, Aleksey Shipilev <sh...@openjdk.org> wrote:

> Current vmTestbase/nsk/monitoring/stress/thread tests contains 21 tests, each 
> running exclusively. This drags the tier4 test times up. There seem to be no 
> reason to run these tests exclusively, though: they complete in reasonable 
> time, are lightly-threaded, and consume the usual amount of memory.
> 
> We should consider enabling parallelism for them and get improved test 
> performance. Currently it is blocked by TEST.properties with 
> exclusiveAccess.dirs directives in them.
> 
> Current run on 18-core machine:
>  1025.63s user 97.23s system 586% cpu 3:11.60 total
> 
> Fully parallel:
>  1125.04s user 107.60s system 1583% cpu 1:17.83 total 
> 
> Additional testing:
>  - [x]  50x iterations of Linux x86_64 fastdebug 
> `vmTestbase/nsk/monitoring/stress/thread`

I think this is fine, but it may inconvenience some test maintainers since we 
may now run against  /proc/sys/kernel/threads-max or whatever other limits 
there are. Or do we do worse things already in other tests?

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

Marked as reviewed by stuefe (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/15508#pullrequestreview-1604843845

Reply via email to