RFR: 8290531: Loom: Parallelize a few tests more deeply

Jaikiran Pai jpai at openjdk.org
Tue Jul 26 10:18:49 UTC 2022


On Tue, 26 Jul 2022 10:06:31 GMT, Jaikiran Pai <jpai at openjdk.org> wrote:

>> Some of the newly added Loom tests are long-running, and thus they delay the completion of otherwise very parallel tier1/jdk_loom. We can parallelize them a bit better to avoid these testing stalls. 
>> 
>> Improvements on Linux x86_64, TR 3970X, `jdk_loom hotspot_loom`:
>> 
>> 
>> # release before
>> real 4m41.424s
>> user 24m18.064s
>> sys 1m16.440s
>> 
>> # release after
>> real 2m47.769s   ; -40%
>> user 23m44.622s
>> sys 1m15.240s
>> 
>> 
>> # fastdebug before
>> real 5m38.078s
>> user 67m23.516s
>> sys 1m56.446s
>> 
>> # fastdebug after
>> real 4m9.249s    ; -26%
>> user 67m21.940s
>> sys 1m57.625s
>
> test/jdk/java/nio/channels/vthread/BlockingChannelOps.java line 36:
> 
>> 34:  * @test id=useDirectRegister
>> 35:  * @bug 8284161
>> 36:  * @summary Basic tests of virtual threads doing blocking I/O with NIO channels
> 
> Hello Aleksey, I see that previously the `@bug` and `@summary` was only there on one test definition. So maybe remove this duplicated `@bug` and `@summary` from this new test definition?
> Other than that, this test change looks OK to me.

On a related note - I had a brief look at the jtreg docs including the FAQ https://openjdk.org/jtreg/faq.html and I don't see any mention about how it treats multiple test definitions in a single source file, when it comes to running them concurrently. But I believe you have more experience with it and have already tested the difference in parallelism, so I think what you suggest here is fine.

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

PR: https://git.openjdk.org/jdk/pull/9554


More information about the core-libs-dev mailing list