[concurrency-interest] Fwd: Some fork/join threads idle when workload is non-uniform

Arul Dhesiaseelan aruld at acm.org
Tue May 29 15:17:46 EDT 2012


Forgot to copy the list.

---------- Forwarded message ----------
From: Arul Dhesiaseelan <aruld at acm.org>
Date: Mon, May 28, 2012 at 12:37 PM
Subject: Re: [concurrency-interest] Some fork/join threads idle when
workload is non-uniform
To: Doug Lea <dl at cs.oswego.edu>


Out of curiosity, I ran the tests with jdk7 and noticed
"ForkJoinPool-1-worker-1" thread was waiting in the non-uniform workload
mode (as reported by Peter).

With jdk8 (build 1.8.0-ea-b40) on Mac OS X 10.7.4, I noticed an interesting
behavior though. Both tests were running fine until about 55 mins, the
ForkJoinWithNonUniformWorkLoad JVM was shutdown automatically while
ForkJoinWithUniformWorkLoad JVM was cruising along without any problems.
Any clues what might be happening here?

-Arul


On Sat, May 26, 2012 at 2:01 AM, Doug Lea <dl at cs.oswego.edu> wrote:

> On 05/26/12 04:55, Peter De Maeyer wrote:
>
>> We have a multithreaded application which relies heavily on the F/J
>> framework.
>> When threadprofiling the application we've observed that not all F/J
>> threads are
>> kept busy. Some threads spend a lot of time waiting, thus making
>> suboptimal use
>> of the available CPUs in the system.
>>
>
> As far as I can tell by running your example code, the underlying issues
> have been addressed for upcoming jdk8 version, as well as in the
> standalone jsr166y version that you can use/run now by grabbing
> jar file and changing imports from "java.util.concurrent" to "jsr166y".
> (See http://gee.cs.oswego.edu/dl/**concurrency-interest/index.**html<http://gee.cs.oswego.edu/dl/concurrency-interest/index.html>
> )
> Please try this and let me know if not.
>
> The main underlying issue is that thread compensation was overly
> conservative, which avoided unbounded spare thread construction
> in the worst cases of misuse, but at the expense of overly limiting
> parallelism in other cases.
>
> -Doug
>
> ______________________________**_________________
> Concurrency-interest mailing list
> Concurrency-interest at cs.**oswego.edu <Concurrency-interest at cs.oswego.edu>
> http://cs.oswego.edu/mailman/**listinfo/concurrency-interest<http://cs.oswego.edu/mailman/listinfo/concurrency-interest>
>
<http://twitter.com/aruld>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.oswego.edu/pipermail/concurrency-interest/attachments/20120529/806fa34c/attachment.html>


More information about the Concurrency-interest mailing list