[Mpi3-hybridpm] thread persistence

[Snir Marc]

Thanks for the observation. Yes, it would be nice to maintain  
persistence in inner parallel constructs -- but I do not think this is  
essential for the model(s) I propose. In a static, programming to the  
iron, model, I suspect that users will not use nested parallelism, but  
will simply start a thread on each core, and go from there. On the  
other hand, in the dynamic model, you make no assumptions on task to  
thread association.

>
> Message: 2
> Date: Thu, 6 Aug 2009 10:33:41 -0700 (PDT)
> From: "Bronis R. de Supinski" <bronis at llnl.gov>
> Subject: Re: [Mpi3-hybridpm] Mpi3-hybridpm Digest, Vol 5, Issue 4
> To: "mpi3-hybridpm at lists.mpi-forum.org"
>        <mpi3-hybridpm at lists.mpi-forum.org>
> Cc: OpenMP Language Committee <omp-lang at openmp.org>
> Message-ID: <Pine.LNX.4.58.0908061030590.11162 at tux213.llnl.gov>
> Content-Type: TEXT/PLAIN; charset=windows-1252
>
>
> Marc:
>
> It is true that you can force persistence of thread
> numbering (and threadprivate data) for the first level.
> However, if you use OpenMP nested parallelism then the
> numbering of threads in inner regions (and threadprivate
> data) is not guaranteed to persist. It seems like you
> require that in some cases (and this topic is one that
> the OpenMP language committee has discussed although
> it has not been a major point recently).
>
> Bronis
>

Marc Snir
4323 Siebel Center, 201 N Goodwin, IL 61801
Tel (217) 244 6568
Web http://www.cs.uiuc.edu/homes/snir




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-hybridpm/attachments/20090807/910a7bc5/attachment.html>