Objet : Developers list for StarPU
Archives de la liste
- From: Samuel Thibault <samuel.thibault@ens-lyon.org>
- To: Jeff Hand <jeffrey.hand@gmail.com>, starpu-devel@lists.gforge.inria.fr
- Subject: Re: [Starpu-devel] StarPU Library Instances
- Date: Tue, 14 Oct 2014 13:47:17 +0200
- List-archive: <http://lists.gforge.inria.fr/pipermail/starpu-devel/>
- List-id: "Developers list. For discussion of new features, code changes, etc." <starpu-devel.lists.gforge.inria.fr>
Hello,
I'm not sure to understand what kind of application you are running and
what its parallelism looks like.
Jeff Hand, le Thu 09 Oct 2014 16:39:15 -0500, a écrit :
> How do I sample multiple input sources, run it through dynamic number of
> computing blocks streams, and maintain different configurations for
> computing
> blocks with minimal disruptions to other streams?
What do you mean by disruption? Performance loss? Crash?
I also don't really understand the difference between your proposed
option 2 and 3.
> Simply have forked processes/threads per stream.
> - Likely disturbs calibration and task priorities for starpu.
Yes, but that may not be an issue. Schedulers often work quite well with
not so good calibration, as long as it's representative enough of the
performances.
Samuel
- [Starpu-devel] StarPU Library Instances, Jeff Hand, 11/10/2014
- Re: [Starpu-devel] StarPU Library Instances, Samuel Thibault, 14/10/2014
- Re: [Starpu-devel] StarPU Library Instances, Jeff Hand, 14/10/2014
- Re: [Starpu-devel] StarPU Library Instances, Samuel Thibault, 15/10/2014
- Re: [Starpu-devel] StarPU Library Instances, Jeff Hand, 14/10/2014
- Re: [Starpu-devel] StarPU Library Instances, Samuel Thibault, 14/10/2014
Archives gérées par MHonArc 2.6.19+.