Accéder au contenu.
Menu Sympa

starpu-devel - Re: [Starpu-devel] Concurrent CUDA kernel execution

Objet : Developers list for StarPU

Archives de la liste

Re: [Starpu-devel] Concurrent CUDA kernel execution


Chronologique Discussions 
  • From: Samuel Thibault <samuel.thibault@ens-lyon.org>
  • To: Keisuke Fukuda <fukuda@matsulab.is.titech.ac.jp>
  • Cc: starpu-devel@lists.gforge.inria.fr
  • Subject: Re: [Starpu-devel] Concurrent CUDA kernel execution
  • Date: Wed, 9 Apr 2014 09:17:54 +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>

Keisuke Fukuda, le Wed 09 Apr 2014 12:55:27 +0900, a écrit :
> I believe this feature is very helpful for many applications that have
> trade-off issues of task size and load balance.

Yes!

> Are you going to include it in the next release, or will it be experimental
> for
> a while?

It will be in 1.2. It is working as expected in terms of giving the
correct results and overlapping kernels as expected. Scheduling things
properly so it gives good performance is another question, which we can
hope to have a look at during the 1.2 RC period :)

Samuel





Archives gérées par MHonArc 2.6.19+.

Haut de le page