Accéder au contenu.
Menu Sympa

starpu-devel - Re: [Starpu-devel] [BUG] tests/datawizard/reclaim.c: hangs forever in task_wait_for_all when using OpenCL.

Objet : Developers list for StarPU

Archives de la liste

Re: [Starpu-devel] [BUG] tests/datawizard/reclaim.c: hangs forever in task_wait_for_all when using OpenCL.


Chronologique Discussions 
  • From: Cyril Roelandt <cyril.roelandt@inria.fr>
  • To: Samuel Thibault <samuel.thibault@ens-lyon.org>, starpu-devel@lists.gforge.inria.fr, ludovic.stordeur@inria.fr
  • Subject: Re: [Starpu-devel] [BUG] tests/datawizard/reclaim.c: hangs forever in task_wait_for_all when using OpenCL.
  • Date: Fri, 09 Mar 2012 13:43:31 +0100
  • 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>

On 03/09/2012 12:25 PM, Samuel Thibault wrote:
Cyril Roelandt, le Fri 09 Mar 2012 03:36:59 +0100, a écrit :
With CUDA, things start to take an awful lot of time :

$ STARPU_NCPUS=0 STARPU_NCUDA=1 STARPU_NOPENCL=0 time -p
./tests/datawizard/reclaim
...
real 46.55
user 33.84
sys 15.13

This is actually expected. The test is meant to exercise the memory
reclaiming code, which is used only for gpus.


OK. Could we use less tasks, so that the test does not run for ages ?

With OpenCL, the program just hangs for ever: the program is stuck in
starpu_task_wait_for_all().

OpenCL's port might not actually support reporting memory outage in all
situations.

Is that something we can fix ? This really looks like Ludovic's bug, so this is likely to be a problem in real applications.

Cyril.






Archives gérées par MHonArc 2.6.19+.

Haut de le page