Re: [Starpu-devel] Worker Binding Problem

[Andra Hugo <andra.hugo@inria.fr>]

Yes it is a nicer solution :) but be careful if at some point you will use 
openmp(mkl) parallel tasks in starpu, that would mess up some things. We rely 
on having those threads alive to do this.

----- Mail original -----
> De: "Berenger Bramas" <berenger.bramas@inria.fr>
> À: "Samuel Thibault" <samuel.thibault@inria.fr>
> Cc: starpu-devel@lists.gforge.inria.fr
> Envoyé: Mercredi 9 Septembre 2015 14:45:03
> Objet: Re: [Starpu-devel] Worker Binding Problem
> 
> So, the OpenMP variable
> OMP_WAIT_POLICY={PASSIVE|ACTIVE}
> let control that and now it works well,
> Thanks again.
> 
> Bérenger Bramas
> 
> HiePACS Project
> 
> Tel (05 24 57) 40 76
> INRIA BORDEAUX Sud Ouest
> 
> 
> ----- Mail original -----
> | De: "Samuel Thibault" <samuel.thibault@inria.fr>
> | À: "Berenger Bramas" <berenger.bramas@inria.fr>
> | Cc: starpu-devel@lists.gforge.inria.fr
> | Envoyé: Mercredi 9 Septembre 2015 14:41:13
> | Objet: Re: [Starpu-devel] Worker Binding Problem
> | 
> | Berenger Bramas, le Wed 09 Sep 2015 14:32:55 +0200, a écrit :
> | > I set OMP_DYNAMIC=false to ensure the OpenMP thread to not spin while
> | > waiting
> | > (even if I am out of the OpenMP parallel section when I execute StarPU
> | > test
> | > so I suppose the GCC implementation put the thread to sleep anyway)
> | > So I clearly do not understand why there is this behavior.
> | 
> | As discussed in real-life, we tend to thing that gcc doesn't actually
> | put threads to sleep, but makes them spin, ready for another parallel
> | section.
> | 
> | Samuel
> | 
> _______________________________________________
> Starpu-devel mailing list
> Starpu-devel@lists.gforge.inria.fr
> http://lists.gforge.inria.fr/mailman/listinfo/starpu-devel
>