Re: [Starpu-devel] [Starpu-commits] r8197 - in trunk: doc/chapters gcc-plugin/examples/cholesky src src/core src/sched_policies tests/sched_policies

[Samuel Thibault <samuel.thibault@ens-lyon.org>]

Samuel Thibault, le Tue 04 Dec 2012 12:27:52 +0100, a écrit :
> Ludovic Courtès, le Tue 04 Dec 2012 11:18:31 +0100, a écrit :
> > Samuel Thibault <samuel.thibault@ens-lyon.org> skribis:
> > > Ludovic Courtès, le Tue 04 Dec 2012 10:13:12 +0100, a écrit :
> > >> croelandt@users.gforge.inria.fr skribis:
> > >> > Users should now use dmda. This commit removes the heft scheduler, 
> > >> > updates the
> > >> > documentation and gets rid of heft in the tests.
> > >> 
> > >> My understanding is that HEFT and DMDA are different scheduling
> > >> algorithms.
> > >
> > > They actually use the same heuristic (and indeed get the same
> > > performance on magma for instance).
> > 
> > But they’re not completely equivalent, are they?
> 
> They are meant to be. Differences were only implementation issues.
> 
> > (I failed to find a paper describing DMDA.)
> 
> DMDA ((task) modeling and data aware) is the name that Cédric coined
> in papers to get a name progression from eager to dm then dmda, dmda
> being equivalent to heft (which does both task modeling and data
> modeling).

(scratch latter, it was part of a previous formulation)

Samuel