Le 16/12/2011 16:03, Jeff Squyres a écrit :
> There have been a bunch of updates that have (rightfully) not been back-ported to the 1.2-ompi branch.
> Is 1.3.x stable/mature enough for me to move into OMPI?
> I have two targets:
> * Open MPI v1.5 series: we're just about to release the final Open MPI 1.5.x and transition it to the v1.6 series (i.e., bug fixes only after it is renumbered to 1.6). Is hwloc 1.3.x a good candidate here? As Paul noted, without some of the recent fixes, OMPI 1.5.5 won't run on older Linuxes (e.g., Fedora 8).
> * Open MPI trunk (which will eventually become the v1.7 series): it's ok to be a bit more forward-looking here. I'm actually pretty convinced that this one should be upgraded to hwloc 1.3.1 (or whatever the latest set of fixes is) in the near future.
For OMPI 1.7, you want to upgrade to 1.3.1 for sure. I don't know the
OMPI 1.7 timeframe, but hwloc 1.4 might even be a better idea by then.
For OMPI 1.5/1.6, I am not sure. If fixing redhat 8 (not fedora) is
really important, we can probably backport some patches.
Upgrading to 1.3.1 just because of this seems like overkill. I am not
saying that 1.3.1 won't be stable. We had a lot of testing recently.
However upgrading to 1.3.1 would likely ease the maintenance of the OMPI
1.6 later because hwloc patches would be easier to backport. IIRC
there's way more changes between 1.2 and 1.3 than between 1.3 and 1.4
(and likely 1.5).