Open MPI logo

Hardware Locality Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Hardware Locality Development mailing list

Subject: Re: [hwloc-devel] PPC64 problem with hwloc 1.1.1
From: Samuel Thibault (samuel.thibault_at_[hidden])
Date: 2011-02-22 18:02:31


Jeff Squyres, le Tue 22 Feb 2011 23:49:31 +0100, a écrit :
> Brad Benton from IBM is reporting a problem with hwloc 1.1.1 on PPC64. See the details here:
>
> https://svn.open-mpi.org/trac/ompi/ticket/2732
>
> The printf that he's referring to is coming directly from hwloc. I can certainly suppress this printf in OMPI's copy of hwloc, but it does seem to be weird. He also includes the native lstopo output, which shows the same printf noise.

Note the "/* TODO: how to report? */" comment in the code: we definitely
_want_ to get users to see the warning and report it. It means the
kernel is giving contradictory information, and hwloc just happened to
take the first one and not the other.

Brad, could you post the output of the hwloc-gather-topology script from
hwloc? That's permit to know why the warning shows up.

Samuel