As I said in the last mail, for the time-being you have to use a non-broken MPI. I listed the tickets. Open MPI has had multiple reduced test cases for several years now, but they haven't done anything to fix it yet.
I will write an alternate implementation of PetscSF that does not use one-sided, but until then, you need a working MPI for the new MatTranspose(). On Wed, Oct 17, 2012 at 2:57 PM, Thomas Witkowski < thomas.witkowski at tu-dresden.de> wrote: > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20121017/7a5a8ee3/attachment.html>
