[petsc-users] Petsc cannot be initialized on vesta in some --mode options

[Jeff Hammond]

On Wed, Jan 22, 2014 at 12:38 PM, Jed Brown <jed at jedbrown.org> wrote:
> Jeff Hammond <jeff.science at gmail.com> writes:
>> I've always thought that MPI_Bcast wasn't actually the problem anyways
>> but rather it was the POSIX calls that were breaking PETSc_Init.
>
> Can you elaborate?  PETSc uses fopen on rank 0 (these are small
> user-configuration files).  In this case, it made it through three of
> these, then got an error in a size_t MPI_Bcast.  Setting
> PAMID_COLLECTIVES=0 used to fix this.  I'll update later when I have
> time to reproduce.

I cannot.  It is just my intuition.  I have not verified it
experimentally.  It just seems rather unlikely that MPI_Bcast is
broken given how much it is used.  On the other hand, I believe that
multiple people (including PETSc developers) have demonstrated that
the getpwid etc. stuff is buggy on BGQ.

Jeff

-- 
Jeff Hammond
jeff.science at gmail.com