Re: mpb-mpi slows down with increasing number of nodes (SMP)

On May 7, 2009, at 3:22 PM, Matt Eichenfield wrote:
> In addition, the mpb-mpi output contains some weirdness that I  
> wouldn't necessarily expect.  In particular, the iterations of the  
> calculations return separately for each node.

That's an indication that something is terribly wrong; it should only  
output things once.

Is it possible that mpb-mpi is compiled with one version of MPI (e.g.  
MPICH) but you're using "mpirun" from another version of MPI (e.g.  
OpenMPI)?  That would be my first guess.