On Aug 30, 2007, at 11:43 AM, Nestor Waldyd Alvarez Villa wrote:
> [waldyd_at_simeca ~]$ mpirun -np 8 -v -s n0 a.out
> 6728 a.out running on n0 (o)
> 4967 a.out running on n1
> 4968 a.out running on n1
> 6729 a.out running on n0 (o)
> 4969 a.out running on n1
> 4970 a.out running on n1
> 6730 a.out running on n0 (o)
> 4971 a.out running on n1
> Hellow, MPI! (3/8)-- simeca.gialea
> Hellow, MPI! (0/8)-- simeca.gialea
> Hellow, MPI! (6/8)-- simeca.gialea
> [waldyd_at_simeca ~]$
>
> As you can see, node "n1" is not answering. Any ideas about how to
> solve
> this issue?
By "not answering", I assume you mean that you don't see the "Hellow,
MPI!" messages from those MPI_COMM_WORLD ranks, right?
Are the two nodes homogeneous, meaning that they are the same
hardware, OS, etc.? One reason that you might not see the printf's
is if there is some difference between them such that the a.out
compiled on one node won't run completely properly on the other.
See this FAQ category: http://www.lam-mpi.org/faq/category11.php3
--
Jeff Squyres
Cisco Systems
|
