Am 18.02.2013 um 13:37 schrieb Jeff Hammond:

I've informed the appropriate people about the Trac issue.

Thanks.

I recall that MPI stipulates only that MPI_UNIVERSE_SIZE exist; it need not be defined.

I think that MPICH might set MPI_UNIVERSE_SIZE in the process manager. What PM are you using?

I started with a plain `mpiexec -np 4 ./program`, hence it should use Hydra as it's a symbolic link to mpiexec.hydra. -- Reuti

Jeff

On Mon, Feb 18, 2013 at 5:52 AM, Reuti <[email protected]> wrote:

...

Hi,

a call:

error=MPI_Comm_get_attr(MPI_COMM_WORLD, MPI_UNIVERSE_SIZE, &universe_sizep, &flag);

worked in MPICH2 1.4.1p1. But the "flag" returned unsupported in 1.5 and 3.0.2. Was this info removed from being retrieved?

-- Reuti

PS: I found http://trac.mpich.org/projects/mpich/ticket/1426 mentioning MPI_UNIVERSE_SIZE but it throws an error:

TracError: The Trac Environment needs to be upgraded.

Run "trac-admin /www/trac.mpich.org/projects/mpich upgrade" _______________________________________________ discuss mailing list [email protected] To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

-- Jeff Hammond Argonne Leadership Computing Facility University of Chicago Computation Institute [email protected] / (630) 252-5381 http://www.linkedin.com/in/jeffhammond https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond _______________________________________________ discuss mailing list [email protected] To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss