Hi Frederico,
> I'm having some issues with MPICH, this is the error that SAM presents
> to me:
>
> 0 - MPI_SEND : Invalid rank 1
> message size transfertime bandwidth
> p0_2493: p4_error: : 8262
> [0] Aborting program !
> [0] Aborting program!
> mpiexec: Warning: poll_or_block_event: evt 3 remote system error.
> mpiexec: Warning: task 0 exited with status 1.
>
> I've already referred to
> (https://twiki.cern.ch/twiki/bin/view/EGEE/MpiTools
> <https://twiki.cern.ch/twiki/bin/view/EGEE/MpiTools> and
> http://wiki.ifca.es/e-ciencia/index.php/MPI_Errors
> <http://wiki.ifca.es/e-ciencia/index.php/MPI_Errors>), but I found no
> hints there.
>
> After searching the internet, and suspect that there is a problem with
> the mpich-p4 communicator and new kernels.
> Has anyone experienced the same problem?
> The only 'solution' I've found so far is to use mpich2. However,
> according to the SA1 twiki, only mpich and openmpi are supported.
MPICH2 is OK, the TWiki page is out of date.
Beware of a bug in YAIM: the "MPI_CE" target should be given _before_
other targets on the YAIM command line. Ticket:
https://gus.fzk.de/ws/ticket_info.php?ticket=55447
|