The common case is to have the image executable stored on all nodes.
But if you don't have a shared filesystem, you can use the "-s" switch
to mpirun to have LAM send your executable around to all nodes. Note,
however, that LAM does not use a scalable method to send your
executable around, so if you go to more than a few nodes, it could be
quite slow.
mpirun -s n1 C image
See the man page for mpirun(1) for more details.
You might want to check out the LAM/MPI User Guide and/or the LAM/MPI
FAQ. See:
http://www.lam-mpi.org/using/docs/
http://www.lam-mpi.org/faq/
On Feb 12, 2005, at 10:09 AM, Toufeeq Hussain wrote:
> HI,
>
> If I have compiled a program on N0 ( on a 2-node
> cluster; N0 and N1 ), do I need to have the program in
> both the nodes to execute it ?
>
> Can this be resolved ?
>
> Currently I get the following error when I have the
> program on N1 but not on N0
>
> mpirun: cannot start image on n0: No such file or
> directory
>
> "image" is the program.
>
> Regards,
> Toufeeq Hussain
>
> =====
> ############################################
> # ring me @ 98401-96690 #
> # mail me @ toufeeq at computer dot org #
> # Debian Sarge \w 2.6.10-ck5 #
> ############################################
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
--
{+} Jeff Squyres
{+} jsquyres_at_[hidden]
{+} http://www.lam-mpi.org/
|
