Dear Tim,
On Saturday 03 November 2007 12:16, Tim Reis wrote:
> I have a code that seems to build fine with mpif90. I can submit it to a
> queue using mpirun with no problems but once it has transferred from qw to
> r state it dies very quickly (I say die rather than finish because this is
> a large code that ran for days on an old cluster).
>
> And when I try to look at the .e and .o files that are defined in my script
> I notice they are both empty.
>
> I am quite puzzled by this so any advice will be most gratefully received.
This sounds like a general problem with the cluster, not with MPI, right?
So the first step would be to run any non-MPI application, if that is working
fine (producing output in Your PBS?-output files), then check, whether
mpicc-compiled applications work and finally try the mpif90-compiled
application again.
With best regards,
Rainer
--
----------------------------------------------------------------
Dipl.-Inf. Rainer Keller http://www.hlrs.de/people/keller
HLRS Tel: ++49 (0)711-685 6 5858
Nobelstrasse 19 Fax: ++49 (0)711-685 6 5832
70550 Stuttgart email: keller_at_[hidden]
Germany AIM/Skype:rusraink
"Emails save time, not printing them saves trees!"
|
![[Flag]](../../../../../community/mirrors/flag-us.gif){kind=small}