On Fri, Apr 21, 2017, sylvester kisembo wrote:
> After running the job with a longer wall time than 30 minutes, the
> segmentation fault re-appeared.
> Below is the full
> error message for the run with no cuda and no mpi.
Segmentation faults are of very little help without the actual files. But
some questions may help:
1. Which version of Amber?
2. In the previous emails, you found end-of-file messages in reading the
prmtop file....did those go away? Is this is different prmtop file?
3. It is still quite odd that you are getting repeated error messages: are
all these systems running through a queueing system? Is it possible that
you are running the same program many times simultaneously (e.g. by using
an "srun" command that invokes MPI, even though your are not using the MPI
verison of the program)? Can you see an error just running a short test
from the command-line (with no slurm or other scheduler involved)?
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Apr 21 2017 - 19:30:02 PDT
