[gmx-users] Problems running gmx314 on SunOS/sparc: core dump

[Marc Baaden]

Hi,

I am losing my nerves over strange problems with running Gromacs
jobs on a SunOS Generic_108528-15 sun4u sparc SUNW,Sun-Fire-880
machine.


The job crashes in the very first instances with
[..usual blurb..]
1250000 steps,   2500.0 ps.

Signaled (core dumped).


I have run -debug, but the file is *HUGE* .. and I wonder if there is
anything I could/should grep for ?


The core dump happens, whether I use 1, 2, 12 or 16 processors,
so it's ?probably? not an mpi problem.


Strangely, a colleague reported running things successfully on his
system. But I have checked, on another cluster and on other boxes
(all running linux, though) my system runs absolutely fine.


The people administering the cluster here have looked into it,
and it does not seem to be a problem with the cluster. But one
of the comments was quite interesting:

> A warning though: In my last job I had to port a lot of s/w from linux to
> solaris and vice versa and it's not always easy. I had numerous occaisions
> where code compiled flawlessly on both but core-dumped at run time. They
> weren't always quick to fix even with a debugger. I hope you are not in
> this situation but I wanted to warn you that porting from one system to
> the other doesn't always work first time even when the compiler says
> otherwise!

So, are people using Gromacs successfully on these
OSC SunFire 6800 Shared Memory Systems ?


(The detailed hardware description is at
http://www.osc.ox.ac.uk/hardware.html   )
 

Any help is *greatly* appreciated, as I am trying to work out a
solution for this for quite some time now, with help from several people.

Cheers,
  Marc Baaden

-- 
 Dr. Marc Baaden - Laboratory of Molecular Biophysics, Oxford University
 mailto:baaden at smplinux.de  - ICQ# 11466242 -  http://www.marc-baaden.de
 FAX/Voice +49 697912 39550  -  Tel: +44 1865 275380  or  +33 609 843217