[gmx-users] Gromacs job locks up computer (reproducibly)
feenstra at chem.vu.nl
Wed Aug 20 08:59:01 CEST 2003
Marc Baaden wrote:
> I am currently onto a very strange behaviour of a Gromacs run
> on a dual athlon box. The job only used a single processor and
> for the second time it locks up the computer without any
> error message or distinct sign of what went wrong.
> It is a long job, and for the second time it locked the machine
> after about 10 days on exactly the same timestep.
> As I said, no error message, the machine remains pingable, but
> no possibility to login neither via the network nor on the console.
> Keyboard is dead.
> The gromacs log/data files just stop at that given point.
> Has anybody ever experienced something similar ?
Can't say that I did, but I'll dumb-guess at some possible causes for
this behaviour (this is my usual routine for finding explanations for
strange behaviour ;-). How many steps has your simulation run, i.e. is it
a very large number like 2^31, 2^32 or 2^63 or so? Have any of the output
files grown to about 2Gb? Does it depend on the output (nst*out) settings?
Is it dependent on the system you simulate? Does it happen on another,
identical, dual athlon? Could you swap CPU's and reproduce the error (i.e.
could it be a faulty CPU)? Does it happen on a non-identical dual athlon?
Is it dependent on memory (type, size, usage), machine load, gromacs
version, compiler options?
P.S. don't start checking all of this if you haven't done it already or
can't do it easily. It's meant as a couple of pointers, one of them might
actually be helpful...?
| | |
| _ _ ___,| K. Anton Feenstra |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|( | )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands |
| \_/ \_/ | | | Tel: +31 20 44 47608 - Fax: +31 20 44 47610 |
| | Feenstra at chem.vu.nl - www.chem.vu.nl/~feenstra/ |
| | "If You See Me Getting High, Knock Me Down" |
| | (Red Hot Chili Peppers) |
More information about the gromacs.org_gmx-users