[gmx-users] Segmentation fault on HPC

Tomek Wlodarski tomek.wlodarski at gmail.com
Sat Jan 3 13:56:17 CET 2015


I am trying to set up simulation on HPC.
Energy minimization in vacuum and in water works great (using 4 nodes with
96 cores altogether)
But when I am trying to run standard md with:

aprun -n 96 mdrun_mpi -v -deffnm md >& out

I always end up with this error (I tried it also on different gromacs
versions: 4.6.5, 5.0, 5.0.4):

starting mdrun 'Protein in water'
500000 steps,   1000.0 ps.
step 0

NOTE: Turning on dynamic load balancing

_pmiu_daemon(SIGCHLD): [NID 00383] [c1-0c2s15n3] [Sat Jan  3 12:26:44 2015]
PE RANK 72 exit signal Segmentation fault
_pmiu_daemon(SIGCHLD): [NID 00380] [c1-0c2s15n0] [Sat Jan  3 12:26:44 2015]
PE RANK 1 exit signal Segmentation fault
_pmiu_daemon(SIGCHLD): [NID 00382] [c1-0c2s15n2] [Sat Jan  3 12:26:44 2015]
PE RANK 48 exit signal Segmentation fault
[NID 00380] 2015-01-03 12:26:44 Apid 12376091: initiated application
Application 12376091 exit codes: 139
Application 12376091 exit signals: Killed
Application 12376091 resources: utime ~16s, stime ~8s, Rss ~63064, inblocks
~363834, outblocks ~1173496

Do you have any suggestion what is wrong (why I can run EM not MD?)? I
checked different system and gromacs run ok.


More information about the gromacs.org_gmx-users mailing list