MPI was killed after several steps in npt MD, running at vasp5.3.3

Message

Emerald · #1 Post by **Emerald** » Wed May 29, 2013 8:29 pm

Dear all,

I am running a npt MD using the latest 5.3.3 version. I am using 4 nodes with 12 cores each. A gamma point sampling is used. The following is the assignment of those cores:
------------------------------------------------
distrk: each k-point on 48 cores, 1 groups
distr: one band on 6 cores, 8 groups
------------------------------------------------

It could run for two frames, and only two. The energy was converged in each main loop, and the temperature and the volume... was given in REPORT. So far so good. It appeared that everything was fine and the code could run forever.

However, after these 2 steps. The mpi is dead. The error messages:
-----------------------------------------------------------------
forrtl: error (78): process killed (SIGTERM)
........
........
mpirun noticed that process rank 32 with PID 47525 on node n23 exited on signal 11 (Segmentation fault).
-------------------------------------------------------------------
I don't know what is going on. If the parallelism was wrong, why it could work for two frames w/o any problem? I tried to use only one node, the same error exactly took place after two steps as well.

I am confused at this moment. Could someone help me out?
Thank you so much.

Best,
Tian

#2 Post by **admin** » Mon Jun 03, 2013 1:23 pm

Have you checked "segmentation fault" issues?
E.g. proper setting in .bashrc file?