segmentation fault because NSW is set too high

Problems running VASP: crashes, internal errors, "wrong" results.

Moderators: Global Moderator, Moderator

Post Reply
Message
Author
j.meyer
Newbie
Newbie
Posts: 3
Joined: Wed Jun 15, 2005 1:10 pm
License Nr.: 372
Location: Hannover

segmentation fault because NSW is set too high

#1 Post by j.meyer » Fri Dec 23, 2005 3:11 pm

For this system
http://www.fkp.uni-hannover.de/~meyer/N ... system.tgz
and some similar that I have tried VASP crashes with a segmentation fault
after the first ionic step on several architectures (Athlon, Opteron, Power4)
when NSW is set 'too high'.
The critical value for this system is different from system to system
(Athlon: >=415, Opteron: >=464, Power4: >500 and <1000).
In the supplied INCAR file NSW is set 1000 which should trigger this behaviour
anyway. A (converged) WAVECAR file is supplied to fasten up experimenting
with different values for NSW.
Of course, 1000 are rather many ionic steps. However, more around
500 steps may be necessary for systems like slabs for simulation of the
adsorption of complex molecules (~ 50 non fixed atoms...).
Last edited by j.meyer on Fri Dec 23, 2005 3:11 pm, edited 1 time in total.

admin
Administrator
Administrator
Posts: 2922
Joined: Tue Aug 03, 2004 8:18 am
License Nr.: 458

segmentation fault because NSW is set too high

#2 Post by admin » Tue Jan 10, 2006 8:28 am

sorry, I could not reproduce your crash with NSW=1000 on any of the machines I tried (PentiumIV, Opteron). I rather think that the crashes are not related to NSW itself but to numerical instabilities to find the next trial step of a fully converged system.
I can therefore only give you the advise to try one of the following:
-- concerning NSW: split your job into a sequence of calculations with a 'safe' choice of NSW (and continue from WAVECAR and CONTCAR (copied to POSCAR) of the last step of the previous run
-- try a different ionic relaxation algorithm or
-- set NFREE to the number of steps which you want to keep in the 'relaxation history'
Last edited by admin on Tue Jan 10, 2006 8:28 am, edited 1 time in total.

Post Reply