[Pw_forum] Time_max variable in phonon calculation
Alcantara Ortigoza, Marisol
alcantar at phys.ksu.edu
Tue Apr 26 07:02:52 CEST 2005
Dear all,
I am calculating the phonon frequencies of a 20-atom system. The
calculation (in parallel with 16 processors) for high symmetry points
take between 90 hours and 200 hours. However, the computer system in
which I'm working only allow us to run jobs for less than 24 hours;
after that time it kills the job. That is way I am using the time_max
variable so that the job stops smoothly. After the number of seconds I
allowed it to run has been reached ( time_max = 85000) it stops showing:
"Stopping for time limit 85048. 85000."
That is like 23 hours and a half. Then, it indeed does nothing but the
machines keep pretending they are doing something until the 24 hours I
asked for have passed. Then, of course, I get an error file from the
cluster system saying the job exceeded the allowed time etc., etc.
My question is, after the time_max has been reached (and the job stops),
can I be sure that the code will safely restart later without any
problem?
Before using time_max variable I wasted many hours given that sometimes
the system kills the job without any problem to restart, but sometimes
it spoils everything and one loses all the information previously
obtained from earlier runs.
Thank you very much.
Marisol Alcantara Ortigoza.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/attachments/20050426/47d0e50f/attachment.htm
More information about the Pw_forum
mailing list