Dear Gabriele, Thank you very much. I'll try what you suggest.
Trinh On 3/12/13 5:48 AM, "Gabriele Sclauzero" <gabriele.sclauzero at epfl.ch> wrote: > >I don't see why the max_seconds option should not work. >You might just need to decrease the max_seconds value by a proper amount >of time, in order to be sure that the elapsed time is checked by the >program before the end of the actual walltime of your batch job. This >amount depends on the time needed by the code to complete one iteration >within the self-consistent loop. > >In order to stop the job "interactively", you have to create a >$prefix.EXIT file in the directory where you started pw.x ($prefix should >be replaced with the actual value of that input variable). Anyway, the >program checks the presence of the file and the elapsed time at the same >moment, so this does not solve the above mentioned problem (i.e., the >code cannot be stop "immediately"). > >HTH > >GS > >> Dear PwSCF Users, >> >> I am sorry that I have one more question. Since the job I am running is >> for a very large metallic system, and I only run a scf calculation. The >> problem I have is that my job is constrained by the time limit (I use >> shared computing resource, which has 12 cores/node). When the time is >> met, my job is killed by administrator. Since the job is killed >>brutally, >> the wave functions are corrupted afetr a number of restarts. I want to >> interactively stop the job nicely before it is killed by administrator. >> I >> have used the option "max-seconds" as mentioned by instruction, but it >>did >> not work. Thus, I have to find a different way to stop the job without >> corrupting the wave function files. It will be helpful very much if you >> could suggest to me what is the best way to do so or which subroutines >>of >> the code that I should look at to modify in order to stop the job >>nicely. >> >> Thank you in advance very much, >> >> Trinh > > >? Gabriele Sclauzero, EPFL SB ITP CSEA > PH H2 462, Station 3, CH-1015 Lausanne > > > > > > > > >_______________________________________________ >Pw_forum mailing list >Pw_forum at pwscf.org >http://pwscf.org/mailman/listinfo/pw_forum
