[Pw_forum] process killed

Giuseppe Mattioli giuseppe.mattioli at ism.cnr.it
Mon Jun 16 14:51:09 CEST 2014



> i do PBE0 calculation for the metal(Fe) and the oxyde (MgO), with 16 atoms

Perhaps not a good idea to use an unscreened functional like PBE0 for metals, but it is your system and you know what you are looking for...

> and everything seems to be OK now when i want to do calculations for the
> junction Fe/MgO with 38 atoms.

You are ~doubling the number of electrons, which is the bottleneck of EXX calculations. You may run short of memory. Use more nodes, if you can, and 
try disk_io='high': the calculation should be slower but it also should use less memory. Remember to force some soft exit (like max_seconds=40000, or 
something suitable), if you want to properly restart the calculation.

> more clarifications, please

The EXX routines have been extensively modified in QE 5.1. This often implies that the calculations a) are faster; b) use less resources.

HTH

Giuseppe


On Monday 16 June 2014 12:06:46 toufik esssakhri wrote:
> Dear Prof. Giuseppe
> Thank you
> 
> > It seems some kind of segmentation fault. It is difficult
> 
> to say more than this without knowing more about the simulation and the
> machine architecture.
> i study the very known metal/oxyde junction.
> i do PBE0 calculation for the metal(Fe) and the oxyde (MgO), with 16 atoms
> and everything seems to be OK now when i want to do calculations for the
> junction Fe/MgO with 38 atoms.
> 
> this is part of my script pbs
> ====================================================
> #!/bin/bash
> #PBS -N test-pbe-supercell
> #PBS -l nodes=4:ppn=8,pmem=2000000kb
> #PBS -l walltime=11:59:59
> #PBS -r n
> #PBS -l nice=11
> #PBS -m abe
> #PBS -M tousak at hotmail.fr
> #PBS -o master0:$PBS_O_WORKDIR/$PBS_JOBNAME.$PBS_JOBID.pbslog.o
> #PBS -e master0:$PBS_O_WORKDIR/$PBS_JOBNAME.$PBS_JOBID.pbslog.e
> export JOBID=$PBS_JOBID
> ulimit -s unlimited
> source /opt/intel/composer_xe_2013_sp1.3.174/bin/compilervars.sh intel64
> module load openmpi-1.8.1
> module load espresso-5.0.1_openmpi-1.8.1
> =====================================================
> 
> > However, try to use the brand new 5.1 release: I've found some
> > improvement in EXX.
> 
> more clarifications, please
> 
>  cheers
>  taoufik


********************************************************
- Article premier - Les hommes naissent et demeurent
libres et ègaux en droits. Les distinctions sociales
ne peuvent être fondèes que sur l'utilitè commune
- Article 2 - Le but de toute association politique
est la conservation des droits naturels et 
imprescriptibles de l'homme. Ces droits sont la libertè,
la propriètè, la sùretè et la rèsistance à l'oppression.
********************************************************

   Giuseppe Mattioli                            
   CNR - ISTITUTO DI STRUTTURA DELLA MATERIA   
   v. Salaria Km 29,300 - C.P. 10                
   I 00015 - Monterotondo Stazione (RM)          
   Tel + 39 06 90672836 - Fax +39 06 90672316    
   E-mail: <giuseppe.mattioli at ism.cnr.it>
   http://www.ism.cnr.it/english/staff/mattiolig
   ResearcherID: F-6308-2012




More information about the users mailing list