[Pw_forum] process killed

toufik esssakhri tousak at hotmail.fr
Wed Jun 18 15:07:05 CEST 2014


thank you very much
I am very happy for this discussion
I will consider your suggestions and try 

cheers
taoufik


> From: giuseppe.mattioli at ism.cnr.it
> To: pw_forum at pwscf.org
> Date: Wed, 18 Jun 2014 14:35:13 +0200
> Subject: Re: [Pw_forum] process killed
> 
> 
> A general suggestion: try to simplify the calculation to the bone and check if it works. Then add features until you feel you have reached 
> convergence, or until the calculation crashes down...
> 
> Some specific suggestions: 
> 
> >     verbosity='high'
> not necessary
> 
> > wf_collect = .True.,
> not strictly necessary, and sometimes source of i/o or post processing error.
> 
> > disk_io=.true.
> 
>    +--------------------------------------------------------------------
>    Variable:       disk_io
> 
>    Type:           CHARACTER
>    Default:        'low'
>    Description:    Specifies the amount of disk I/O activity
>                    'high':   save all data to disk at each SCF step
> 
>                    'medium': save wavefunctions at each SCF step unless
>                              there is a single k-point per process (in which
>                              case the behavior is the same as 'low')
> 
>                    'low' :   store wfc in memory, save only at the end
> 
>                    'none':   do not save anything, not even at the end
>                              ('scf', 'nscf', 'bands' calculations; some data
>                               may be written anyway for other calculations)
> 
>                    Note that the amount of needed RAM increases as the amount
>                    of I/O decreases! IMPORTANT: default has been changed to 'low'.
>                    It is no longer needed to specify 'high' in order to be able
>                    to restart from an interrupted calculation (see "restart_mode")
>                    but you cannot restart from disk_io='none'
>    +--------------------------------------------------------------------
> 
> >   ecutwfc =80,
> I suppose that you have checked the convergence of your PPs wrt ecutwfc in smaller systems, but 80 Ry may be a bit low...
> 
> > nqx1=2,   nqx2=2,    nqx3=2,
> > K_POINTS {automatic}
> >    8 8 2 1 1 1
> 
> A very demanding sampling! Start with something lower, e.g.
> 
> nqx1=1,   nqx2=1,    nqx3=1,
> K_POINTS {automatic}
> 4 4 1 1 1 1
> 
> or also
> 
> nqx1=1,   nqx2=1,    nqx3=1,
> K_POINTS {automatic}
> 2 2 1 1 1 1
> 
> And check if it works. Then try to use the same setup for your separate oxide and iron calculations (you reported that such calculations were fine in a 
> previous message), and check whether the results obtained by using the "high" setup are so better than those obtained by using the "low" setup. 
> Finally, if you still need a better sampling, then start using a denser k-point grid rather then a denser q-point grid, e.g.,
> 
> nqx1=1,   nqx2=1,    nqx3=1,
> K_POINTS {automatic}
> 8 8 2 1 1 1
> 
> Finally, don't forget to use this:
> 
>    +--------------------------------------------------------------------
>    Variable:       adaptive_thr
> 
>    Type:           LOGICAL
>    Default:        .FALSE
>    Description:    If .TRUE. this turns on the use of an adaptive conv_thr for
>                    the inner scf loops when using EXX.
>    +--------------------------------------------------------------------
> 
> It gives a consistent speedup...
> 
> HTH
> Giuseppe
> 
> On Wednesday 18 June 2014 12:06:52 toufik esssakhri wrote:
> > thank you very much!
> > 
> > > I'm afraid I cannot help you more than this, at least without seeing your
> > > input file.
> > 
> > this is part of my input file:
> > ======================================================
> >   &control
> >     calculation='scf',
> >     prefix='tst-pbe0-fm5fo',
> >     pseudo_dir='/workdir/share/pseudo_sakhraoui',
> >     outdir='/home/ipcms/sakhraoui/scratch',
> >     restart_mode='restart',
> >     verbosity='high',     wf_collect = .True.,
> >     disk_io=.true., max_seconds=40000,
> >  /
> >  &system
> >   ibrav=6,  A=4.18,  C=24.72,
> >   nat =38,   ntyp=4,
> >   ecutwfc =80,
> >   occupations='smearing',    smearing='mv',    degauss=0.01,
> >   nspin = 2,   starting_magnetization(1)=0.5,
> > starting_magnetization(2)=0.5, starting_magnetization(3)=0.3,
> >    input_dft='pbe0',   nqx1=2,   nqx2=2,    nqx3=2,
> >    exxdiv_treatment='gygi-baldereschi',
> > /
> >  &electrons
> >      mixing_mode = 'local-TF' ,
> >      mixing_beta = 0.4,     diagonalization = 'david' ,
> >      conv_thr = 1.0d-6,     electron_maxstep=10000,
> > /
> > ATOMIC_SPECIES
> >     Fe1 55.85    Fe.pbe-mt_fhi.UPF
> >     Fe2 55.85    Fe.pbe-mt_fhi.UPF
> >     Rh 102.91    Rh.pbe-mt_fhi.UPF
> >     Mg 24.305    Mg.pbe-mt_fhi.UPF
> >     O  10.999    O.pbe-mt_fhi.UPF
> > ATOMIC_POSITIONS {angstrom}
> > ..................
> > K_POINTS {automatic}
> >    8 8 2 1 1 1
> > 
> > 
> > ======================================================
> 
> 
> ********************************************************
> - 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
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20140618/3544ae09/attachment.html>


More information about the users mailing list