The ABINIT Users Mailing List ( CLOSED )

[Michel Côté <michel.cote@umontreal.ca>]

Title: Re: [abinit-forum] too much memory demand

I am surprised this problem requires that much memory. The unit cell contains only 2 atoms! It is true that you are using a very large k-point grid. Maybe you do not need that much and you should try with smaller grid. I think BaO is an insultator and a grid of 6x6x6 should be enough. In anycase, a convergence test with respect to k-point grid should be done.

Michel 


Le 11/12/06 08:04, « Anglade Pierre-Matthieu » <anglade@gmail.com> a écrit :

> If you can use a parallel version. The parallelisation ovber k-point
> reduce fdramatically the memory needs per nodes.
> regards
> PMA
>
> On 12/10/06, pablo <pablod@fq.edu.uy> wrote:
> > Hi,
> >
> >          I have used mffmem=0 to reduce a little the mem (700mb to 500mb), I
> > will be happy to hear something better, mem is a huge problem...
> >
> > regards,
> >
> >                    pablo
> >
> > ----- Original Message -----
> > From: <mailoliver@163.com>
> > To: <forum@abinit.org>
> > Sent: Monday, December 11, 2006 3:48 AM
> > Subject: [abinit-forum] too much memory demand
> >
> >
> > > Dear all:
> > >
> > > I try to calculate the optical property of BaO in b1 phase. I modify the
> > > input file of toptic_1.in. And I have done convergence studies for
> > > ecut(37.5Hatree). When I set ngkpt=15*15*15, it needs more than 10G
> > > memory. How can I reduce the memory demand?
> > >
> > > best wish
> > >
> > > tieyu lv
> > >
> > > my input::
> > > **************************************
> > > ndtset 6
> > >
> > >   iscf1  3
> > > #  nband1  13
> > >  nstep1 9999
> > > kptopt1 1
> > > nbdbuf1 0
> > > prtden1 1   getden1 0   getwfk1 0    ! Usual file handling data
> > > toldfe1 1.0d-6
> > >
> > > #Second dataset : NSC run with large number of bands, and points in the
> > > IBZ
> > >    iscf2 -2
> > >   nband2  24
> > >   nstep2 9999
> > >  kptopt2  1
> > >  getwfk2  1   getden2 1
> > >  tolwfr2 1.0d-20
> > >
> > > #Third dataset : NSC run with large number of bands, and points in the the
> > > full BZ
> > >    iscf3 -2
> > >   nband3  24
> > >   nstep3 9999
> > >  kptopt3  2
> > >  getwfk3  2   getden3 1
> > > tolwfr3 1.0d-20
> > >
> > > #Fourth dataset : ddk response function along axis 1
> > >   iscf4 -3
> > >  nband4  24
> > >  nstep4  1  nline4  0
> > > kptopt4  2
> > >
> > >   nqpt4  1  qpt4  0.0d0 0.0d0 0.0d0
> > >  rfdir4  1 0 0
> > > rfelfd4  2
> > > getwfk4  3
> > > tolwfr4 1.0d-20
> > > nbdblock4 4
> > > wfoptalg4 1
> > >
> > > #Fifth dataset : ddk response function along axis 2
> > >   iscf5 -3
> > >  nband5  24
> > >  nstep5  1  nline5  0
> > > kptopt5  2
> > >
> > >   nqpt5  1  qpt5  0.0d0 0.0d0 0.0d0
> > >  rfdir5  0 1 0
> > > rfelfd5  2
> > > getwfk5  3
> > > tolwfr5 1.0d-20
> > > nbdblock5 4
> > > wfoptalg5 1
> > >
> > > Sixth dataset : ddk response function along axis 3
> > >   iscf6 -3
> > >  nband6  24
> > >  nstep6  1  nline6  0
> > > kptopt6  2
> > >
> > >   nqpt6  1  qpt6  0.0d0 0.0d0 0.0d0
> > >  rfdir6  0 0 1
> > > rfelfd6  2
> > > getwfk6  3
> > > tolwfr6 1.0d-20
> > > nbdblock6 4
> > > wfoptalg6 1
> > >
> > >
> > > #Data common to all datasets
> > > nshiftk 1
> > > shiftk
> > >         0.0 0.0 0.0
> > > ngkpt  3*15
> > >
> > > acell   3*1.0493241067E+01
> > > amu 137.33 15.9994
> > > diemac 12.0
> > > ecut 37.5
> > > iscf 3
> > > natom  2  nbdbuf 2
> > > ntypat  2
> > > rprim   0 .5 .5  .5 0 .5  .5 .5 0
> > > xred 3*0.00d0 3*0.5d0
> > > tnons 144*0.0
> > > typat  1 2
> > > znucl  56 8
> > >
> >
> >
> >


-- 
*************************************************************** 
  Michel Cote                          tel: +1 (514) 343-5628     
  Professeur agrégé                   fax: +1 (514) 343-2071               
  Département de physique 
  Université de Montréal 
  C.P. 6128, succ. Centre-ville 
  Montréal (Québec) H3C 3J7          Michel.Cote@umontreal.ca 
  Canada                   www.phys.umontreal.ca/~michel_cote 
***************************************************************