The ABINIT Users Mailing List ( CLOSED )

[Yann Pouillon <yann.pouillon@ehu.es>]

Hi,

Latévi Max LAWSON DAKU <Max.Lawson@unige.ch> ha escrito:

> I would like to use the band/FFT/k-point parallelisation, and several
> distributions have been proposed for the number of processors.
> The distributions retained read (npkpt, npband, npfft)=(2, 150, x),
> with x a positive integer.
>
> But the calculations tend to be extremely slow even when using
> 3000 cpus. In this last case, the code actually segfaults after having
> output:
>
>    ITER STEP NUMBER     1
>     vtorho : nnsclo_now=  2, note that nnsclo,dbl_nnsclo,istep=  0 0  1
>      ****  In vtorho for isppol= 1
>     starting lobpcg, with nblockbd,mpi_enreg%nproc_band 1 150
>     condition number of the Gram matrix= 59476636.484308563
>     Lobpcgccwf: restart performed

I have the same problem with nanotubes using PAW on a different  
architecture. The segfault disappears when I reduce the cell size  
(which in absolute I don't want). I've been trying to change the  
following parameters:

  * increasing nband
  * varying npband, npfft, bandpp
  * increasing diecut, dielam
  * varying iprcel, iprcch
  * increasing nnsclo
  * trying both metallic and non-metallic nanotubes

I also very carefully checked the positions of the atoms. But the  
electronic ground-state SCF never converges.

I'm pretty sure there is a memory leak somewhere, but I was unable to  
locate it. Smells to me like a problem in the preconditioning, but may  
be elsewhere.

If somebody has more insight about this, I would greatly appreciate any hint.

Best regards,

Yann.

--
Yann Pouillon               European Theoretical Spectroscopy Facility (ETSF)
Tel: (+34) 943 01 83 94     Centro Joxe Mari Korta, Avenida de Tolosa, 72
Fax: (+34) 943 01 83 90     20018 Donostia-San Sebastián (Gipuzkoa)
Web: http://www.etsf.es/    España