[CP2K:8472] cp2k/3.0 optimisation gives CRASSERT error in cp2k/4.1

Iain Bethune i.be... at epcc.ed.ac.uk
Mon Dec 12 10:49:00 CET 2016


Dear Julia,

I am working on some code changes that will reduce the chance of this occurring if you are using large numbers of OpenMP threads.  However, it will still occur if you are using a grid size of roughly 4 billion points, which is huge.  As Juerg said, this is most likely due to having a very large unit cell and/or high CUTOFF setting (or possibly, for MPI parallel jobs, having some very diffuse basis functions).

You can try adding

  &PRINT
    &GRID_INFORMATION ON
    &END
  &END

inside the FORCE_EVAL section. This should output some useful information about the grid sizes and decomposition settings, before your job exits with the CPASSERT error.  Also check the output near the start for the number of MPI processes and OpenMP threads that CP2K is using.  It’s possible if you are running an OpenMP-enabled build of CP2K that OMP_NUM_THREADS defaults to the number of cores in your machine, which is probably not what you wanted.

- Iain

--

Iain Bethune
Project Manager, EPCC

Email: i.be... at epcc.ed.ac.uk
Twitter: @IainBethune @PrimeGrid @CP2Kproject
Web: http://www2.epcc.ed.ac.uk/~ibethune
Tel/Fax: +44 (0)131 651 7183/6555
Mob: +44 (0)7598317015
Addr: 2404 JCMB, The King's Buildings, Peter Guthrie Tait Road, Edinburgh, EH9 3FD

> On 8 Dec 2016, at 17:43, donaldducksdaughter <donaldduc... at gmail.com> wrote:
> 
> Hi,
> 
> Thanks a lot for your reply! At least I know what to look for now! I have changed my submission script and even with the support from my technician, however it's still giving me the same CPASSERT error. 
> 
> I'll let you know how it goes.
> 
> regards
> 
> Julia Schmidt
> 
> Am Dienstag, 6. Dezember 2016 10:11:30 UTC+1 schrieb jgh:
> Hi 
> 
> from the error message I can see that you have a Integer overflow problem. 
> Why you get the problem is not possible to determine without additional 
> information. My guess is that you try to run a large box with high cutoff 
> with a small number of MPI tasks and a large number of OpenMP threads. 
> As a first try I would use the popt module (OpenMP threads = 1) with as 
> many as possible MPI tasks. 
> 
> regards 
> 
> Juerg Hutter 
> -------------------------------------------------------------- 
> Juerg Hutter                         Phone : ++41 44 635 4491 
> Institut für Chemie C                FAX   : ++41 44 635 6838 
> Universität Zürich                   E-mail: hut... at chem.uzh.ch 
> Winterthurerstrasse 190 
> CH-8057 Zürich, Switzerland 
> --------------------------------------------------------------- 
> 
> -----cp... at googlegroups.com wrote: -----To: cp2k <cp... at googlegroups.com> 
> From: donaldducksdaughter 
> Sent by: cp... at googlegroups.com 
> Date: 12/05/2016 12:22PM 
> Subject: [CP2K:8455] cp2k/3.0 optimisation gives CRASSERT error in cp2k/4.1 
> 
> Hi all, 
> I tried to run a geometry optimisation in cp2k/4.1. The same optimisation runs smoothly in cp2k/3.0. Hence, I used the same input file for cp2k/4.1 and just called the different executable. Unfortunately, the same input does not work in 4.1 and the following erro was obtained (also see screenshot): 
> CPASSERT failed      - pw/lgrid_types.F:118   
> 
> Additionally, I tried to fix the error comparing with the only other CASSERT error, I could find (https://www.cp2k.org/faq:cholesky_decomp_failed). However, this is a slightly different error and possible solutions such as changing the preconditioner did not fix my error. 
> Please find my input file and error screenshot attached for your reading. 
> Many thanks in advance 
> Julia 
> 
> 
> 
> 
> 
> 
> -- 
> 
> You received this message because you are subscribed to the Google Groups "cp2k" group. 
> 
> To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+... at googlegroups.com. 
> 
> To post to this group, send email to cp... at googlegroups.com. 
> 
> Visit this group at https://groups.google.com/group/cp2k. 
> 
> For more options, visit https://groups.google.com/d/optout. 
> 
> 
> 
> [attachment "Screen Shot 2016-12-05 at 12.07.13.png" removed by Jürg Hutter/at/UZH] 
> [attachment "test.in" removed by Jürg Hutter/at/UZH] 
> 
> -- 
> You received this message because you are subscribed to the Google Groups "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns... at googlegroups.com.
> To post to this group, send email to cp... at googlegroups.com.
> Visit this group at https://groups.google.com/group/cp2k.
> For more options, visit https://groups.google.com/d/optout.


-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.




More information about the CP2K-user mailing list