[CP2K-user] [CP2K:12796] 120680 atom system - CPASSERT failed
hut... at chem.uzh.ch
hut... at chem.uzh.ch
Tue Jan 21 11:06:26 UTC 2020
Hi
what Matt was pointing to is
1) there might be not enough memory
2) the plane wave part of the code might be buggy for this large
size of code
As we have done calculations of similar size cells, my guess would be 1).
If I run your cell with one single water molecule and a cutoff of 100,
the job fits on my workstation with 256 GB memory. For a cutoff of
800 you will need approx. 8^1.5=22.6 times more memory (= 6 TB).
However, these are very rough estimates and my jobs stop with another
error message if there is not enough memory.
In any case, the CPASSERT points to a problem with the plane waves.
regards
Juerg
--------------------------------------------------------------
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: cp... at googlegroups.com
From: "Pierre Cazade"
Sent by: cp... at googlegroups.com
Date: 01/21/2020 11:39AM
Subject: Re: [CP2K:12796] 120680 atom system - CPASSERT failed
Dear Matt,
I can hardly scale the dimension of the cell as it is a crystal with 8 protein units (space group 96). The crystal has been tested and relaxed with a classical model to remove bad contacts with water molecules. As for the large cutoff, I don't know what you mean by that. Even for smaller systems (280 atoms), applying the methodology described in the howtos, I have a hard time identifying a satisfactory combination of cutoff and rel_cutoff providing good accuracy and convergence around these values (I mean by convergence that, by slightly changing one or the other cutoff, the accuracy remains the same). I usually find that very high values for the cutoff are necessary. In fact, in the present context I feel it is actually quite small but I need the system to run first before I can optimize them anyway.
Regards,
Pierre
On 21/01/2020 10:23, Matt W wrote:
Your large cutoff and huge cell have probably never been tested fully. Try running with just a single atom in the cell and see what happens, maybe scale the cell dimensions to see where it breaks.
Matt
On Monday, January 20, 2020 at 9:00:07 AM UTC, Pierre Cazade wrote: Hi Tiziano,
My bad. Please find below the most recent error message I got:
*******************************************************************************
* ___ *
* / \ *
* [ABORT] *
* \___/ CPASSERT
failed *
* | *
* O/| *
* /| | *
* / \ pw/pw_grids.F:853 *
*******************************************************************************
===== Routine Calling Stack =====
8 pw_grid_distribute
7 pw_grid_setup_internal
6 pw_grid_setup
5 pw_env_rebuild
4 qs_env_rebuild_pw_env
3 qs_env_setup
2 qs_init_subsys
1 CP2K
I tried to change the values of the CUTOFF and REL_CUTOFF, as well as
the values of EPS_DEFAULT and EPS_PGF_ORB. I tried the SZV and DZVP
MOLOPT-SR basis sets. I also changed NGRIDS from 4 to 6. Nothing seems
to get me through. I am looking forward to you insight.
Regards,
Pierre
On 20/01/2020 08:04, Tiziano Müller wrote:
> Hi Pierre,
>
> I haven't looked at your input (yet), but it is always helpful to also
> post the exact error message: CPASSERT usually shows at which source
> code line the error occurred, which helps to pinpoint the problem.
>
> Best regards,
> Tiziano
>
> On 17.01.20 18:08, Pierre-André Cazade wrote:
>> Dear CP2K users,
>>
>> I am trying to run an energy calculation for a 120680 atom biological
>> system. I run this on a node with almost 2TB of RAM. I tried with and
>> without LS_SCF and I always get a CPASSERT failure despite all the
>> tweeting of the threshold I may try. Could anyone point what is wrong
>> in my system and what I need to address to make it work. Please find
>> my input file at the following link:
>>
>> https://drive.google.com/open?id=1ibpjZuupTjNBCUAoSDQKEbh-v-YxqSCR
>>
>> Regards,
>> Pierre
>>
>> --
>> 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 c... at googlegroups.com
>> <mailto:c... at googlegroups.com>.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/cp2k/eae745b6-ac09-437f-b96c-391a35152340%40googlegroups.com
>> <https://groups.google.com/d/msgid/cp2k/eae745b6-ac09-437f-b96c-391a35152340%40googlegroups.com?utm_medium=email&utm_source=footer>.
>>
>
--
Dr Pierre Cazade, PhD
AD3-023, Bernal Institute,
University of Limerick,
Plassey Park Road,
Castletroy, co. Limerick,
Ireland
--
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 cp... at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/93fb0100-2a7d-4a79-be6b-4dbef3682c83%40googlegroups.com.
--
Dr Pierre Cazade, PhD
AD3-023, Bernal Institute,
University of Limerick,
Plassey Park Road,
Castletroy, co. Limerick,
Ireland
--
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 cp... at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/138a652b-eddc-64b6-fcf4-671c6430a16b%40gmail.com.
More information about the CP2K-user
mailing list