[QE-users] 'Cholesky failed in aceupdate.' when using npools

[Lars Blumenthal]

This is for future reference: With Paolo's help, I found out that I had 
to recompile QE.
At first I was running the pwscf v.6.1 that was precompiled on ARCHER 
and in that case the parallelisation over k-points didn't work when 
using hybrid functionals.
However, it does work with the pwscf version (also v.6.1) I compiled myself.

Many thanks to Paolo and best wishes,

Lars
PhD Student
EPSRC Centre for Doctoral Training on Theory and Simulation of Materials
Imperial College London


On 30/05/18 12:36, Paolo Giannozzi wrote:
> I made a quick test on a reduced version of your job and found no 
> problems, but the original job requires a larger machine and I have no 
> time to work on it now.
>
> Paolo
>
> On Wed, May 30, 2018 at 11:58 AM, Lars Blumenthal 
> <lars.blumenthal11 at imperial.ac.uk 
> <mailto:lars.blumenthal11 at imperial.ac.uk>> wrote:
>
>     Does anyone have any advice/feedback?
>
>     Many thanks,
>
>     Lars
>     PhD Student
>     EPSRC Centre for Doctoral Training on Theory and Simulation of
>     Materials
>     Imperial College London
>
>
>     On 25/05/18 17:03, Lars Blumenthal wrote:
>>     Hi everyone,
>>
>>     I am trying to do scf calculations using the HSE functional with
>>     PWSCF v.6.1 (svn rev. 13369).
>>
>>     When I don't use parallelisation over k-points, i.e. when I don't
>>     specify npools, the calculation runs successfully.
>>     However, as soon as I try to make use of npools, the calculation
>>     crashes with:
>>
>>     DPOTRF exited with INFO= 7
>>     Error in routine DPOTRF (1):
>>     Cholesky failed in aceupdate.
>>
>>     I have attached the corresponding output file. Previously I have
>>     had the same issue with another compound but in that case npools
>>     = 2 actually did work and it only crashed with the above error
>>     when npools > 2. So it's not necessarily that the parallelisation
>>     with npools doesn't work at all.
>>
>>     Not using the ACE algorithm makes the calculation painfully slow
>>     so I'd like to avoid that. Do you have any advice on how to
>>     optimise the parallelisation of hybrid DFT calculations in general?
>>
>>     Many thanks and best wishes,
>>
>>     Lars Blumenthal
>>     PhD Student
>>     EPSRC Centre for Doctoral Training on Theory and Simulation of
>>     Materials
>>     Imperial College London
>>
>>
>>
>>     _______________________________________________
>>     users mailing list
>>     users at lists.quantum-espresso.org
>>     <mailto:users at lists.quantum-espresso.org>
>>     https://lists.quantum-espresso.org/mailman/listinfo/users
>>     <https://lists.quantum-espresso.org/mailman/listinfo/users>
>
>
>     _______________________________________________
>     users mailing list
>     users at lists.quantum-espresso.org
>     <mailto:users at lists.quantum-espresso.org>
>     https://lists.quantum-espresso.org/mailman/listinfo/users
>     <https://lists.quantum-espresso.org/mailman/listinfo/users>
>
>
>
>
> -- 
> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
> Phone +39-0432-558216, fax +39-0432-558222
>
>
>
> _______________________________________________
> users mailing list
> users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20180530/a8110de6/attachment.html>