how to improve CP2K scaling for small systems

[Matt W]

Also, could you post the timings information at the end of completed runs? 
I.e. the

 SUBROUTINE                       CALLS  ASD         SELF TIME        TOTAL 
TIME
                                              AVERAGE  MAXIMUM  AVERAGE 
 MAXIMUM
 CP2K                                 1  1.0    0.056    0.064 9136.160 
9136.192
 geoopt_bfgs                          1  3.0    0.086    0.110 9135.135 
9135.363
 cp_geo_opt                           1  2.0    0.000    0.001 9135.135 
9135.363
 cp_eval_at                          10  4.0    0.000    0.001 8805.294 
8805.323
 qs_forces                           10  5.0    0.356    0.686 8805.275 
8805.301
 qs_energies_scf                     10  6.0    0.010    0.022 7714.098 
7714.144
 scf_env_do_scf                      10  7.0    0.001    0.004 7560.596 
7560.621

etc. bits at the end.

It would be helpful to see if FFTs or some other routines are unexpectedly 
dominating the calculation, which might help track down what is going on.

Matt

On Monday, March 17, 2014 10:28:27 AM UTC, Florian Schiffmann wrote:
>
> I know it's a stupid question but what's about memory of the machine? The 
> system seems quite large, could it be that you run out of memory and use 
> the swap? 
>  
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140317/f7d9efb9/attachment.htm>