Libinit compilation

Matt W mattwa... at
Fri Feb 17 21:48:40 UTC 2017


just a general note: you do need to be quite careful running performance 
tests. It can easily be that your test is not representative of the systems 
you are are going to run in practice.

Whilst all these codes will run this C-240 system. It may be that their 
ability to use symmetry, run 3D periodic systems, scalability to large 
number of processors and other factors, such as local knowledge to get 
help, are going to vary a lot. Depending on what you need for the whole of 
your project, just seeing which one runs this system fastest might not be a 
very sensible cost function.


On Friday, February 17, 2017 at 6:07:22 PM UTC, Reza wrote:
> Dear my friends
> Thank you very much for your kindness and time to guide me. 
> I tested ORCA, Gamess-US, Gaussian09, NWChem, CP2K, FHI-aims for PBE0 
> (hybrid) calculation of C240 with nearly similar parameters (same basis 
> sets, grid, ...). The Fastest Code is Gaussian till now. It is welcomed an 
> optimized input to test CP2K speed. total energy of SCF-converged 
> calculation should be the same (-9143 Hartree).
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the CP2K-user mailing list