[CP2K:29] latest modification to improve 2d particle distribution

Teodoro Laino teodor... at gmail.com
Wed May 2 04:44:41 UTC 2007


The last update is mostly focused on avoiding that a molecule is  
totally present on 1 cpu... (this was our default scheme.. 2d  
distribution was molecular based)
This means that now you can use a molecular topology in QS without  
any terrible load balance (for QS runs the default topology was
1 atom= 1molecule)..
You may observe some speed up with the new implementation, but in  
most of cases it should be pretty small..

So unless you were explicitly using the keyword CONNECTIVITY GENERATE  
for QS runs or providing PSF files for QS runs
don't expect any relevant increase of performance..

cheers
teo

As Joost pointed out:


On 1 May 2007, at 21:55, Nichols A. Romero wrote:

> Hi,
>
> Question about the latest updates by vondele. Are the recent  
> changes in the parallel 2d particle distrubution algorithm
> aimed at improving the load balancing for increased scalability on  
> massive-parallel platforms?
>
> Thanks,
>
> Nichols A. Romero, Ph.D.
> 1613 Denise Dr. Apt. D
> Forest Hill, MD 21050
> 443-567-8328 (C)
> 410-306-0709 (O)
>
> >




More information about the CP2K-user mailing list