[gmx-users] (no subject)

[David van der Spoel]

On Wed, 2004-06-30 at 14:07, Fiona Reid wrote:
> Dear Gromacs users, 
> 
> Does anyone know how the total number of atoms is 
> divided between processors? 
> 
> I have a system with 97787 elements, for 4 and 16 
> processors the division of atoms/nodes seems to be ok:
> 
> Division over nodes in atoms:		(for 4 processors)
>  24447 24447 24447 24446
> 
> Division over nodes in atoms:		(for 16 processors)
>   6137  6118  6118  6110  6108  6108  6108  6108  6108  6108  6108  6111
> 6108  6111  6108  6110
> 
You probably have a molecule with 6137 atoms, which can not be split if
you use constraints. Use either constraints=h-bonds with a time step of
1 fs, or (more economically) use fewer processors.


> However, for a 32 processor run or any run with more 
> than 32 processors the division of atoms per node is no 
> longer equal, e.g. 
> 
> Division over nodes in atoms:		(for 32 processors)
>      0  6137  3036  3036  3036  3036  3058  3057  3057  3057  3057  3057
> 3057  3057  3057  3057  3057  3060  3057  3060  3057  3060  3057  3060
> 3057  3060  3057  3060  3057  3060  3057  3059
>  
> The first processor always gets zero atoms, the second 
> processor always gets 6137 atoms and the remaining N-2 
> processors get approx 91650/(N-2) atoms. 
> 
> Is this correct? how large an effect will the uneven 
> atoms/node distribution have on the runtime? 
> 
> Regards, 
> 
> Fiona Reid
> 
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
-- 
David.
________________________________________________________________________
David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596, 75124 Uppsala, Sweden
phone:  +46 18 471 4205  fax: +46 18 511 755
spoel at xray.bmc.uu.se    spoel at gromacs.org   http://xray.bmc.uu.se/~spoel
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++