Hi,
See notes in section 3.6.6 in the manual, under "parallel simulations";
because density matrix build and the matrix diagonalisation are not
parallelized, scalability is limited.
In my (limited) experience, ~50 QM atoms scale reasonably ok up to 8 CPUs,
but probably not much further. The same will be true (for the same reasons)
for other semi-empirical methods than AM1.
(See further also:
Walker, et al. J. Comput. Chem. 2008, 29, 1019. - similar statements are
made there)
--Marc
On 24 April 2012 16:12, Jacopo Sgrignani <sgrigna.sissa.it> wrote:
> Dear all
> could you give me infos about scalability of MD calculations with
> QM(AM1)/MM in amber12.
> Does anybody experience for this type of calculations on a blue-gene
> machine.
>
> Thanks a lot
>
> Jacopo
>
> --
> Jacopo Sgrignani PhD
> CNR-IOM-Democritos National Simulation Center
> c/o International School for Advanced Studies (SISSA/ISAS)
> via Bonomea 265,
> 34136
> Trieste
> Italy
> email: sgrigna.sissa.it
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Apr 24 2012 - 08:30:04 PDT
