Just to complement on the answers you have already received. In our
experience with large numbers of processors, it is usually a good idea
to leave some cores idle during the calculation. With 8-core nodes,
for example, we get better throughput by allocating the whole node and
only using 1-2 MPI processes per node. Still, the best thing is for
you to run some test calculations with your own system.
Gustavo Seabra
Postdoctoral Associate
Quantum Theory Project - University of Florida
Gainesville - Florida - USA
-------------------------------------------------------------------------------------------
Q: Why do mountain climbers rope themselves together?
A: To prevent the sensible ones from going home.
On Sun, Jun 7, 2009 at 11:19 PM, Catein Catherine<askamber23.hotmail.com> wrote:
>
> Dear Sir/Madam,
>
>
>
> Could you mind to share your experience with me about the parallel efficiency of amber10. It seems to me that we can run AMBER10 efficiently in 8 processors on the same node.
>
>
>
> However, the jobs may have problem to run on 2 nodes with 8 processors each.
>
>
>
> Is it better to do the sander within the same node? Could you pls kindly share your experiences with me?
>
>
>
> Best regards,
>
>
>
> Cat
>
> _________________________________________________________________
> More than messages–check out the rest of the Windows Live™.
> http://www.microsoft.com/windows/windowslive/_______________________________________________
> 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 Jun 09 2009 - 01:09:46 PDT
