On Sat, Dec 08, 2012, Jason Swails wrote:
>
> The fastest you can get is using pmemd.cuda. Or using a GB model without
> the neck correction (e.g., 2, 5, or 1).
This is correct if you really want no cutoff. The gb implementation in NAB
has some features that are attractive for very large systems: it uses a
reasonably efficient pair list, and there is also the Hierarchical Charge
Partitioning (HCP) option, which can be quite effective for large molecular
assemblies. You will need to run timing experiments with your systems to find
good compromises between accuracy and efficiency.
...dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Dec 08 2012 - 19:30:02 PST
