Re: [AMBER] trajectory of minimization using generalized Born

From: Jason Swails <>
Date: Mon, 23 Mar 2015 14:36:20 -0400

> On Mar 23, 2015, at 7:47 AM, Edwin Helbert Aponte Angarita <> wrote:
> Dear All,
> I have installed AmberTools14 (and applied all updates and passed _all_
> tests) in my home directory on this cluster and used its sander to run the
> minimization using 16 processors and got again "ERROR IN SETPAR() upon atom
> distribution"
> Thus, as pointed out before, this error should be because the peptide is
> very short and there is not enough workload to spread across 16 processors,
> shouldn't it?

I think so. Minimizations also use different parallelization scheme than MD (although not *that* different). So it’s possible you won’t see this limitation in MD.

In any case, 16 CPUs for a 34-residue peptide in implicit solvent is most likely overkill. I would stick to 8 (or maybe even 4 -- you would have to benchmark to find the optimum number of CPUs for your system).


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Mon Mar 23 2015 - 12:00:02 PDT
Custom Search