On Tue, Sep 4, 2012 at 3:16 PM, David A Case <case.biomaps.rutgers.edu> wrote:
> On Tue, Sep 04, 2012, Gustavo Seabra wrote:
>>
>> How can we compile a parallel version of ptraj?
>
> We no longer support parallel ptraj. As you have discovered, you would need
> to install and configure pnetcdf. If you still have old versions of Amber,
> you could get ptraj.MPI from there.
>
> But more to the point: why do you want this? Our experience was that for most
> applications, there were much better approaches to parallelism, such as
> processing parts of trajectories with simultaneous runs of the serial code.
> Plus, the cpptraj code is parallelized using openmp, which is often more
> effective than the parallelization model used in ptraj itself.
>
> If you describe the sorts of things you want to do that don't work well with
> serial ptraj, people on the list may have some suggestions.
>
> ...dac
Hi Dave,
Thanks for your reply. We're trying to cluster a large trajectory, and
clustering is taking forever for it. That's why I tried to compile
ptraj in parallel. Can cpptraj do clustering as well?
Gustavo.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Sep 05 2012 - 20:00:03 PDT
