Charm++ parallel programming system

[Phil Miller <mille121 AT illinois.edu>]

On Mon, Mar 21, 2011 at 07:17, Kannan Govindarajan
<kannan.gridlab AT gmail.com>
 wrote:
> I am G.Kannan, working as research associate in MIT, Anna University. I am
> trying to test  our Care Resource Broker performance using NAMD application.
>
> I have successfully tested the NAMD application using charm++.
>
> I am using NAMD-2.6 and mpich-1.2.6 version. Is it possible to run the namd
> application using mpich?  and PBS as local scheduler ?

Before going any further, I'll note two things:
1. NAMD 2.6 is rather old, and was superseded by NAMD 2.7 several months ago
2. This (ppl@) is not the mailing list for NAMD. That list is
namd-l AT cs.uiuc.edu
 (CCed). The NAMD developers also have an excellent
list of problem-solving resources [0]

Now, to the core of your questions:
NAMD is implemented in Charm++. Charm++ has an MPI backend, but we
don't recommend its use, since the machine-specific backends tend to
perform better. Even when using a non-MPI version of Charm++, it is
straightforward to run Charm++ applications in an MPI-centric
environment, by telling Charm++ to take advantage of the available MPI
job launcher. This is described in the manual [1], conveniently linked
from [0].

If you have suggestions for how we can improve our documentation so
that it would have answered your question, please let us know.

-- 
Phil

[0] http://www.ks.uiuc.edu/Research/namd/bugreport.html
[1] http://charm.cs.uiuc.edu/manuals/html/install/4_1.html