Re: [AMBER-Developers] near-final testing

From: Daniel Roe <daniel.r.roe.gmail.com>
Date: Mon, 12 Apr 2010 11:04:56 -0400

On Mon, Apr 12, 2010 at 9:04 AM, case <case.biomaps.rutgers.edu> wrote:

> 5. Dan: as ptraj expert, have you been trying the ptraj_comprehensive
> test on 8 threads? (See error reported by Ben on the wiki.)
>

The issue that Ben reported was that with 8 threads the ptraj_comprehensive
test case is failing, specifically by not processing the last 4 frames of
the trajectory. The reason for this is the current limitation of ptraj.MPI
that the # of frames must be divisible by the # of processors. This is from
ptraj output on said test with 8 threads:

WARNING in ptraj(): Number of sets to be processed is not a multiple of
offset * number of ranks for input file trajectory.mdcrd.
This is a current limitation of parallel ptraj. Stop frame being set to: 96

And there go the last 4 frames. So this test should probably be restricted
to either 2 or 4 processors - guess you could go up to 10 as well, but why
would you want to?

I'm attaching a patch that will modify the Run.comprehensive test script to
check the number of threads.

-Dan

PS - Currently checking cygwin...

-- 
-------------------------
Daniel R. Roe
Postdoctoral Associate
SAS - Chemistry & Chemical Biology
610 Taylor Road
Piscataway, NJ   08854



_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers


Received on Mon Apr 12 2010 - 08:30:04 PDT
Custom Search