On Thu, Apr 15, 2010 at 5:53 PM, David A. Case <case.biomaps.rutgers.edu> wrote:
> On Thu, Apr 15, 2010, Jason Swails wrote:
>>
>> Is the following common for trying AmberTools parallel with
>> DO_PARALLEL="mpirun -np 4" ?
>>
>> make: Target `test.parallel2' not remade because of errors.
>> 13 file comparisons passed
>> 15 file comparisons failed
>> 5 tests experienced errors
>>
>
> No: not common. Look at the diffs--if you are getting duplicated outputs,
> my guess is that you are really testing a serial version of NAB.
> Unfortunately, we don't have nab and nab.MPI, so unless you know your history,
> you could be trying a parallel test on a non-parallel nab.
Ah. Serial nab. I recompiled everything in serial after I was done
to check sander.APBS, and that's when I decided to poke around with
higher proc counts for ambertools. Didn't even think about that.
i still have that one weird nab failure that's printing out some value
like 10^200 in each parallel case using openmpi-1.4.1 on linux
(doesn't show up on Mac OS X).
Thanks!
Jason
>
> Of course, it could be something completely different, but all 27 tests pass
> for me using either 2 or 4 cpus.
>
> ....dac
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
--
---------------------------------------
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Apr 15 2010 - 15:30:02 PDT
