Dear all,
I haven't seen this on the mailing-list so far, but may be I missed it.
I'm trying to test the CUDA version of Amber (amber11 and amber12). I've
tested different combinations and I have a problem just for one: in the Amber11
test suite (available at
http://ambermd.org/gpus/benchmarks.htm), if I replace
ntt=1 by ntt=2 in the PME/FactorIX_production_NVT directory, pmemd never stops
when using pmemd.cuda_SPDP.MPI. Other pmemd versions are OK: pmemd.MPI (8
cores) and pmemd.cuda.SPDP with intel 12 (or gnu 4.4.6) + mpich2-1.4.1p1 +
nvcc 4.2. Same problem occurs either with amber 11 and amber 12: only with
ntt=2 (ntt=1 runs fine). I've tested others systems (mine are about the same
size as the FactorIX test) and the problem stays.
Does anybody have the same problem as I have? Am I missing something here?
Sincerely,
G. Monard.
P.S.: ./test_amber_cuda_parallel.sh only reports one minor diff.
--
____________________________________________________________________________
Prof. Gerald MONARD
Theoretical Chemistry and Biochemistry Group
SRSMC, Lorraine University, CNRS
Boulevard des Aiguillettes B.P. 70239
F-54506 Vandoeuvre-les-Nancy, FRANCE
e-mail : Gerald.Monard.univ-lorraine.fr
tel. : +33 (0)383.684.381
fax : +33 (0)383.684.371
web : http://www.monard.info
____________________________________________________________________________
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jun 07 2012 - 14:30:03 PDT
