Re: [AMBER] cpptraj.OMP vs cpptraj

From: Daniel Roe <>
Date: Wed, 10 Jan 2018 09:20:43 -0500


Some actions don't scale perfectly - users are encouraged to do some
benchmarking on a handful of frames first to get a sense for how much
speedup you'll get for a given number of threads. The number of
threads used can be controlled with the OMP_NUM_THREADS environment
variable (this is mentioned in the manual). So run the analysis on a
few frames and set OMP_NUM_THREADS to 1, 2, 4, etc and see what
speedup you actually get. Note that cpptraj does not tend to benefit
from hyperthreading, so if the number of actual cores you have is 8
you probably won't see any speedup beyond that.

Note also that 'watershell' is one of two actions (the other is
'closest') that have been CUDA-accelerated, so if you have access to a
GPU you may want to try the CUDA build.


On Wed, Jan 10, 2018 at 6:06 AM, Saikat Pal <> wrote:
> Dear all,I have performed watershell analysis using cpptraj and cpptraj.OMP .In case of cpptraj time is very much less consuming than cpptraj.OMP (cpptraj is almost 5 times faster than cpptraj.OMP).For cpptraj.OMP I have used 16 core processor.Both the cases the results are same.Is it my installation problem ?? Please help me out.Thanks And Reargds,Saikat Pal
> _______________________________________________
> AMBER mailing list

Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
AMBER mailing list
Received on Wed Jan 10 2018 - 06:30:04 PST
Custom Search