Andreas Svrcek-Seiler wrote:
> Hi all,
> I've just gotten my hands on 2 dual quad core machine with 2.33GHz Xeons.
> With NAB and openmp-parallelization, I get a nearly 8-fold speed increase
> for implicit solvent (GB) runs and the big example (~18000 atoms)I just
> ran.
> For the "cox2" benchmark (100 timesteps, cut=18, rgbmax=17) I get
> 1 CPU 742s
> 2 CPU 375s
> 4 CPU 190s
> 8 CPU 97s
>
> So despite the architecture's weaknesses, 8 cores are about 7.6 times
> faster that one core. A quick test with two independent jobs on four
cores
> each also showed that they don't seem to interfere with each other.
>
> ...just 2 cents more
>
> Andreas
>
Thanks for the info. Just to make sure I have this right, this is
for NAB molecular dynamics, not AMBER, correct?
Is this the regular 5.1.2 version of NAB? I had forgotten that
NAB MD was parallelized, I guess.
Wonder why the simulation engine in NAB seemingly parallelizes a
little better than AMBER? If I am not making too much out of
this.
Bud Dodson
--
M. L. Dodson
Email: mldodson-at-houston-dot-rr-dot-com
Phone: eight_three_two-56_three-386_one
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Aug 15 2007 - 06:07:28 PDT
