[gmx-users] Apples-to-Apples comparison for dpcc benchmark

[David]

On Wed, 2005-06-15 at 13:53 -0500, Nathan Moore wrote:
> I want to make sure that I understand the meaning of the GROMACS benchmarks.
> 
> When I look at the logfile from a 32processor dpcc run, I see the
> following text near the end of the file:
> 
> 
>         M E G A - F L O P S   A C C O U N T I N G
> ...
>                NODE (s)   Real (s)      (%)
>        Time:   1464.000   1464.000    100.0
>                        24:24
>                (Mnbf/s)   (GFlops) (ps/NODE hour) (NODE hour/ns)
> Performance:     79.584      2.878     24.590     40.667
> 
> 
> Do I understand correctly that the number to compare is total ps/day,
> 24.590 ps/Node-hour * 24 hours/day * 32 processors = 18,888 ps/day ?
No I'm sorry...
24.59 * 24 = 600 ps/day.
which is roughly the same as 4 Xeons at 4 times the clock.

> 
> Additionally, the published parallel dpcc benchmarks use single precision
> and grompp options -sort and -shuffle, 5000 timesteps?  Are there any
> other optimization options I should be aware of when making a comparison?
> 
This is all.


-- 
David.
________________________________________________________________________
David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596,          75124 Uppsala, Sweden
phone:  46 18 471 4205          fax: 46 18 511 755
spoel at xray.bmc.uu.se    spoel at gromacs.org   http://xray.bmc.uu.se/~spoel
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++