[MITgcm-devel] Hi-res cube stats

Dimitris Menemenlis menemenlis at sbcglobal.net
Wed Jan 12 02:53:54 EST 2005


> Where can I find stats on the cube run. e.g. processor type,

Alistair, there is no such stats in one place.  Description of Columbia 
processors: http://www.nas.nasa.gov/Resources/Systems/columbia.html

> cpu-hours/model year,

17 hours 19 minutes on 216 CPUs / model year
Chris and Jim Taft have achieved close to
10 yr/day on 512 CPUs without I/O, seaice, and KPP

> disk foot-print for i/o, etc..

domain is 510x510x6x50, so
6 Mb per 2-D field
312 Mb bytes per 3-D field
187 Mb for sea-ice pickup and
5 Gb for basic pickup
212 Gb for pretty basic output for a year

D.

P.S. You never replied to question re viscC4Leith
What's appropriate thing to do if I change model
resolution or time step?

-- 
Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
Jet Propulsion Lab, California Institute of Technology
MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099
tel: 818-354-1656;  fax: 818-393-6720



More information about the MITgcm-devel mailing list