[gmx-users] Re: Itanium 2

[Anton Feenstra]

Hi Peter,


A bit belated but I was looking into benchmarks myself now anyway.

We run on an IBM regularly, and for the Power4 nodes I get this for 1 CPU:
                (Mnbf/s)   (MFlops) (ps/NODE hour) (NODE hour/ns)
Performance:     16.581    544.615     20.639     48.452

And this for 6 CPU's:
                (Mnbf/s)   (GFlops) (ps/NODE hour) (NODE hour/ns)
Performance:     60.922      2.003     76.215     13.121

i.e. scaling ~62%, which isn't at all bad for a ~40k atom system
(455 residue protein, heme group and ligand in ~12k molecule water box).

For comparison, this sytem runs on an AMD 1GHz like this:
                (Mnbf/s)   (MFlops) (ps/NODE hour) (NODE hour/ns)
Performance:      9.914    327.066     12.871     77.694

And this on an SGI R12k 270MHz:
                (Mnbf/s)   (MFlops) (ps/NODE hour) (NODE hour/ns)
Performance:      4.188    140.747      5.217    191.694


-- 
Groetjes,

Anton
  _____________ _______________________________________________________
|             |                                                       |
|  _   _  ___,| K. Anton Feenstra                                     |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|(   |   )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands   |
| \_/ \_/ | | | Tel: +31 20 44 47608 - Fax: +31 20 44 47610           |
|             | Feenstra at chem.vu.nl - www.chem.vu.nl/~feenstra/       |
|             | "If You See Me Getting High, Knock Me Down"           |
|             | (Red Hot Chili Peppers)                               |
|_____________|_______________________________________________________|