> a system I found at the noaa had four processors and was rated at 200 MFlop/s
> http://www.ncep.noaa.gov/NCO/NCCF/NCEP-HPC/tsld005.htm
>
> ORNL eval of opteron showed 2*Mhz. their test was on a 1800mhz or 3600 MFlop /
> sec
> http://www.csm.ornl.gov/~dunigan/opteron/
>
> sounds like the opteron might win.
While the Opteron might in fact win (let's face it - the Js were not
really supercomputers), that comparison simply does not hold water.
Mflops are basically meaningless unless they can be sustained.
William Donzelli
aw288_at_osfn.org
Received on Mon Mar 08 2004 - 22:04:03 GMT