Newsgroups: sci.physics.plasma
From news@cornell.edu Sun Jan 8 06:39:48 1995
From: rilee@astrosun.tn.cornell.edu (Michael L. Rilee)
Organization: Laboratory for Plasma Studies
Subject: Re: speed of cray vs unix work stations

From: brooker@toka.ireq-ccfm.hydro.qc.ca
>The million dollar question is how much slower is a powerful UNIX work
>station than a CRAY? A factor of 10? 100? 1000?

Depends on your needs (cpu/disk speed, memory, throughput. . .) One
thing, though, if you need long running times, you might get a better
turnaround time with your own machine because you won't be in the long
queues. My code is relatively small, and restarts well, so I putter
along in the 15-min queues, and chain the batch jobs. That goes rather
quickly, and I get a much better cpu-time/wall-clock ratio.

Heard at a colloquia on finding pulsars in radio data:

For the amount of money we were allocated in supercomputing
time, we could have bought a good workstation, and we would
have finished analyzing the data in one year, as opposed to
two.
Paraphrased, of course. . . :)

Regards,

Mike Rilee
--
Laboratory of Plasma Studies
369 Upson Hall
Cornell University
Ithaca, New York 14853-7501
Phone: 607 255 4723
Fax: 607 255 3004

>obviously, IBM DOES NOT SUPPORT any use of these connectors.
>
>i politely asked, if i pay for hardware, shouldn't i expect all the parts
>to work. the answer was a clear "no"....
(from a certain mailing list. . .)

The birds sing just before sunrise. Did you hear them today?

Smile!