tyan tiger i7501
Robert G. Brown
rgb at phy.duke.edu
Thu Feb 20 13:39:08 PST 2003
On Thu, 20 Feb 2003, Dr. Rochus Schmid wrote:
> since the 760MPX boards dont have GigE on board, the price difference between
> AthlonMP and Xeons seems to be roughly the pure processor price diff. due to
> the heat problems of the Athlons and SSE2 i tend towards the Xeon.
> am i wrong with this?
Not wrong, but remember that a benchmark is worth a thousand tendencies,
especially of your own code.
Oddly enough, my own favorite application runs MUCH faster on an Athlon
than on a P4. It even beats AMD's extravagant pseudoclock scaling (as
in an Athlon 1900+ beats a 2.4 GHz P4). Other (possibly more
memory-local) applications go the other way entirely.
So if at all possible, test your code on alternative node prototypes,
don't rely on hearsay or guestimated performance.
It is, however, worthwhile to look for evidence, anecdotal or not, of
problems with any given motherboard and CPU. Even a better performing
system gets less work done if it crashes all the time. As in, the Tyan
Robert G. Brown http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb at phy.duke.edu
More information about the Beowulf