[geeks] Intel's new bechmark idea

[Joshua D Boyd]

On Wed, Mar 06, 2002 at 01:27:31AM +0000, Kris Kirby wrote:

> What's a good general benchmark for comparative performance? All of the
> "old" machines I have read about are quoted in MIPS.

FLOPS is kinda informative.  SPEC numbers are informative, if you work looks
like the work SPEC tests do, assuming that the machine was ever good enough
to run the tests.

There aren't really any good general benchmarks, but understanding what the
benchmark does on the machine goes a long way to gaining usefull information
from it.

So, how are FLOPS figured out anyway?  The number of times that a FP chip
can do the floating equivelent of adding two hardwired zero registers and 
writing the result to the same?
 
> [0]: I can't afford a serious UNIX machine right now. :(

Can you afford shipping on a 715/50?  On a pizza box Quadra?  On a lunch box
sun?  If so, then I can hook you up with a serious unix machine.  They really
are pretty easy to come by.

-- 
Joshua D. Boyd