-23 Mar 97  17:40:02-
=Niels Jonker babbled to Mike Phillips about Re: MHZ VS MFLOPS=
 NJ> If you're working with a Pentium processor this isn't exactly right
 NJ> True, almost all floating point instructions do not pair ( you can
 NJ> only pair FXCH with stuff like FADD,FSUB,FMUL etc.. ) but instructions
 NJ> can overlap, this means that the next instruction can be started while
 NJ> the first is still being executed.
I didn't know that.
 NJ> Ofcourse to get the most out of this your code must follow a few
 NJ> rules,( the second instruction does not overlap if it needs the result
 NJ> of the first, a FMUL can't start the clockcycle after another FMUL,
 NJ> etc... ) 
Well, that goes without saying.
 NJ> This means the number of FLOPS (for the Pentium processor) depends a
 NJ> lot on how you write your code / the intelligence of your compiler
 NJ> and so to simply divide the Mhz by the number of clockcycles of a
 NJ> single floating point instruction takes, will you not give you a very
 NJ> meaningfull result.
Even if the Pentium operated as I described, the performance of your
code would still be related more to the way you write your code than to
the performance of the processor's FPU.  If you're dealing with a
multitasking situation, the processor's performance is almost
irrelevant.  The processor power that your app sees can vary greatly
from run to run.
Mike Phillips
INTERNET:  phil4086@utdallas.edu
... Learn the basics of Clintonomics.  Call 1-900-SUK-U-DRY
--- FMail/2 1.22
---------------