Re: Athlon 1.2Ghz results == P4 1.5

[R Clint Whaley <rwhaley@cs.utk.edu>]

>What are the *numerical* tradeoffs between SSE and 3DNOW2?? :))

I think Peter's response answered the question, but the question seems to
imply the timings I sent had something to do with 3DNow!.  They don't:
all Athlon timings were using Julian's x87 code.  If I had compared to
3DNow!, the Athlon would have a theoretical peak of 4*mhz, just like the
P4.  I pretty much never do anything with 3DNow!, since it is not numerically
stable.

Regards,
Clint