One final P4 spec number from billybob @ aces' message board. All numbers for 1.5ghz part
Specint2000 502
SpecFP2000 525
By comparison, the best posted scores for the P3 at spec.org are for 1ghz cumine, on an OR840 (dual rambus channel) system, Intel 5.0 compiler.
CINT2000 438 base, 442 peak
CFP2000 327 base, 335 peak
Again, the int scores don't look very impressive, 14.6% increase from a 50% clock increase. FP looks better, a 60.5% increase from a 50% clock increase. I think a fair amount of the FP jump comes from the faster bus, the spec FP things seem to be getting stream-like.
The couple posted Athlon scores are FP only, and aren't on the face of it competitive, till you go back and compare equivalent compilers. Again, it's nice that Intel has a whole compiler operation just to generate good spec numbers, but I don't think that'd be the best way for AMD to spend its money.
Cheers, Dan. |
