> I know we're just speculating here because there are no published benchmarks from Intel but that INT number of 502
> would be very close to the highest score ever posted, 511. That's impressive to me.
It is impressive right now. But in Q1, when the processor starts approaching volume availability (imho, and that's kinda important in my books), I'm not so certain that it will be as impressive as it seems now.
The only real problem here is that AMD is too wussy to post up specint scores. I'm going to beat up on my local rep now to ask about that.
Actually, my other problem is that I don't know the ramp of the two processor manufacturers. I know that Intel is shooting for a groovy 2.00GHz in Q2, but AMD's strategy is a bit up in the air. Also, there are a number of improvements in the upcoming "Mustang" revision of the Athlon which could have a material effect in average ipc for specint2000, but I do not know enough to be certain.
I agree with what you've been saying, that performance per process is what's important in examining the design, not performance per clock. Well, that and to some extent performance per price, that sort of thing. But I am very interested in seeing Willamette's max spec performance in the P858 (wait, was it P860? I always get fracking confused about which process is which -- assistance would be welcomed here).
Still ... if the Thunderbird variant of the Athlon is anywhere remotely near the specint2000 performance of the PIII (I will check with my sources at AMD, and I will look up some third party Tbird/Cumine spec2000 comparisons), then it may be a closer match than you suspect, even if the Mustang offers no higher specint2000 ipc than Thunderbird.
-JC |
