JC heeft wat info gepost over de PIII en Intel's PII variant voor mobiel gebruik. Zijn conclusie: de mobiele PII is door z'n snellere cache waarschijnlijk sneller dan de PIII! Intel blijft me verbazen...
In defense of Intel: You know, I'm usually hard on the fellows, but there's one thing I have to back them up with lately. The new Pentium II PE mobile processor, earlier known to us by the cooler codename "Dixon", packs in 256k of on-chip L2 cache, instead of 512k of much slower L2 off the main silicon.
As a result, these notebook processors pack more of a punch (eg..they're faster) than their desktop counterpart Pentium II. In fact, they're likely even faster than the Pentium !!!. I'm starting to hear complaints here and there that Intel is holding back the design from the desktop market, and people are hence getting screwed with slower chips.
I want to just take a couple seconds aside to explain why these faster chips are only being offered for notebooks. Basically, the size of the die of the Dixon is rather immense, being probably about 35% larger than the Pentium II die and over twice the size of the K6-2. Because of a larger size, not only can Intel fit fewer of the chips onto each production wafer (hence they can't make as much), but the percentage of bad chips shoots up as the size goes up.
So, the yields of the PE chip are likely not capable enough for the desktop market. Intel can only relegate it to a market with higher profit margins and less demand -- the notebook market. Basically, Intel will port these chips over to the desktop market in the second half of this year (with KNI/SSE added, of course). This will happen after Intel goes through with their die shrink from 0.25um to 0.18um (which'll basically have the effect of chopping the die sizes in half). The 0.18um chips, by the way, are currently codenamed Coppermine.