Paul, Re: Anyone think a 1.4GHz Willamette would hold a candle to a 1.4GHz PIII?
Very interesting comparison. In fact, the Athlon 64 has many similarities. Overall, AMD improved the Athlon 64 in IPC quite a bit. At the very least, it receives a small advantage over an Athlon XP at the same clock, and at the most, it gains by leaps and bounds. Of course, in aiming for IPC, AMD chose several paths that has so far limited the second most important variable to performance: frequency.
Willamette had all the performance it needed above the Pentium III, but it needed to grow to a certain frequency before it could top the Pentium III in every single way. Eventually, it reached that point around 1.8GHz or so. No other .18u Pentium III on any application could outperform a 1.8GHz Willamette.
AMD now has a product that takes a leap forward in IPC and a leap backward in frequency, which brings them to the same problem. Athlon 64 needs to grow into its target frequency, such that it can outperform a .13u Athlon XP in every single way. I don't think this is going to happen until 2.2GHz or so, even though some apps will have no problem outperforming.
Willamette was similar. Some applications, such as Quake III and Windows Media Encoder, outperformed the Pentium III quite easily, even from the beginning, and even at the paltry frequency of 1.3GHz. The reason why Willamette left such a bad taste in the mouths of the enthusiast community was that it couldn't outperform Pentium III in such applications as Winstone, even at higher frequencies, like 1.6GHz.
Athlon 64 will leave a similar bad taste if it cannot outperform the fastest Athlon XP, and this will happen if AMD is still stuck at lower frequencies by the time it launches. I think AMD realizes this, which is why they pushed out the launch to September to give them extra time to hit 2.2GHz or so. In the end, AMD's end users may have nothing to worry about, but even though this is a solution, it is not the ideal. AMD was supposed to have Athlon 64 on the market more than a year ago, and now they have another 5 months.
The conclusion that I and others have been trying to make is that frequency is AMD's main limiter. Athlon 64 is not some magic technology that will allow them to "Hammer" the competition. It requires as much frequency as the Athlon XP to maintain a forward advantage, or else you will only have an improvement in some apps, and a loss in others.
The other problem comes when you consider the appropriate PR rating for such a chip. If you have a 2.0GHz Athlon 64 that performs like a 2500+ in some apps and a 3400+ in others, will you be fair and call it a 2900+? So far, AMD has indicated that they will call it a 3400+, and I think that's stretching the truth far more than the market will handle. Realistically, however, they still have several months to figure out their marketing strategy. I suppose I'll have to reserve my judgement until then.
News 
Market Data 
Discover 
AI
