Chris, re: I think a good indicator will be the voltage and OC'ability of the first 65nm K8s. As I have said, if the wires has been scaled significantly, the 65nm K8 will be faster than 90nm K8 from the start. If voltage really has dropped to 1.25V, that either means the 65nm transistors are very different from 90nm, or that AMD is trading the drive strength from next gen. strain, for lower power.
The first possibility could be good or bad. It's good if their transistor design simply calls for lower voltage. It's bad if they leak, and voltage reduction is to keep leakage in check.
The second possibility either suggests that AMD is going for lower power instead of performance, or that wire scaling has significantly reduced the need for transistor drive strength.
Since AMD is in dire need of performance, I don't see why they would go for lower power. ...
If AMD has 3GHz dual cores at 65nm release, I would guess they pulled it off. Otherwise, I will wait and see.
The Inquirer is showing a somewhat dire picture for initial 65nm processors based on initial samples. They say the design was targeted at 1.35V but that the samples are 1.4V, both being significantly higher than the 1.25 you said would be nice. So although intial 65nm products will use relatively high voltages the article goes on to say that early next year AMD will use a newer 65nm process that will allow dropping voltages and power dramatically leading the way to QC parts. uk.theinquirer.net
Basically the initial 65nm process will allow to continue to shift to dual core processors reasonably fast (but won't allow for 2.8GHz I think), while the second 65nm process looks to support the amount of more performance and lower power that we just talked about.
I was slightly curious about the short succession of processes when I first read it.
Regards,
Rink |
