wbmw, you seem to be about 2000 posts behind (your post 33441 responded to 31615). We had extended discussions on this topic with some very interesting insights offered by the board. Bottom line, although there are many design differences between Dothan and Prescott, junction leakage generated by increased frequencies was identified as the element which can cause power consumption to become non-linear as you push the envelope on clock frequencies. (Some types of leakage actually go down as frequency goes up, but this decline is bounded as leakage can not go below zero.)
It is not known (here) whether the design of the junctions is different between Prescott and Dothan. However, my initial statements have held up very well under scrutiny (and chipguy really did want to disprove it) - you cannot draw negative conclusions about Dothan from Prescott, and you cannot draw positive conclusions about future Prescotts from a low power Dothan. The leakage behavior of junctions at such different frequencies makes any such conclusions unwarranted.
(On a tangent discussion we revisited the famous Intel engineer's surface of the sun warning and quickly debunked it.)
So, I am no hardware engineer but my empirical observations seem to be pretty good in this case.
Edit: I see from your answer to Paul that you really did misinterpret my original statement. I was trying to indicate that Dothan may very well scale better than Prescott because it is starting at a much lower frequency - I was arguing that Dothan may be a killer product even if Prescott is a performance scaling dud!
Dothan, on the other hand, won't get near 4GHz, but there's no reason to assume that it will be process limited, just because the common myth is that Prescott is process limited. This also applies to the FUD that gets tossed around here about Dothan being higher power than Banias just because Prescott is higher power than Northwood.
A little defensive, aren't we?
:)
Market Data 
Markets 
AI
