90nm paths in the same substrate is going to have higher resistance than 130nm paths will (I think you know that). BTU/Watt will be higher (I think you know that as well). Prescott cores do not have a significant transistor count over the Northwood core, yet power consumption is approximately 40% higher at the same clocke rates.
You simply cannot say power consumption is directly related to transistor count, but I also figure you are using general statements as well. While transistor count is a factor, it is not the only gauge of power consumption.
Take a look at GPU's today. They are using 3 times as many transistors as they were 3 years ago, with higher clock rates, and yet using less power (ATI's 800XL is a good example. It uses less power than an NVidia GF4Ti4600, but has 3.2 times the transistor count and is running a higher clock rate). Funny thing is, ATI has licensed a good deal of thier process from Intel.
If Intel sticks with that design, which they have not shown they are going to abandon, then they need a low-k process to reduce the power consumption to reasonable levels.
Given that Prescott cores are slower, clock for clock, than Northwood (by design), my hope is Intel will revamp and dump Prescott. Damn thing looks like a hyped up Willamette core for all the performance it delivers.
And, like you, I use terms most people are familiar MD. This is a topic which could easily overwhelm most people. Bottomline: The Prescott design is horrible and should have never been placed in the marketplace. Intel can do better.
I hope that FAB they are currently revamping will house some better processes.
Since Intel rolled out that processor, I have pretty much stopped keeping up with Intel until I see that design fade away. Which is why I asked about Nocona. Is it a 90nm chip? What family is it branched from or based on? What is the targer system for Nocona? Does it actually run cooler than Prescott?
