If you want to talk at the level of choices deciding 3%
changes in computational power efficiency, well yes, my
description is oversimplified.
Add a few of these up, add in the other things I commented on in another post, and you can end up with a 30% difference in power per unit of real work between various cores. All we have to do is look at power output and some benchmarks for each core, and we'll see this difference.
So let's see what Prescott and 90nm Athlon64 bring to the table. Rumors are already saying 90nm Prescott is going to consume more power than 130nm Athlon64 at the same level of performance. We'll see if this holds true. If it does, AMD has more power headroom, which was what this was all about, right?