Pentium 4 was supposed to scale up to 10GHz, but they failed to reach 4GHz. I believe it's a bit premature to shout this off the rooftops.
That was different, Intel were relying on natural silicon scaling which dropped sharply at 90nm. They also had nasty leakage problems and worst of all they were trying to take a complex machine up to a high frequency and that caused it's power consumption to skyrocket.
IBM are using completely different methods to boost the clock speed. The core is simpler than before, they're switching to "hand made" logic and there's a lot of clever silicon technology being used which Intel doesn't have.
In any case they already have second revision chips running.
So IBM are gowing narrow with high clock speed. Although the Alpha did high clock speeds, I think with the Failure of the P4, I think this is a bad idea. I personally profer wider with lower clocks.
You can only go so wide before the law of diminishing returns kicks in. Most code can't make sustained use of wide issue machines.
They've gone back to a less aggressive OOO engine (it's said to be comparable to the 604e), this drops power consumption and allows part of the the frequency boost.
The IPC (Instructions per cycle) will drop but probably not by a great deal.
Frankly, the topic should be fixed. It's 'Power6 chips', not 'PowerPC chips'.
Actually I reckon the "PowerPC" brand is going to be dropped, IBM and now Freescale are saying "Power architecture" these days.
BTW you do know there is a "low end" version planned at an even higher frequency...
