But nothing is yet engraved in stone, it appears. Intel, as former IA-32 architect Bob Colwell reveals in this entertaining talk which is still online at Stanford, typically takes the multiple team approach on design of CPUs and also has folk in the fabs getting their gloves wet messing about with chemicals and trying to achieve the impossible.

As Colwell pointed out in his talk, rather than creating One Big Thing, Intel is now looking to bung a whole heap of smaller things together to achieve the near impossible, even though it's evident to most that Moore's Law is close to the end of the road.

As he said in his illuminating talk: "Because Intel tries to present a unified front, you can't tell where the different parts of the process come from".

He also said: "We're moving away from that one big idea. We don't have any big ideas, so we'll take 10 or 15 small ideas and put them all together."

A chemistry change, however, is likely to give Intel room on the thermal front and they don't want any other pipelines to cause timing delays. Maybe, speculates one bloke that understands this stuff, they'll give us back the 4K micro-ops that are missing if the clever architects and engineers manage to expand trace cache.

Tejas, he speculated, is also likely to include execute protected memory, cryptographic processing built in, and - as we've suggested in previous reports - quite a lot of elements of its Dothan miracle Pentium M to keep the power low as possible.

There are still questions about large cache technology in Dothan but maybe INTC will bung 2MB level two cache into the now-delayed Potomac which could be migrated down to Tejas given time.

The real question now is whether Intel can manage to make a 4GHz Prescott, as it promised late last year. Given that they're still fussing about with Prescotts trying to make them better, this now looks a megahurts too far. But no doubt everyone is working double time to achieve the targets those on high insist on.

As Colwell notes in the speech, the last thing anyone wants is a call from Dr Craig Barrett - or worse, Paul Otellini, with a roadmap for your career which doesn't appear to have a backstop.

He also said: "Sooner or later this game is going to grind to a halt. Exponential trends in general are not sustainable. You're going to hit a wall and you're going to break. Usable CPU performance is levelling off".