Re: I can't tell you why, but I can add 2 and 2 and it's pretty obvious that AMD's 130nm SOI process is producing chips that run faster than INTCs best.
AMD's chips don't run faster (i.e. clock higher); rather, they are higher performance, which is an important differentiation. The myth is that SOI is the reason, but this is absurd. AMD has better performance because the 31-stage Prescott core has too high of a performance penalty to compensate for the clock speed it runs at. AMD made a different micro-architecture, and it has paid off.
Once again, you guys are left with plenty of proof in the form of Dothan, which is constrained to a 21W TDP, has less than half as much memory bandwidth as Athlon 64, and does not have the benefit of an on-die memory controller; yet in spite of this, it still performance comperably on many benchmarks. And it doesn't use SOI.
But rather than using this as completely obvious proof, the critics think up ways that Dothan is the exception to the rule, and that Prescott should be compared instead. Once again, this is absurd logic that the Internet community has formed into an urban legend. Prescott is the exception, because it is so unlike any processor that has preceded it. It has tons of performance enhancing logic that generates plenty of heat for the sole reason of reversing some of the immense performance penalties of the 31-stage pipeline. The thermal characteristics have stunted frequency growth, and all the hard work and engineering that went into the design has effectively been for naught.
Prescott may have been a great science project, but in pragmatic terms it was a mistake for Intel of epic proportions. You really should blame the CPU planning entirely, rather than the manufacturing process or the engineers involved. SOI has nothing to do here, and neither does any of the other red herrings invented to throw more mud at Intel for what is really just a single - gigantic - mistake in planning.
