Intel Corp (INTC)

[spokeshave]

fingolfen: Re: "When AMD moved to copper for 0.18 micron, many individuals made the case that 1) they would enjoy a massive speed advantage due to copper and 2) their move to 0.13 micron would be easier than Intel's because "they already learned the lessons of copper.""

I would argue that AMD did enjoy a significant speed advantage that coincided with the transition to copper and 0.18u. Thunderbird was beating P/// hands-down, and P4 was only beginning to ramp. It really was not until Northwood that Intel regained the undisputed speed crown. Whether or not AMD ever had a "massive" speed advantage is an argument over semantics.

"Analysis of the evolution of the AMD 0.18 micron process shows that the only reason they maintained performance parity with Intel was through shortening the channel lengths, not because of any intrinsic speed benefit from copper. Copper, it appears, is truly there primarily for electromigration issues, just like Intel said way back when."

Well, Intel was not the only one saying what the benefits of copper were. I don't think that anyone who has at least a rudimentary understanding of the physics could conclude that copper in and of itself has any intrinsic speed benefit. Copper provides reduced electromigration and lower electrical resistance. These characteristics were necessary for a shrink to be successful. The shrink allowed for better speeds, copper allowed the shrink.

As for the migration to 0.13u, AMD simply dropped the ball. It could be that they were trying to do too much at one Fab, it could be poor management, it could be poor engineering, it could be a myriad other things. I do not know for sure what the real reasons were (are). I do know that AMD went from hero to goat in relatively short order, while Intel executed flawlessly. However, if AMD can get Hammer ironed out quickly, that could reverse just as quickly.

I agree that Intel has a strong design and development group. But, they have not been without their own gaffes. Like AMD, Intel has a long list of "oops" in their history. I simply maintain that strained silicon is a very new medium, and there is at least the potential for it to be a roadblock for Intel.