<And you think going from Pentium to PentiumPro to Pentium II to Katmai is a breakthrough?>
The transition from Pentium to Pentium Pro *is* a breakthrough. Out of order execution, register renaming, backside L2 cache, three-issue superscalar design, 12-stage pipeline, GTL+ processor bus, etc., etc. All these are features not present in the Pentium, and all of them helped contribute to a 30% increase in speed at the same clock speed for 32-bit applications. Its weakness, however, was 16-bit apps, which was addressed by core changes within the Pentium II.
Meanwhile, I still don't see anything different in the K7 over the P6, except for the 3D-Now instructions, the Slot A EV6 bus, and a symmetric (as opposed to asymmetric) three-issue decoder. In effect, the K7 is more like a P6 plus. That's still enough to leapfrog Intel, given that AMD can match Intel's frequencies, but it's not what I call 7th generation technology.
Meanwhile, Intel is working on some wacko technologies for the Willamette processor. One such technology was already mentioned, an instruction trace cache. I must admit that I don't know how an ITC will benefit x86 code, but I'm willing to guess that the K8 team will be looking at it very closely, among other things.
Tenchusatsu |
