I just found this article on the K8 architechture. - The english isn't great - but then mine ain't particularly good either (you gotta hear my accent to appreciate it)

The article taught me a lot about how these things work in PC-land (I'm more up to date on the SPARC stuff). In particular it is clear that a bunch of architechtural improvements have been made which are independant of the 64-bit-ness fo the processors! Memory bandwidth, instruction decoding, and cache architechture all provides for performance improvement!

I know at least some people on this forum will enjoy the article as much as I did :-)