Ummm...Hasn't there been a good bit of fury over the fact that Sisoft has reportedly optimized their software in favor of the Intel products? And not to rain on anyone's parade, but that -only- shows you a single operation execution, and that does =not= equate to anything like real world performance. This doesn't excercise either the L1 or L2 cache, and how they perform can have a tremendous effect on the results. It doesn't seem complex enough to stress the branch predictor and prefetch predictor systems either; another area where there clock cycles are secondary to architecture. Then you have to take into account frontside bus latency, access time for both the ram and the hard drive(s), what OS is controlling all of this and what it is doing with and to the system resources, PCI and AGP bus latencies, interrupt and DMA requests, possibly even thermal conditions in the case and room. All this really shows is that a 3 ghz P4 is clocking %55 faster than an XP-2100 (which a glance at a spec sheet will tell you anyway).