1) I believe BrainStorm was posting in regards to the AnandTech benchmark, not the GamePC benchmark. kha just happened to bring up the other one.
2) In the GamePC benchmark, there's a bit of a discrepancy. For one, the Xeons are only winning the synthetic Sandra benchmarks (whoops, synthetic), the half-synthetic SYSmark (he even notes it's not a good real-world SMP scenario), and the Quake benchmarks (OK, Q2 isn't SMP-enabled, but score one or two anyways. Inexplicable--some esoteric bit of Abrash's code hanging around in the engine?)
Out of the true real-world benchmarks, the dual Xeons either win by a hair or get thoroughly slammed. Therein lies the biggest discrepancy...
Kelledin
"/join #hackerz. See the Web. DoS interesting people."