I believe it was two years ago that one of the major hardware sites found out that the mailing address for a multitude of benchmarks (I'm pretty sure it wasn't Sysmark) was a building that Intel owned.
Furthermore, one year (I believe 2002), AMD *won* that particular benchmark. I *think* one of the tests was crunching a lot of excel data. The following year, the benchmark was changed to favor Intel and they, surprise surprise, won. Basically they do better with HUGE excel sheets (that you will almost never use) and AMD was better for everyday use. Guess which test was included.
Now after that debacle, benchmarking has gotten better (mainly b/c Intel is being watched), but it's hardly the perfect tool to use for comparison and never will be. Sites do their best to even things out, but they will never be even.
Now at the same time, I'm not saying to totally disregard them. Instead, take them with a grain of salt or just as advice, not absolution like a lot of people seem to take them as. Benchmarks are made by people and a lot of those people work for Intel

If you *really* wanna know the performance of a processor and have 20,000 dollar simulation software, then I guess you could figure it out for *one particular set of tests that you are testing*.
Can anyone say Monopolistic Business Practices? No wonder why they're being sued by the Japanese Goverment
