Advanced Micro Devices Inc (AMD)

[blauboad]

I think any reasonable, unbiased observer would--at the very least--find Sysmark's choices to be strange ones given its stated purpose of representing common real-world tasks and software setups. I can't *prove* it to a lawyer's satisfaction--it's a matter of judgment. Because of that, benchmarks like that are always going to be susceptible to similar accusations, even unreasonable and biased ones.

If Bapco is serious about measuring real-world performance (evidently not!), they would make all of their choices and rationale transparent. The choices would, furthermore, be based on publically-accessible empirical data on how real people actually are using their hardware (preferably culled recently and by a third-party). Short of that, they are of dubious merit.

Then I'd like to see some surveys of perceived performance vs. real performance. That would be very interesting! I suspect that they are only loosely correlated.