not to say you shouldn't look at benchmarks, but i don't go around trying to compress James Bond DVDs 400+ times.

most pro users are multitaskers, running myriads of apps whose processes are hard to characterize via simple compression/data transfer tests. not to mention the different component configurations that come into play.

the way i see it, they're good for indicating the max power of your system in a given condition. but actual percieved power by the user is another thing altogether, and benchmarks are just indicators of this power but by no means are the results 1-to-1.