I've said this before, but one thing that baffles me is in all the benchmarks and performance comparisons, people tend to compare a fresh copy of Windows against a fresh copy of OS X. I think this is completely false. Every self-respecting reviewer should, first of all install all the software they would normally in a production machine, preferably use the reviewed system as their main system for a while.
We all know what happens to Windows, and to some extent to OS X after you start installing stuff on them. But what I'd like to see is a benchmark of Windows with over 100 third party apps installed on it. This is exactly how many apps I have in my Applications folder (96 items in /Applications + sub-folders for stuff like browsers, games and server tools).
> system_profiler SPApplicationsDataType | egrep ":$" | wc -l
269
That's exactly how many apps I have on my system altogether, not counting system-level stuff. With that many apps in the registry, I would bet it would take the Windows PC longer to boot up than the Mac to finish the benchmark. If it would start up at all.
And this all comes back to the cost issue as well. Ask any Windows-using media professional and they'll tell you that you shouldn't install anything else except your bare necessities on your production workstation. That you should essentially have one machine for your video work and another for your email and a third for P2P. But now you're already paying for 3 machines, aren't you?