The big question in all this is not whether Linux is capable of running games with the same performance as windows, but whether the game dev companies will actually create a linux port of thier game/s in the first place.
There's some big factors in this, most of them revolving around money - development costs, distribution figures etc.
Put it this way, if your intended target base is known to be 95% windows, why the heck expend extra time (and thus cost) for the other 5% ?
While Wine is a solution, it's not the right direction to take unless the performance and stability matches the performance and stablity of the game on it's native platform, which I dought is possible.
Also, if Wine becomes relatively successfull, all the more reason for game developers not to make a native linux version of thier game/s.
id software made a brave move with Quake3, releasing for 3 seperate platforms, but as far as I know, the much anticipated RTCW will not be available in Linux flavour in stores - sure, a binary will be released, but that's hardly the point as what that says to game development companies is that making ports for Linux is not profitable.
I'd like to see Linux succeed as a gaming platform, but not at the expense of it's main strength - the server market.
For Linux to succeed as a gaming platform, it first has to gain wider acceptance as a desktop platform, something analysts are not predicting will happen any time soon.