First, stop trying to use theoretical peak performance numbers; they are meaningless for all intents and purposes. It's like to trying to evaluate F1 cars based on how they perform on a perfectly straight stretch of road, when no track in the world is even remotely straight and it's all dependent on how good a driver they put behind the wheel anyway (hmmm...I rather like that analogy).bbot said:The cell chip can do 218 gflops. The xcpu can do 115 gflops. The RSX can do 136 shader ops. The Xgpu can do 96 shader ops. In a previous interview, Carmack calls the PS3 "marginally" more powerful. Now he says pretty much the same thing: "a bit" more powerful. (raw gpu and cpu power)
Wouldn't you agree that the PS3 is not "marginally" and "a bit" more powerful, but alot more powerful (raw performance) than Xbox 360?
And no, I don't think anyone with any decent technical knowledge who is credible would say that the PS3 is a lot more powerful. Search the forums and the Internets and you'll see that the consensus is that in real world performance these two systems are going to be very close to each other. But anyway, we have to wait until the PS3 is actually released before we can have even slightly reasonable metrics to compare the two system (even then, there's no perfectly fair comparisons). As bad as comparing theoretical specs are, comparing a released product to a yet-to-be-released one is even worse. You just can't do it; all you can do is speculate.
Even then, though, comparing multi-platform games isn't going to be the end-all-be-all. Going back to the F1 analogy, even if you have both cars both drive the same track (e.g. multiplatform games), the times are going to be heavily dependent on the driver behind the wheel. What that means is that the quality of developer behind the game is going to have far more impact on it than the platform it's released on. (I'm actually loving this analogy, I'm going to always use this from now on...)