How can you think that a 10-20% CPU bottleneck in a high-end review is not a problem?
This sentence shows how little you understand. It really pisses me off when people spend the kind of money that you have without doing any research into these things.
CPU bottlenecks are at a fixed number of FPS for any given set of graphical settings, drivers, OS etc. They are not a percentage penalty. If you get a Sempron and couple it with two 7970s, and then with four, you're going to get identical FPS in every game. According to your implied theory, you should get a discounted version of the FPS you'd get on a high-end platform.
Wizz's graphs only show averages. So you're getting the average of the times when the GPU was able to open up, and the times when the CPU was limiting every graphics solution to roughly the same FPS. This can look like this:
But as you increase the resolution, the FPS spends less and less time being capped by the CPU and more and more time being capped by the GPU, so you get results like this:
As an overclocked i7-920 never imposes a bottleneck at anything even near unplayable FPS in any games in the suite, I really don't see the issue. The graphs show that at 1024*768 or in undemanding games, there's a tiny difference between one 7970 and two. Yes, they'd show a bigger difference with a more powerful CPU, but who cares? The graphs also show that the i7 920 is almost never a bottleneck at 2560*1600 - note near perfect scaling in crossfire in many games. And that's what anyone reading this review is likely to actually care about.
Makes me wonder if the 7970 might actually be enough to make a more noticeable difference.
but the cards are too much for my broke ass right now 
