These benches have been criticized all over the net - there's no mention what the settings were when these stats were taken. Also, this is data taken from the user machines, which could have a number of unknown things affecting the performance, besides the simple GPU difference.
All this graph shows is that more expensive GPU can run games with higher fps than compared to cheaper GPUs. That's it, you cannot infer anything further from this data.
Yes, there's a lot of weakness there, but it is a start. Perhaps a Tom's reader's PC is "cleaner" and better optimized than an average user's machine. I think the text from the accompanying article had a little more information, but I'm at work now and can't get to it. I think the settings were pretty high though, if only because people do like to play on the highest possible settings. I'm sure more results will be forthcoming. Hopefully Tom's or HardOCP or another site will bench the game after this weekend.