The issue with the charts is a limitation of the CMS I designed. The system wasn't ever designed to handle more than one product in the same article and by default it takes the title of the review and uses that the primary listing in the charts. For this review I had two choices. 1) create 4 articles and link them together (that is what I normally do with the motherboard reviews, I just don't have to create them all at once) or 2) Create a single article and deal with the chart limitation.
You can see which one I went with.
As for your question about dedicated PhysX it really doesn't matter what GPU you pick as any modern nVidia GPU will do a great job at the calculations. For instance in the review I used a GTX 560 and the GTX 650Ti. I didn't post the score but running full res the two offered the same framerate, almost identical in fact. The only major difference was power draw.
So, knowing this you could go with the normal GTX 650 and likely still get the same performance you would with a 650Ti or even 660Ti.
I believe SiSoft has a benchmark to test PhysX performance, I'll have to fire it up and maybe do up a big chart with the results. It may be moot since PhysX in games still isn't all that demanding.