Every setup and install is going to get a difference FPS, as would my setup in the benches I do for my site. Each bench run itself on the same hardware is going to fluctuate on the average FPS so in theory they are supposed to do multiple runs then average out the bottom line.
What I do is I Google search the card in question IE 660ti review, then go down the lists of each site on the first page. I also then head on over to my site to see what my guys get and go from there. Since I've been doing that so long I can pretty much tell you what card is going to beat what as sites like Guru3D and Techpowerup have a huge list of multiple cards to compare against for each bench.
In regards to the OPs question, that list is terrible and please don't use that site of the list off PCMark either for GPU performance.