Benchmark is in some way associated with viewdistance. Crudely, perhaps you could say that two different comps playing the same mission would give similar performance if the viewdistance was set to the benchmark in each case. However, as you can get different benchmarks just by pressing "autodetect" repeatedly .....
Changing viewdistance on an empty island can have a marked difference in performance. Drawing all these distant objects (and indeed the active part of the game just "knowing" they are there) makes a big difference. I have no idea which bits of your computer handle that, but you would expect that the ones that are are involved in that are also involved in calculating benchmark.