1) Car low
Overall system speed is important. Always bench it as last, as it is the most variable test of them all. When you're testing clocks for the different tests, run this one first. The first run is always semi-good. If you're beyond that FPS, you can call it a good run.
2) Car high
As far as I know, this is determined by the l2 cache. A little increasement of the CPU/GPU frequency doesn't really help to increase the FPS. rerunning the CH will result mostly in lower FPS in contrary to Car Low.
3) Dragothic low
Gpu frequency seems to be one of the most important factors, however increasing the CPU speed really helps. As CL, this test is re-runable
4) Dragothic high
Gpu memory speed. At least, that's my experiance, this is the test where the memory usually craps out. In this bench I always have to bench at a lower GPU memory speed than in others. Re-runable? Yeah, most of the time the FPS keeps +/- stable.
5) Lobby low
Memory bandwidth, no? Pedro Rocha knows what exactly is the key to have this one run right
6) Lobby high
Worst bench tests of them all, imho. I have trouble determing the exact bottleneck, need imput.
7) Nature
Gpu core/shader frequency, very re-runable as the FPS keep the same the most of the time.
Bookmarks