I believe 3DMark does not have actual "default" settings, it just puts things on the resolution of the screen so the screen will mess kinda with the results (lots of different resolutions). I don't know about you, but IMHO if we get thru all this testing and comparing we should do it "right". However, "right" could also mean "in the normal circumstances like during daily use". So I think I will go with your solution in case of 3DMark, and perhaps go for a standard resolution in windowed mode for the progs we have more influence on (like FurMark in benchmark mode). I will give some room in the public test data spread sheet to fill in driver settings, assuming not everyone is willing to go on default with that. Might be interesting as well, to see if different driver settings significantly influence performance differences between comparable set-ups.