As this uses the actual game engine from X3 Reunion (it's a demo & benchmark) which uses SM 2 & 3, depends which you activate, IMHO it is a slightly better testing/tweaking your graphics system than 3DMarks synthetic systems can provide you at times.
Because 3Dmark can vary +/- 50 - 250 a run with the user changing nothing, how can you tell if any changes you made did anything with variance like that?
The in game timetest type benchmarks make much better tweaking systems than 3Dmark could in it's current form IMHO. Even if you avg the 3Dmark runs the variance is still to high sometimes to actually tell if anything has changed or not.