Making an apples-to-apples comparison isn't exactly trivial. I'd start with the same sound files on the different systems and do the same simple mix, minus plug-ins at first, to evaluate the mix engines. Since the pan law implementations might be a factor, it would be good to pan some tracks full left-right and some half-way left-right. It is also important to check gain change behavior, so the gain settings for the mix tracks should be moved from 0dB.
Then the mixed tracks can be compared bit-for-bit to look for any difference. It would also be good to do a blind listening comparison to see if anyone hears a difference that is/isn't indicated by the data comparison. I will volunteer to do this for Logic Pro and ProTools TDM (but I can't say when exactly we'll get to it.) It'll make a good student exercise if nothing else.
So many variables, so little time...
-Jay