The highest dyno numbers are from hub dynos (vs rollers), which very few tuners utilize. In addition to giving higher numbers they are also probably more accurate as they eliminate the losses from the wheels/tires (grip, strapping technique, friction, heat etc). The few hub dynos that been done peg the M3 at least in the 370 rear hub horsepower range bone stock, with ~10% drivetrain loss.
In general, when one hears about dynoing the M3 and the dashboard lighting up like a christmas tree, how can one be confident that the ECU isn't responding to prevailing conditions and detecting any out-of-the ordinary parameters and making changes to protect the engine? This would suggest that any dyno of performance modifications needs to be done with a baseline to really get any indication of whether those mods actually add power (and even then, the dyno conditions are still changing with each run).
Last edited by FogCityM3; 11-20-2012 at 02:30 PM.