What is your source for the 3.7 with no rollout? Both Car and Driver and MotorTrend, which are the primary publications reporting 3.7's both admit to rolling the car in their published methodologies. They've published stories to this effect, but the quickest Google link I can find that has a staffer admitting to it (Read Michael Austin "C/D Staff" post).
Acceleration Testing Methodology? - Car and Driver Backfires
Here is the quote -
I don't feel like running down every publication, but anyone reporting a 3.7 is rolling the car.
Also, Michael Austin is both BSing in that post and being inaccurate as well. They use 3mph not 0.3mph and the rollout time they publish is more like 0.3-0.5 seconds faster.
Note that the methodology he admits to is not an actual test metric. They just introduce a fudge factor to make the times look better because it helps them sell magazines.
And as to their attempt to "approximate" the 1 foot rollout, their accelerometers provide actual data for that so there isn't any need to fudge except that it allows them to REALLY fudge. Many independent reviewers trying to replicate these tests believe that the published times at Car and Driver and MotorTrend actually correspond more to a 16" rollout (which is agressive, but possible in actual racing, especially with a larger tire), and that they just take a 16" time from the accelerometer as opposed to fudging anything (this time is readily displayed as one of the default options in the software).
Edmunds is the biggest reviewer which does not publish the rollout (at least without clearly labeling it, and deriving it directly from the accelerometer). They have published 4.3 for manual, 4.1 DCT but I think that is based on the factory data for the 2013 M5 (they haven't gotten to do a detailed test yet). BMW doesn't publish rollout numbers because rollout is only something that happens in the NHRA. So 4.3 and 4.1 are "true" numbers and would include launch control in the case of the DCT.
Here are useful links that I put in another post about rollouts -
Edmunds testing methodology -
How We Test Cars and Trucks
Racer X helpfully lists publications and manufacturers which use rollout as well as discussing the whole controversy -
0-60 times - 6speedonline.com Forums
Quote from Car and Driver -
One reason 3mph is bogus is that it's a rule of thumb that comes from NHRA. It has no basis in reality. Racer X is wrong in the sense that they aren't even measuring 3-60. They are just fudging.
Here is a detailed discussion of staging and rollout written by a nice guy I know who deserves the web hits, even if he isn't savvy enough to sell advertising for his content
-
http://wediditforlove.com/techtalk3.html
Rollout is a real thing which can be precisely quantified with an accelerometer, but magazines tend to just introduce a fudge factor instead that tends to be more complimentary than they admit.. to sell magazines.
Bottom line, MSP compares quite nicely in 0-60 acceleration with the 2013 M5. It matches the manual transmission and is a hair slower than the DCT with launch control. Both are a bit faster than MSP in a quarter mile and would crush it in anything longer (MSP traps at ~109mph while I think M5 is ~119mph-ish)