To each his own
Using Sparky's to emulate NGC's setup (true ellip sliding collision with instant all group feedback) is slower here (from the DBP code and not Sparky's itself).
As I said before the fast stack options in NGC along with the exclusive replacement commands work great here. Matthew just doesn't give any examples showing this, which will change in the next update if i have time to help him. There are a few bugs still in NGC, but they will be fixed in time.
I would love to see a version of both Sparky's and NGC in DGDK.net as I use Sparky's for applications that use non-sliding or sphere sliding collisions
I refuse to use more than one at a time whether it is faster or not, I hate wasting memory using a physics and collision or 2 collision systems at the same time. Nothing is really wrong with it I guess in todays computers, it's just a quirk of mine.