Swamp ran the same set of simulations, and posted here:
http://www.m3post.com/forums/showthread.php?t=107582
Did you see those? He's got faster outcomes. I don't know how you guys are setting up the simulation and what's different.
One thing Swamp caught earlier with regards to DCT based on a tangential comment of mine was the time step. What is your time step, and is it significantly lower than the DCT shift time?