(1) you don't need or want to use -resid with -Clustsim, since the program will do that for you when you use -Clustsim.
(1) 3dttest++ -Clustsim does NOT use -acf or any smoothness estimates at all. It doesn't model the spatial structure of the noise directly. It uses a randomization of residuals approach to compute noise-only data, and then runs that stuff through 3dClustsim
(2) The results from 3dttest++ -Clustsim can be very different (and have much larger cluster size thresholds) than 3dFWHMx -acf + 3dClustSim -acf. This is the price for obtaining accurate False Alarm Rate (FAR) control -- this is our current approach to dealing with the issues in the Eklund et al. PNAS paper -- for more details, see the manuscript
http://biorxiv.org/content/early/2016/07/26/065862
(3) I am working on developing a better thresholding tool that still maintains FAR control, but that will be some time off -- weeks, at best, considering that even after implementing it, I have to test it, and every test case (as in Eklund et alii and as in our bioRxiv manuscript) takes several days to run -- and there are a lot of cases to test. And I have other responsibilities.