Hi Chris,
The problem is now fixed. It was an evil combination of sparse data, cmask, and a compiler optimizer bug! This should all be sorted out now.
AFNI binaries postdating this message will contain the fix.
Note, that your detrending order is a little low. This detrending is meant to remove anatomical structure in the image and reduce activation signal components. You want to measure the spatial smoothness of the noise, not the anatomy or the activation.
I suggest you leave the detrending order out and let the program decide, also -bmall is overkill, given that you have more than 240 samples in your time series.
A simpler command would be:
SurfSmooth -met HEAT_07 -spec std.SM11_rh.spec \
-surf_A std.rh.smoothwm.asc \
-input v2s_std.rh.RP_run01.niml.dset \
-blurmaster v2s_std.rh.RP_run01.niml.dset \
-output rh.v2s.blr.niml.dset \
-target_fwhm 4 \
-cmask '-a v2s_std.rh.RP_run01.niml.dset -expr bool(a)' \
-detrend_master
cheers,
ziad