> Jeepers adding a mask speeds up the program considerably.

Yes, masking would avoid some unnecessary computations. Also you can increase the threshold number of nonzero values to save runtime. For example, in your case you could change from 20 to 25.

> I still get a few error messages at boundary conditions like you mention.

That's a little weird because I failed to duplicate such an error here. Get the most recent version of 3dMEMA here: [afni.nimh.nih.gov], and see if it would be fixed. Be realized that this new version will also have a new option to detect outliers at individual subject level, plus the capability to include covariates in the model.

> The exact error related to specifying multiple processors is:
> ssh: Could not resolve hostname 2: No address associated with nodename
> ...

I also failed to get on my computer your error message associated with multiple jobs. What kind of system are you running? I just saw somebody who had a similar problem on Debian, but no solution was shown there:

[tolstoy.newcastle.edu.au]

Gang