Hello,
I tried out the solution using 3dZcutup and that reduces memory load significantly - but big glms still take a lot of memory. I run 3dDeconvolve seperately for all 30 slices, which yields a memory usage of about 120 MB for small glms (34 regressors, 28 output subbriks). For big glms (158 regressors, 397 output subbriks) memory usage is still high, almost 1 GB. Is this normal? Why is so much memory needed? Do I do something wrong?
Nick