Hello Jim:
I've probably asked this question before:
Why are you using 3dDeconvolve on "data that's been normalized and resampled
to +tlrc."?
If your Orig voxels are 4x4x4, then converting to Tlrc 1x1x1 means an increase
in processing time (not to mention memory requirements) by a factor of 64.
I don't understand the reason for doing this.
However, to answer your question: All of the 3dDeconvolve source code is
written in C. Probably most of the real work that is done by 3dDeconvolve
occurs in the matrix arithmetic routines contained in file matrix.c.
Doug Ward