I recently realized that GCC 4.2 on Mac OS X 10.5 supports OpenMP, a relatively easy-to-use parallelization (multiple CPU threading) paradigm. I have started using OpenMP, firstly in program 3dAllineate.
As a result, anyone attempting to compile against the Makefiles
Makefile.macosx_10.5_Intel
Makefile.macosx_10.5_Intel_64
will need to do one of the following
(a) delete the OpenMP special compilation line in the Makefile.* file -- the line starting with 'CCOMP ='
**OR**
(b) switch from using the default GCC 4.0 to GCC 4.2, by changing the symbolic link that defines 'gcc' in /usr/bin to point to gcc-4.2 instead of gcc-4.0, as in
cd /usr/bin
sudo rm gcc
sudo ln -s gcc-4.2 gcc
The corresponding 3dAllineate binaries in the AFNI distributions are now compiled with OpenMP. The program runs somewhat faster with the default settings on my machine, but your results may vary.
I would also like to use OpenMP to speed up 3dREMLfit, but have been unable to get any decent results there so far, for reasons that I don't understand.