Thank you Bob for looking into this.
It did run with '01' as the pen (took 2 hours with 64-bit; however it failed on 32-bit: malloc failure for image space: 535442080 bytes; in file: mri_new.c at line 193; mri_new_7D_generic; mri_matrix_psinv; 3dTfitter).
So my next question (I guess to Gang) is how much of an impact using the pen as '01,' instead of '012' would have on the PPI analysis, esp. on the accuracy of the deconvolution of the seed timeseries. I can't really compare since I currently can't run it with 012.
Also, what about just rounding up the 1/10s to 1s in my 1D files so that I don't need to upsample my seed timeseries (TR=2) so much?
Thank you!!