We have been able to establish that we are getting identical results when running recon-all -all on the same scan on different machine for the following: All .mgz files in directories mri, mri/orig, and mri/transforms, by comparing using mri_diff ... All of the .label and .annot files in directory label with md5sum ... Many of the files in directory surf with mris_diff.

We see differences in the BA*.annot files. We are still seeing difference in about 20 of the files in surf including [lr]h.inflated|orig|pial . We do see identity for lh.area.pial but not for lh.pial for instance. I've attached the output of our test routine.

Your thoughts would be welcome on this pesky question.

Best - Don