Different results on different systems

I did HCP pipeline preprocessing for same data running on 2 computers:
a computer with LINUX CentOS6 64bit
a computer with LINUX CentOS7 64bit

I processed a subject on both systems, and checking the results.
I noticed that the results were visibly different.
From my less experience, I think that the difference probably come from recon-all step.
What could be the reason for this failure at reproducibility?
Note that I changed the program slightly so as not to use the T2 image.

Thank you so much,


Could be the OS could be the lack of T2 in the second run. Both are plausible explanations.

Hi Chris,
Thank you for your comment!

Yes, the lack of T2 could be plausible explanations. But, I processed a subject with lack of T2 on both systems and got different results.
Then I would like to know how the difference in OS affects the results and what is the solution to replicate in different system.


You might find this paper relevant: https://www.frontiersin.org/articles/10.3389/fninf.2015.00012/full

Thank you so much!!
This paper is very useful for me!