Hi Bill - You should not have a problem at all with the 24 hour limit. Tracula runs independently of the main FS stream (actually has to be run after the main FS stream). Times vary with different configuration parameters and data size, and we should get around to posting some nominal processing times on the wiki soon anyway, but with the basic options none of the 3 steps below should take more than a few hours. Step 3 is the fastest, it takes less than an hour for us. Step 1 is more variable depending on data quality - part of it might run for several tries until it gets things right. Step 2 (FSL's bedpost) can also take less than an hour for a subject if it parallelizes all the subject's slices (it's set up to do that but you may have to modify it to work that way on your cluster).
I hope this helps for now! a.y
On Thu, 1 Sep 2011, Bill Taylor wrote:
I looked at the wiki and the FAQ but still need an answer.
In my work to get more of Freesurfer available to researchers using our cluster, I am concerned about the 48+ hours for a run on one subject. Our 1700 slots which can run these jobs only promise a 24 hour time slice before evictions. Is it likely that all three steps will fit within a 24 hour time frame.
What are the outside limits for: Pre-processing FSL's bedpostX Reconstructing white-matter (WM) pathways
If they run over, is it possible to due the part which runs over in pieces?
Thanks in advance! There are currently 3 groups on campus interested in our automation involving thousands of scans.