Hi Greg,
This is an issue with the way FEAT currently handles models with large numbers of confounds, we are working on a version with much lower RAM requirements that will be released in a future patch.
Kind regards
Matthew
> I have an MR task that is approximately 1700 volumes. When running this task through a simple analysis in FEAT, which includes adding motion parameter regression, it takes about 9 hours and about 4GB RAM. If I run the motion_outliers script and add in the 100~300 outliers as confounds, the task then takes 12GB RAM and 3-4 days to complete...
>
> I'm wondering what I can do to reduce the RAM usage or compute time?
|