 |
 |
Hi Mark, you can edit the script bedpostx_gpu. Basically, you will need to change the three lines that use fsl_sub. These lines are the ones that submit the jobs to the cluster. You can use your own qsub command. The jobs at the second fsl_sub (xfibres) must wait until the first job has finished (split). The last job (merge) must wait until all the xfibres jobs have finished. Moises. On 25 May 2017 at 17:02, Mark Bergman <[log in to unmask]> wrote: > We've got an existing SGE cluster with various GPUs configured as > consumable resources (ie., "-l > gpu") within our default queue, not in distinct SGE queues (ie., "-q > $FSLCUDAQ"). > > Are there any suggestions about what needs to be changed in the FSL > scripts to configure FSL to use > SGE in this way? > > Thanks, > > Mark > > -- > Mark Bergman voice: 215-746-4061 > [log in to unmask] fax: 215-614-0266 > https://www.cbica.upenn.edu/ > IT Technical Director, Center for Biomedical Image Computing and Analytics > Department of Radiology University of Pennsylvania > PGP Key: https://www.cbica.upenn.edu/sbia/bergman >