I am trying to run bedpostx_gpu on our university cluster with Nvidia nodes. The fsl version is 5.0.9 and the cuda version is 6.5. The system of cluster is Centos6. When I run bedpostx_gpu, it basically creating folders and log files without doing any real calculation. The following is what terminal shows:

---------------------------------------------

------------ BedpostX GPU Version -----------

---------------------------------------------

subjectdir is /scratch/gaoy3/Diesel/b3000_GPU

Making bedpostx directory structure

Copying files to bedpost directory

Pre-processing stage

Queuing parallel processing stage

----- Bedpostx Monitor -----

/bin/sh: line 1: /scratch/gaoy3/Diesel/b3000_GPU.bedpostX/logs: is a directory

Queuing post processing stage

Would you please give me some suggestions on this?

Yurui Gao