Hi FSLers, I’m running bedpostx_gpu on some diffusion data from the Human Connectome Project. I have CentOS 6.5 configured with an NVIDIA 780 ti running Cuda 64 bit cuda libraries 5.0. I use: bedpostx_gpu /SUBJID/T1w/Diffusion --model=2 --rician to run the data and in a few hours it completes with an error: —————— 57 parts processed of 58 57 parts processed of 58 58 parts processed of 58 For some reason the bedpostX process DOES NOT appear to have successfully completed. Please examine your results carefully. Queuing post processing stage /usr/local/fsl/bin/bedpostx_gpu: line 235: kill: (4319) - No such process —————— I’ve seen the “For some reason…” error appear on this forum before but I’m wondering (i) why the '58 parts process of 58' is only listed once when all other parts before are listed twice as per the default 2 fibres per voxel and (ii) why there is nothing listed after “Queuing post processing stage” when I’ve seen in other posts on this forum list parts processed after this step (e.g. https://www.jiscmail.ac.uk/cgi-bin/webadmin?A3=ind1403&L=FSL&E=0&P=4323437&B=--047d7bd759209b08e604f4e477f8&T=text%2Fplain;%20charset=ISO-8859-1&header=1 ) What is the best way forward? Cheers, Linden —————————————————————— Linden Parkes PhD Candidate Monash Clinical & Imaging Neuroscience (MCIN) School of Psychological Sciences Monash Biomedical Imaging Monash University A: 770 Blackburn Rd, Clayton, 3168, Vic, Australia T: +61 3 9902 9435 | 0402 516 574 E: [log in to unmask] W: www.med.monash.edu.au/psych/mcin/