I will also run another subject with 100gb of memory to see if there may be a truncation problem, though I don’t see an error in the log files but it’s worth a try.
I’ll reach out to the FSL people as well to see if they have any ideas for how to fix this - I’ll check in with what I find soon.
If it helps in your communication with FSL folks, I am attaching the exact probtrackx2_gpu call that creates outputs of different dimensions:
/opt/qunex/qx_library/etc/fsl_gpu_binaries/probtrackx_gpu_cuda_9.1/probtrackx2_gpu
--loopcheck
--forcedir
--fibthresh=0.01
--cthr=0.2
--sampvox=2
--randfib=1
--nsamples=3000
--nsteps=2000
--steplength=0.5
--pd
--ompl
--samples=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/T1w/Diffusion.bedpostX/merged
--mask=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/T1w/Diffusion.bedpostX/nodif_brain_mask
--meshspace=caret
--seed=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/MNINonLinear/ROIs/Whole_Brain_Trajectory_ROI_2
--seedref=/opt/fsl/fsl/data/standard/MNI152_T1_2mm_brain_mask
--xfm=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/MNINonLinear/xfms/standard2acpc_dc
--invxfm=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/MNINonLinear/xfms/acpc_dc2standard
--stop=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/MNINonLinear/Results/Tractography/stop
--wtstop=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/MNINonLinear/Results/Tractography/wtstop
--waypoints=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/MNINonLinear/ROIs/Whole_Brain_Trajectory_ROI_2
--omatrix3
--target3=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/MNINonLinear/Results/Tractography/Mat3_targets
--dir=/gpfs/project/fas/n3/Studies/Connectome/subjects/105115/hcp/105115/MNINonLinear/Results/Tractography