Summary of what happened:
When i add the option --omatrix1, the software displays the following error:
“CUDA Runtime Error: OS call failed or operation not supported on this OS”
Without this option, everything works fine.
Command used (and if a helper script was used, a link to the helper script or the command generated):
probtrackx2_gpu -s input/Diffusion.bedpostX_mrtrix3denoised/merged -m input/Diffusion.bedpostX_mrtrix3denoised/nodif_brain_mask -x input/seed/seeds.txt --meshspace=freesurfer --seedref=input/seed/orig.nii.gz --xfm=input/seed/freesurfer2fa.mat --dir=input/result_surf --sampvox=0.29 -S 2857 --loopcheck --omatrix1
Version:
System: Debian GNU/Linux 10
I tried on fsl 6.0.4 and fsl 6.0.6
For the GPU : i tried with cuda 11.2, 10.1, 9.4
Screenshot of nvidia-smi below
Environment (Docker, Singularity, custom installation):
FSL is installed locally.
Relevant log outputs (up to 20 lines):
PROBTRACKX2 VERSION GPU
Log directory is: input/result_surf
Running in seedmask mode
Loading tractography data
read gifti
Number of Seeds: 47015
read gifti
Dimensions Matrix1: 47015 x 47015
Time Loading Data: 296 seconds
...................Allocated GPU 0...................
Device memory available (MB): 32195 ---- Total device memory(MB): 32510
Memory required for allocating data (MB): 4823
Device memory available after copying data (MB): 27357
Running 175252 streamlines in parallel using 2 STREAMS
CUDA Runtime Error: OS call failed or operation not supported on this OS
