Why GPU is not detected?

[zhould1990]

Hello,

I used the following command to run a test case:

./run_fastsurfer.sh --t1 $datadir/P008/P008_T1_mri_orig.mgz
--sid P008 --sd $fastsurferdir

But no GPU is used when the program is running. I have several GPUs available, the recon-all.log file shows:
Cuda available: False, # Available GPUS: 0, Cuda user disabled (--no_cuda flag): False, --> Using device: cpu

Does anyone know why this happens?
Thanks.

[LeHenschel]

Hey,

it looks like pytorch is not recognizing your GPUs. Can you post the output of nvidia-smi and the pytorch version you used? I suspect it has to do with the CUDA version or drivers etc.

In addition, you can test if the FastSurferCNN docker is working for you.

Best,
Leonie

[tashrifbillah]

Hi @LeHenschel , I am going to Google but I have a different GPU discovery error:

[root@pnl-z840-2 fs_test]# docker run --gpus all -v FastSurfer2:/root/FastSurfer2 -v pwd:/root/data --rm fastsurfer:gpu --seg /root/FastSurfer/aparc+aseg.mgz --sid 003 --parallel --t1 /root/data/003_T1w_resampled.nii.gz
docker: Error response from daemon: could not select device driver "" with capabilities: [[gpu]].

Checking with you in case you know a potential solution.

[LeHenschel]

Hey @tashrifbillah,

looks like the nvidia-toolkit might be missing (see: NVIDIA/nvidia-docker#1034).

Best,
Leonie

[tashrifbillah]

Right! People can navigate from that issue or just do--

https://nvidia.github.io/nvidia-container-runtime/
and then
NVIDIA/nvidia-docker#1034 (comment)

[LeHenschel]

Ok, thanks for the pointers. I am closing this issue for now then.