Thanks, 3.8 seems to have fixed the issue, must've been a driver thing like you say.
Can you tell me how you fixed this? I am running into the same issue currently and on drivers v510.73.05 and CUDA 11.6. I'm decently familiar with building new imagine and have used the Dockerfile to build with `nvidia/cuda:11.6.2-base-ubuntu20.04` but still no success.
Of note: `sudo docker run --rm --gpus all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi` does return the output of `nvidia-smi` so the gpu can be passed to the container.
EDIT: Ok new information. Tag 3.8 from Docker Hub works with my drivers stated above but 4.2 and beyond do not.
[Support] PTRFRLL - Docker images
in Docker Containers
Posted · Edited by ksnell
Update
Can you tell me how you fixed this? I am running into the same issue currently and on drivers v510.73.05 and CUDA 11.6. I'm decently familiar with building new imagine and have used the Dockerfile to build with `nvidia/cuda:11.6.2-base-ubuntu20.04` but still no success.
Of note: `sudo docker run --rm --gpus all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi` does return the output of `nvidia-smi` so the gpu can be passed to the container.
EDIT: Ok new information. Tag 3.8 from Docker Hub works with my drivers stated above but 4.2 and beyond do not.
[ptrfrll/nv-docker-trex Tags | Docker Hub](https://registry.hub.docker.com/r/ptrfrll/nv-docker-trex/tags)
Apologies, the number 3.8 threw me off before as it is the same version my compose files are on!