GPU didn't work after restart, but it was working before the restart, any ideas?

[matthewbaynham]

I started a Podman container with the flag to use the GPU and it worked, but then when a few days later I restarted the container and the GPU doesn't seem to be working.

How do I get the GPU to continue working when I restart the Podman container?

The details...

I create the container with this command.
podman run -ti -p 8443:8443 --name comfy-ui-20250221 --gpus all ubuntu:latest

When I had the container running I installed Python and the torch library, which enabled me to check the GPU.

If I run...

import torch
print (torch.cuda.is_available())

I get true which is good it means it works

If I run...

import torch
print (torch.cuda.device_count())

I get 1 which is good it means it works.

So I managed to get all this to work. But then today I switched on my computer and tested this again before I was going to do something else, and this failed to work, I got false and I also got 0. So torch is not seeing my GPU and the only thing that has happened is I switched off my computer for a couple of days.

So any ideas how I can get the GPU to work when restarting the Podman container?