Hi there,
I’ve been having problems with accessing my host’s nvidia card from containers , I think it might have stopped working for me with Fedora 42 but I’m not totally sure. I’ve tried several different container images with no success, below I’m trying bazzite-arch-gnome.
I’m running this image:
ostree-image-signed:docker://ghcr.io/ublue-os/bazzite-gnome-nvidia-open:stable
Digest: sha256:eb2983ddedf93b3c8f0217014e863cd932ce5b6cbd2f4a1bb8ef8474e4d476bb
Version: 42.20250501 (2025-05-01T19:25:32Z)
Running nvidia-debugdump --list on the host and the container shows the card
Found 1 NVIDIA devices
Device ID: 0
Device name: NVIDIA GeForce RTX 2070 SUPER (*PrimaryCard)
GPU internal ID: GPU-3b4e64d6-fd82-05c7-3d09-74d70b017424
On in the container if I run vulkaninfo --summary I get
ERROR: [Loader Message] Code 0 : setup_loader_term_phys_devs: Failed to detect any valid GPUs in the current config
ERROR at /usr/src/debug/vulkan-tools/Vulkan-Tools/vulkaninfo/./vulkaninfo.h:248:vkEnumeratePhysicalDevices failed with ERROR_INITIALIZATION_FAILED
vulkaninfo does work as expected on the host.
It does seem like the card is visible on the container but maybe vulkan drivers are missing? Any suggestions on what might be going wrong?