Changes for version 0.005 - 2026-10-21
- detect (lspci and sysfs) now lists AMD Instinct accelerators, which enumerate as PCI class 0380, under "amd" (compute 0, still no driver install); gpu_setup logs its "AMD GPU detected" warning for them. Class 0380 is accepted for vendor 1002 only; NVIDIA detection is unchanged. The lspci read command string changed to match.
- generate_cdi_specs no longer logs "[ok] CDI specs written" when nvidia-ctk cdi generate for nvidia.com/gpu=all fails; it warns and carries on, like the management spec.
- _reboot_and_wait now compares /proc/sys/kernel/random/boot_id before and after the reboot and reports the host back only on a different boot; a host that cannot read it falls back to the old answer-only check.
- verify_nvidia (the last gpu_setup step) warns when `crun --version` reports a crun older than 1.15 (1.14.1 is known broken: with the GPU Operator's NVIDIA toolkit making crun the first runtime, sandboxes fail with "unknown version specified"). Warning only, nothing is changed; a host without crun is silent. It adds one `can_run: crun` probe.
Modules
GPU detection and driver management for Rex
GPU hardware detection via PCI class codes
GPU detection from sysfs, without lspci or pciutils
NVIDIA GPU driver and container toolkit management
What NVIDIA driver a GPU generation needs (experimental)
Base class of the per-distro NVIDIA driver setups (experimental)
apt/dpkg packaging layer of the NVIDIA driver setups (experimental)
NVIDIA driver setup for Debian (experimental)
NVIDIA driver setup for RHEL, Rocky, AlmaLinux and CentOS Stream (experimental)
rpm packaging layer of the NVIDIA driver setups (experimental)
NVIDIA driver setup for openSUSE Leap (experimental)
NVIDIA driver setup for Ubuntu (experimental)
NVIDIA driver setup for Ubuntu, package named by ubuntu-drivers (experimental)
Which PCI subsystem IDs mark an NVIDIA vGPU guest device