Hacker News Reader

But they are not in the actual Linux kernel, and are a separate module that does not adhere to Linux kernel code conventions. It also has no user-space driver that isn't NVIDIA's proprietary driver. Anything further open-source such as NVK is not being worked on by NVIDIA, but by other 3rd parties.

Compared to AMD and Intel, NVIDIA is very much not an 'out of the box', or stable experience.

Saying "and future" is like taunting fate

Anyway, in case someone was interested it seems the code itself is cited as MIT, however it has a "when it becomes a Linux .ko it becomes GPLv2" clause https://github.com/NVIDIA/open-gpu-kernel-modules/blob/580.7... and they do go out of their way to say "lol, needs binary blobs" https://github.com/NVIDIA/open-gpu-kernel-modules/blob/580.7...

That XFree86.run has always struck me as "you're gonna what*?"

To be fair, AMD's driver also needs proprietary firmware blobs. This is still infinitely better than having mistery sauce running in kernel space directly. A bigger problem is that there is no performant open source user-space counterpart like there is for AMD (Mesa with radeonsi for OpenGL and radv for Vulkan).

Kernel driver only.

Is AMD different, do they OpenSource firmware as well?

They have an open source userspace component.