I noticed, that my GPUs don’t show in any resource monitor I’ve tried (btop, stacer, monitor, command center). I use the nvidia proprietary drivers and nvidia-smi does see the GPU:
If nvidia-smi is working the hardware monitors are probably a red herring. Maybe if they in fact properly support nvidia and that’s actually compiled in, though that’s unlikely as the graphics pipeline/basic openGL libraries are completely separate from nvidia’s monitoring API. Plus, firefox’ WebGL working means that the GL libraries are set up correctly, so it’s probably just lwjgl.
My guess would be something about the environment you launch lwjgl in. What is the context around that?
Unrelatedly, your nvidia config is probably quite suboptimal. A few notes with links to the nvidia docs that actually explain the settings where appropriate:
You shouldn’t manually load nvidia kernel modules. The various hardware.nvidia settings enable them when necessary, manually setting them is a noop at best and adds incompatible or half-configured stuff at worst.
The fbdev thing is a workaround for a bug that has probably disappeared by now, but in fairness setting it should just be a noop as well.
Your iGPU is probably not showin anywhere because you’re using only your nvidia GPU. To use both you’d need to set up offload rendering, which is probably a good idea on devices with an iGPU (and will correctly set up your uvm module).
The hyprland wiki also recommends some environment variables you do not currently set, if memory serves. That may also be relevant, but it’s been a long time since I’ve used it.
Cheers for the information on the NVidia settings, I’ll change that.
And about lwjgl: I was playing around with a libGDX project I had previously developed on a Windows machine in school. So I probably just had something set up the wrong way to run on Linux.
