Yes, same behaviour after shutdown, waiting 5 minutes and turn on again 
This is the current output of my dmesg
[ 5.306031] nvidia: loading out-of-tree module taints kernel.
[ 5.306041] nvidia: module verification failed: signature and/or required key missing - tainting kernel
[ 5.334304] nvidia-nvlink: Nvlink Core is being initialized, major device number 509
[ 5.334309] NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
[ 5.335295] NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
[ 5.335298] NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
[ 5.335299] NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
[ 5.335306] nvidia 0000:01:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem
[ 5.347449] NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64 570.86.16 Release Build (dvs-builder@U16-I2-C04-15-2) Fri Jan 24 21:05:28 UTC 2025
[ 5.573804] nvidia-modeset: Loading NVIDIA UNIX Open Kernel Mode Setting Driver for x86_64 570.86.16 Release Build (dvs-builder@U16-I2-C04-15-2) Fri Jan 24 20:49:16 UTC 2025
[ 5.597037] [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
[ 5.597041] [drm] Initialized nvidia-drm 0.0.0 for 0000:01:00.0 on minor 2
[ 5.627539] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card0/input15
[ 5.627572] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card0/input16
[ 5.627595] input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card0/input17
[ 5.627615] input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card0/input18
[ 6.375567] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
[ 6.426346] nvidia-uvm: Loaded the UVM driver, major device number 507.
[ 19.020851] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
[ 19.021589] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
[ 19.031358] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
[ 19.031985] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
[ 19.034052] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
[ 19.034723] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
[ 19.037124] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
[ 19.037663] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
[ 48.338604] nvidia 0000:01:00.0: Direct firmware load for nvidia/570.86.16/gsp_ga10x.bin failed with error -2
file exists??
do you have secure boot enabled? maybe easiest thing for you is disable if so?
Okay, got it working now with the open moduls verison.
Just had some old manual entries in my grub file. After removing them, it works now.
Thanks for your help
Okay, Gardiner Bryant confirmed Resolve Studio is working good enough on the open headers.
Now I want to know specifics of if all the accelerations are working properly. Specifically 10 bit 4:2:2 video.
Also I guess gaming performance can only be measured using NVK in Mesa since there’s no proprietary Vulkan stack for the 5000 series.
What’s concerning though is the full discarding of the proprietary stack means Maxwell and Pascal GPUs will stop working if drivers are updated.
Pardon their french, but this is a must watch to set expectations.
Pretty frustrating dealing with this. It’s very hit or miss whether I get video out (even for GRUB!), and I don’t want to unseat my 5090 from my sandwich case VR rig to use it in a separate machine for Ubuntu. Wish ASUS would add BIOS options to disable M.2 slots.
With the Linux 5070 Ti benchmarks now out on the Linux channel, I don’t think that 1% low issue is a Nvidia issue. It’s a Mesa NVK issue and it’s down to that team to get performance back up to par. There is no proprietary stack comparison cause it’s not allowed on Blackwell.
This is 100% in the hands of NVK to improve performance.
Just follow the script below on a brand new install of Ubuntu Server 24.04 for an RTX 5080:
sudo apt update && sudo apt upgrade -y
#Download and Install Latest Kernel for Ubuntu 24.04 Server
sudo add-apt-repository ppa:cappelikan/ppa
sudo apt update && sudo apt full-upgrade -y
sudo apt install -y mainline
mainline
sudo apt install pkexec
sudo mainline install 6.13.6
#Install GCC 14 and Build Essencials
sudo apt update
sudo apt install build-essential
sudo apt install gcc-14
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-14 14
sudo apt install g+±14
sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g+±14 100
#Download and install Latest NVidia Drivers (Choose MIT License)
sudo chmod +x NVIDIA-Linux-x86_64-570.124.04.run
sudo ./NVIDIA-Linux-x86_64-570.124.04.run
#Download and Install Cuda
sudo apt install nvidia-cuda-toolkit
nvcc --version
1 Like
Can you explain the reason for choosing the MIT license version over the proprietary version. Advantage/disadvantage one over the other?
Just saw that choice in the installer for the 570 stable drivers. Don’t remember that choice in the 550 drivers.
nvidia isnt really giving much of a choice, “new” development is under the mit license version vs proprietary. so if your kernel is new enough it literally wont compile with the “old” proprietary way.
this change is because its been unduly hard on the linux community to bundle proprietary+gpl linux kernel together in the past because neither license is compatible. You, the end user, can choose to marry them together but thats not a choice someone else can make for you, and distribute, the way open source software works.
so nvidia is trying to right the past wrong of “eh, whatever, we don’t care” proprietary license with an approach that is more compatible with wider distribution
2 Likes