Failed to boot after manually installed NVidia Driver on Ubuntu 22.04 .

**jiapei100** · June 23rd, 2022

Firstly installed Ubuntu 22.04
Then, I installed Official NVidia Driver downloaded from NVidia website, but NOT from Ubuntu repo

Now, Ubuntu 22.04 hangs at booting.... This is so weird .....

I'm guessing it's because there are 2 video cards?

Code:

:~$ lspci | grep VGA
0000:00:02.0 VGA compatible controller: Intel Corporation Alder Lake-P Integrated Graphics Controller (rev 0c)
0000:01:00.0 VGA compatible controller: NVIDIA Corporation GA103M [GeForce RTX 3080 Ti Mobile] (rev a1)
:~$ glxinfo | egrep -i "device|memory"
    Device: Mesa Intel(R) Graphics (ADL GT2) (0x46a6)
    Video memory: 29872MB
    Unified memory: yes
    GL_AMD_performance_monitor, GL_AMD_pinned_memory, 
    GL_EXT_framebuffer_object, GL_EXT_framebuffer_sRGB, GL_EXT_memory_object, 
    GL_EXT_memory_object_fd, GL_EXT_packed_depth_stencil, GL_EXT_packed_float, 
    GL_AMD_pinned_memory, GL_AMD_query_buffer_object, 
    GL_EXT_gpu_program_parameters, GL_EXT_gpu_shader4, GL_EXT_memory_object, 
    GL_EXT_memory_object_fd, GL_EXT_multi_draw_arrays, 
    GL_EXT_memory_object, GL_EXT_memory_object_fd, GL_EXT_multi_draw_arrays, 
:~$ nvidia-smi
Command 'nvidia-smi' not found, but can be installed with:
sudo apt install nvidia-utils-418-server  # version 418.226.00-0ubuntu4, or
sudo apt install nvidia-utils-390         # version 390.151-0ubuntu0.22.04.1
sudo apt install nvidia-utils-450-server  # version 450.191.01-0ubuntu0.22.04.1
sudo apt install nvidia-utils-470         # version 470.129.06-0ubuntu0.22.04.1
sudo apt install nvidia-utils-470-server  # version 470.129.06-0ubuntu0.22.04.1
sudo apt install nvidia-utils-510         # version 510.73.05-0ubuntu0.22.04.1
sudo apt install nvidia-utils-510-server  # version 510.73.08-0ubuntu0.22.04.1

It looks like currently ONLY Intel video card is running for display, but NVidia video card is NOT running at all. That is also acceptable for me. I actually ONLY use NVidia video card as heavy computation, but NOT for display purpose.
However, in any case, I do need to have NVidia driver installed, either for CUDA computation, or for display (not supposed to, but may have to) .

And, I also realize that after NVidia driver NVIDIA-Linux-x86_64-515.48.07.run is installed manually ,

Ubuntu 22.04 won't be able to boot into XMode; and some times it just hangs before that flashing page, even without letting me switch to Text Model.
I remember I happened to boot into X Mode a couple of times, and it seems the running driver is NOT Intel, but NVidia. I can clearly tell NVidia GPU memory was partially used.

What's more, NVidia driver 510.73.05 from Ubuntu 22.04 repo is working properly...

Can anybody give me a hint? How to correctly boot into X Model with Ubuntu 22.04 ? Is there any possibility that I use Intel GPU for display, and NVidia GPU for computation ONLY?
Must I install nvidia-utils-510 from Ubuntu 22.04 repo??

Cheers
Pei

**#&thj^%** · June 23rd, 2022

As a older user, from your Join Date
Jul 2006 I would have thought you would have installed it through

Code:

ubuntu-drivers autoinstall

downloading and installing from outside the repos can prove problematic.

**MAFoElffen** · June 23rd, 2022

I would suggest running XServer as your Linux Graphics Layer graphics engine, then using an xorg.conf file to specify where each GPU is (bus:slot), and which drivers to use for each... That is the detailed helper for those. Though is usually not needed and is overkill.

If your manually installed NVidia drivers (with Cuda support) does not display, are you sure you installed the correct driver for your specific NVidia GPU? Mine all work with multiple, varied GPU's in the same box... That used to be a problem about a decade ago, but hasn't been a problem in at least the last 4 years... (GPU driver conflict between different GPU makes.)

**tea for one** · June 23rd, 2022

Also, Ubuntu defaults to a Wayland session.
There have been reports of less-than-perfect performance with Nvidia cards in Wayland.
You can select a Xorg session when logging in via the gear wheel (lower right).

**MAFoElffen** · June 23rd, 2022

Originally Posted by tea for one

Also, Ubuntu defaults to a Wayland session.
There have been reports of less-than-perfect performance with Nvidia cards in Wayland.
You can select a Xorg session when logging in via the gear wheel (lower right).

Agreed +1 with running XServer instead of Wayland for NVidia GPU's. That's what I do for all my NVidia GPU machines.

**#&thj^%** · June 23rd, 2022

Agreed +2 running XServer instead of Wayland

**jiapei100** · June 28th, 2022

Yeah... You are correct.
By using Ubuntu repos's NVidia driver 510.73.05, it suppots CUDA 11.6 ONLY.
I actually happened to have installed CUDA 11.7 .

Everything went on well (NVidia driver 510.73.05 + CUDA 11.7), until yesterday, I noticed the incompatibility... sorry, didn't take a snapshot but my nvidia-smi clearly shows:

Code:

➜  ~ nvidia-smi
Tue Jun 28 06:07:30 2022       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 510.73.05    Driver Version: 510.73.05    CUDA Version: 11.6     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
......

However, when I run a **cuda-sample** after building it:

Code:

➜  release git:(master) ✗ ./deviceQuery
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 1 CUDA Capable device(s)

Device 0: "NVIDIA GeForce RTX 2080 Ti"
  CUDA Driver Version / Runtime Version          11.6 / 11.7
  CUDA Capability Major/Minor version number:    7.5
  Total amount of global memory:                 10986 MBytes (11519590400 bytes)
  (068) Multiprocessors, (064) CUDA Cores/MP:    4352 CUDA Cores
  GPU Max Clock rate:                            1755 MHz (1.75 GHz)
  Memory Clock rate:                             7000 Mhz
......
deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 11.6, CUDA Runtime Version = 11.7, NumDevs = 1
Result = PASS

I clearly had CUDA 11.7 installed..

It's better to have a Driver compatible to CUDA 11.7 ...

**jiapei100** · June 28th, 2022

God save me... it looks **nvidia-driver-515** supporting **CUDA-11.7** is just out ....

my gosh....

Thank Ubuntu ... ^_^

Thank you ...

**jiapei-nexera** · June 29th, 2022

Problem continues...

At home, my desktop only has a a single GPU that's why a simple

Code:

sudo ubuntu-driver autoinstall

will do.
However, in the office, this laptop comes with 2 GPUS as mentioned above, quoted again:

Code:

➜  ~ lspci | grep VGA00:02.0 VGA compatible controller: Intel Corporation Alder Lake-P Integrated Graphics Controller (rev 0c)
01:00.0 VGA compatible controller: NVIDIA Corporation GA103M [GeForce RTX 3080 Ti Mobile] (rev a1)

Current status:

1. How I'm now able to login GUI ?

Code:

sudo apt purge gdm gdm3
sudo apt install gdm3 ubuntu-desktop
sudo systemctl restart gdm3

2. After that I had theh GUI popped up, but:
- Selecting Ubuntu (This is clearly Wayland) is working
- Selecting Ubuntu X .... (This should be Xorg ) is NOT working with a black screen

3. With the GUI, I'm now replying this message:

Code:

➜  ~ nvidia-smi
Failed to initialize NVML: Driver/library version mismatch
➜  ~ echo $XDG_SESSION_TYPE
wayland

Let's forget NVidia driver for a bit, but come back to the key part:

- Is there a way to select by ourselves: which display card is used for display (with its particular driver), and the other display card is NOT for display but with a prepared driver for CUDA application?
- It looks to be when we installed NVidia driver, it builds into the kernel automatically, which seems to drive for display automatically. That is also acceptable, although that will directly disable Intel integrated display card... However, is it guaranteed that NVidia driver is able to find the correct graphic card for display? Not 100% sure about this... For instance, I already have NVidia driver installed,

Code:

➜  ~ apt list --installed | grep nvidia


WARNING: apt does not have a stable CLI interface. Use with caution in scripts.


libnvidia-cfg1-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
libnvidia-common-515/jammy-updates,jammy-updates,jammy-security,jammy-security,now 515.48.07-0ubuntu0.22.04.2 all [installed,automatic]
libnvidia-compute-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
libnvidia-compute-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 i386 [installed,automatic]
libnvidia-decode-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
libnvidia-decode-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 i386 [installed,automatic]
libnvidia-egl-wayland1/jammy,now 1:1.1.9-1.1 amd64 [installed,automatic]
libnvidia-encode-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
libnvidia-encode-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 i386 [installed,automatic]
libnvidia-extra-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
libnvidia-fbc1-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
libnvidia-fbc1-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 i386 [installed,automatic]
libnvidia-gl-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
libnvidia-gl-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 i386 [installed,automatic]
linux-modules-nvidia-515-5.15.0-40-generic/jammy-updates,now 5.15.0-40.43+1 amd64 [installed,automatic]
linux-modules-nvidia-515-generic-hwe-22.04/jammy-updates,now 5.15.0-40.43+1 amd64 [installed]
linux-objects-nvidia-510-5.15.0-40-generic/jammy-updates,now 5.15.0-40.43+1 amd64 [installed,automatic]
linux-objects-nvidia-515-5.15.0-40-generic/jammy-updates,now 5.15.0-40.43+1 amd64 [installed,automatic]
linux-signatures-nvidia-5.15.0-40-generic/jammy-updates,now 5.15.0-40.43+1 amd64 [installed,automatic]
nvidia-compute-utils-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
nvidia-driver-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed]
nvidia-kernel-common-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
nvidia-kernel-source-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
nvidia-prime/jammy,jammy,now 0.8.17.1 all [installed,automatic]
nvidia-settings/jammy,now 510.47.03-0ubuntu1 amd64 [installed,automatic]
nvidia-utils-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
xserver-xorg-video-nvidia-515/jammy-updates,jammy-security,now 515.48.07-0ubuntu0.22.04.2 amd64 [installed,automatic]
➜  ~

But why there is

Code:

➜  ~ nvidia-smi
Failed to initialize NVML: Driver/library version mismatch

Anyway, I'm going to remove ALL default nvidia packages from Ubuntu, and try this way out:

https://stackoverflow.com/questions/...rsion-mismatch
https://stackoverflow.com/questions/...33147_43022843

4. Finally, by the way, I'm using this laptop: https://rog.asus.com/ca-en/laptops/r...6-2022-series/

**jiapei-nexera** · June 29th, 2022

It seems I solved it... Make sure you always have 1 single version of NVidia driver installed ...
I just uninstalled my manually installed NVidia driver, and the Ubuntu NVidia driver, EVERYTHING...
And now I can log in Xorg, with nvidia-driver enabled...

Finally... my god...