Debian 13 > how to install NVIDIA Corporation GP108M [GeForce MX150]

Graphics / Linux Linux
1

Hello

I try to make work the following card under Linux Debian 13 (trixie):

 $ lspci | grep -i nvidia
01:00.0 3D controller: NVIDIA Corporation GP108M [GeForce MX150] (rev a1)

Several drivers are available from the non-free repository:

(mando@velvet) (~) $ apt search xserver-xorg-video-nvidia
xserver-xorg-video-nvidia/testing 535.247.01-1 amd64
  NVIDIA binary Xorg driver

xserver-xorg-video-nvidia-legacy-340xx/unstable 340.108-25 amd64
  NVIDIA binary Xorg driver (340xx legacy version)

xserver-xorg-video-nvidia-legacy-390xx/unstable 390.157-12 amd64
  NVIDIA binary Xorg driver (390xx legacy version)

xserver-xorg-video-nvidia-tesla/unstable 525.147.05-15 amd64
  NVIDIA binary Xorg driver (Tesla version)

xserver-xorg-video-nvidia-tesla-418/unstable 418.226.00-18 amd64
  NVIDIA binary Xorg driver (Tesla 418 version)

xserver-xorg-video-nvidia-tesla-450/unstable 450.248.02-10 amd64
  NVIDIA binary Xorg driver (Tesla 450 version)

xserver-xorg-video-nvidia-tesla-460/unstable 460.106.00-20 amd64
  NVIDIA binary Xorg driver (Tesla 460 version)

xserver-xorg-video-nvidia-tesla-470/testing,unstable 470.256.02-6 amd64
  NVIDIA binary Xorg driver (Tesla 470 version)

xserver-xorg-video-nvidia-tesla-535/testing 535.216.03-3 amd64
  NVIDIA binary Xorg driver (Tesla 535 version)

xserver-xorg-video-nvidia-tesla-550/unstable 550.54.15-1 amd64
  NVIDIA binary Xorg driver (Tesla 550 version)

However, I have the feeling that no driver support this card --correct me if I’m wrong:

(mando@velvet) (~) $ apt show xserver-xorg-video-nvidia-* | grep MX
 GeForce GT 520MX [GF119M], GeForce GT 525M [GF108M], GeForce GT 530 [GF108],
 GeForce GTX 670MX [GK104M], GeForce GTX 675M [GF114M],
 GeForce GTX 675MX [GK104M], GeForce GTX 675MX Mac Edition [GK104M],
 GeForce GTX 680M [GK104M], GeForce GTX 680MX [GK104M],

I tried to install the default driver xserver-xorg-video-nvidia which install the driver 535.247.01-1. However running nvidia-smi or nvidia-settingshangs.

nvidia-bug-report.log.gz (144.4 KB)

This issue seems very related to this topic – almost the same card, same behavior.

I tried to search the right driver from nvidia.com but I can’t find my model.

  • Am I using the wrong driver?
  • What driver shall I use, if any?

Thanks,
Best regards

2

That driver should be fine. Looking at the debug log, it looks as though perhaps, (I am by no means an expert), it may be using the CPU graphics instead of the MX150.

Have you tried disabling that in the BIOS?

3

Have you tried disabling that in the BIOS?

There is no such option in the BIOS.

it may be using the CPU graphics instead of the MX150

Indeed, that’s the problem. Though, both the Intel and the Nvidia cards are correctly displayed in the BIOS/UEFI.

As I suspect that my GPU is not supported by a nvidia module, I tried to only install the nouveau driver, Wayland uses the Intel graphic card (despite the xserver-xorg-video-nouveauis installed)

(mando@velvet) (~) $ grep -i Intel /var/log/Xorg.0.log
[     6.769] (II) modeset(0): glamor X acceleration enabled on Mesa Intel(R) UHD Graphics 620 (KBL GT2)
[     7.741] (II) config/udev: Adding input device HDA Intel PCH Headset Mic (/dev/input/event18)
[     7.742] (II) config/udev: Adding input device HDA Intel PCH Headphone (/dev/input/event19)
[     7.744] (II) config/udev: Adding input device HDA Intel PCH HDMI/DP,pcm=3 (/dev/input/event20)
[     7.745] (II) config/udev: Adding input device HDA Intel PCH HDMI/DP,pcm=7 (/dev/input/event21)
[     7.746] (II) config/udev: Adding input device HDA Intel PCH HDMI/DP,pcm=8 (/dev/input/event22)
(mando@velvet) (~) $ grep -i nouveau /var/log/Xorg.0.log

Ideas are welcome :)

4

Here are some news!

  • I currently use Debian forky.
  • I’m now using nvidia-kernel-dkms (550.163.01-3), installed as follows:
     sudo apt update
 sudo apt install nvidia-driver
  • I had to generate the keys required by the secure boot:
    sudo apt install openssl
sudo mkdir -p /var/lib/shim-signed/mok
sudo openssl req -nodes -new -x509 -newkey rsa:2048 -outform DER -keyout /var/lib/shim-signed/mok/MOK.priv -out /var/lib/shim-signed/mok/MOK.der -days 3650 -subj “/CN=mando@velvet/”
sudo chmod 600 /var/lib/shim-signed/mok/MOK.\*
sudo mokutil --import /var/lib/shim-signed/mok/MOK.der
  • I also had to add the Microsoft Corporation UEFI CA 2011 certificate, because of this bug. This requires to prepare to store the MicCorUEFCA2011_2011-06-27.der file on FAT32 USB stick and add it from the BIOS interface.
    $ mokutil --db | grep Subject | grep Microsoft
        Subject: C=US, ST=Washington, L=Redmond, O=Microsoft Corporation, CN=Microsoft Corporation UEFI CA 2011
        Subject: C=US, ST=Washington, L=Redmond, O=Microsoft Corporation, CN=Microsoft Windows Production PCA 2011
        Subject: C=US, O=Microsoft Corporation, CN=Microsoft UEFI CA 2023
        Subject: C=US, O=Microsoft Corporation, CN=Microsoft Option ROM UEFI CA 2023
  • I installed bumblebee as explained here:
    sudo apt install bumblebee-nvidia primus-nvidia primus-vk-nvidia
    As explained here, I had to modify /etc/bumblebee/xorg.conf.nvidiaso that the Bus ID (otherwise, optirun complains):
    $ grep Bus /etc/bumblebee/xorg.conf.nvidia 
#   To get the BusID prop, run `lspci | egrep 'VGA|3D'` and input the data
BusID "PCI:01:00:0"
  • The NVIDIA card is correctly displayed when using optirun:
    $ optirun glxgears -info | grep GL_RENDERER
GL_RENDERER   = NVIDIA GeForce MX150/PCIe/SSE2
  • The NVIDIA card is correctly displayed in nvidia-smi and nvidia-settings if only if they are run through optirun, i.e.:
    optirun nvidia-settings &
optirun nvidia-smi
  • Currently, I still face some issue when using torch but it looks this is now purely related to my torch/CUDA installation, but the driver installation now looks OK.
    $ pip install torch
...
$ optirun ipython3
Python 3.13.9 (main, Oct 15 2025, 14:56:22) [GCC 15.2.0]
Type 'copyright', 'credits' or 'license' for more information
IPython 9.1.0 -- An enhanced Interactive Python. Type '?' for help.
Tip: The `%timeit` magic has a `-o` flag, which returns the results, making it easy to   plot. See `%timeit?`.

In [1]: import torch
   ...: assert torch.cuda.is_available()

In [2]: import torch
 ...: torch.zeros(1).cuda()
/home/mando/.local/lib/python3.13/site-packages/torch/cuda/__init__.py:283:   UserWarning: 
    Found GPU0 NVIDIA GeForce MX150 which is of cuda capability 6.1.
    Minimum and Maximum cuda capability supported by this version of PyTorch is
    (7.0) - (12.0)
  ....
1 Like