Solved: Upgrading CUDA 11.6/11.6.2 to CUDA 11.7

Hi,

I upgraded a CUDA 11.6.2 installation to 11.7 using

sudo apt install cuda

After rebooting I started to notice the high fan noise reported in Solved: Fan noise on a Dell XPS 9510 and sure enough, 'Additional Drivers' showed that the nvidia-driver-515 was installed, but

$ nvidia-smi

showed that no drivers were loaded. After some trial and error I found that

$ sudo apt purge cuda
$ sudo reboot
...
$ sudo apt install cuda

resulted in a working nvidia-driver-515 installation:

$ nvidia-smi
Mon May 16 22:49:52 2022       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 515.43.04    Driver Version: 515.43.04    CUDA Version: 11.7     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  NVIDIA GeForce ...  On   | 00000000:01:00.0 Off |                  N/A |
| N/A   46C    P8     8W /  N/A |      4MiB /  4096MiB |      0%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|    0   N/A  N/A      2799      G   /usr/lib/xorg/Xorg                  4MiB |
+-----------------------------------------------------------------------------+

Regards,
GW

1 Like

The problem that the nvidia-drivers were not loaded seems to occur more often: after a kernel update and/or an nvidia update. As of now, executing the command:

$ sudo apt purge cuda

is not enough to trigger recompilation of the DKMS modules, you also have to execute the command:

$ sudo apt autoremove

Reboot and then execute the command:

$ sudo apt install cuda

You should now see 'Building for ..' messages. Reboot and check that the nvidia-driver is loaded by executing the command:

$ nvidia-smi

Regards,
GW

Hi,

For some reason I had to do this twice yesterday (purge, autoremove, reboot, install, reboot) before the nvidia-driver was shown as loaded.

Regards,
GW