Driver errors when installing cuda

Performed this operation on the Nvidia Xavier AGX to install cuda
Operating System Linux
Architecture arm64-sbsa
Compilation Native
Distribution Ubuntu
Version 18.04
Installer Type dev(local)

– My software Versions
Python 3.6
torch 1.9
tensorflow 2.6.0+nv21.9
numpy 1.19.5
torchvision and audio 0.10.0
keras 2.7

Performed these steps

– wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/sbsa/cuda-ubuntu1804.pin
—sudo mv cuda-ubuntu1804.pin /etc/apt/preferences.d/cuda-repository-pin-600
—wget https://developer.download.nvidia.com/compute/cuda/11.5.0/local_installers/cuda-repo-ubuntu1804-11-5-local_11.5.0-495.29.05-1_arm64.deb
– sudo dpkg -i cuda-repo-ubuntu1804-11-5-local_11.5.0-495.29.05-1_arm64.deb
– sudo apt-key add /var/cuda-repo-ubuntu1804-11 5-local/7fa2af80.pub
– sudo apt-get update
– sudo apt-get -y install cuda

Once I do the install cuda I get these errors

The following packages have unmet dependencies:
cuda-drivers-495 : Depends: libnvidia-compute-495 (>= 495.29.05) but it is not going to be installed
Depends: libnvidia-gl-495 (>= 495.29.05) but it is not going to be installed
Depends: nvidia-utils-495 (>= 495.29.05) but it is not going to be installed
libcudnn8-dev : Depends: libcudnn8 (= 8.2.4.15-1+cuda10.2) but 8.2.1.32-1+cuda10.2 is to be installed
libnvidia-decode-495 : Depends: libnvidia-compute-495 (= 495.29.05-0ubuntu1) but it is not going to be installed
nvidia-compute-utils-495 : Depends: libnvidia-compute-495 but it is not going to be installed
nvidia-driver-495 : Depends: libnvidia-gl-495 (= 495.29.05-0ubuntu1) but it is not going to be installed
Depends: libnvidia-compute-495 (= 495.29.05-0ubuntu1) but it is not going to be installed
E: Unmet dependencies. Try ‘apt --fix-broken install’ with no packages (or specify a solution).

Some things that may help in debugging

import torch

torch.version
‘1.9.0’
torch.cuda.is_available()
False
torch.version.cuda
torch.backends.cudnn.enabled
True
torch.cuda.device_count()
0

the command nvidia-smi does not work

The cmd ‘apt fix broken’ does not work. I tried many things on the forums but none work. Is the cuda verison 11.5 good for the Xaiver AGX or should i specify another version

Thank You

FYI, this is the Nano forum, you should probably use the Xavier AGX forum.

You cannot use the desktop PC version of the driver for discrete GPUs (dGPU). These were designed for PCI devices, but the GPU in a Jetson is integrated directly to the memory controller (iGPU). Details such as “nvidia-smi” are only available for PCI-based dGPUs.

The only place you should get CUDA from is JetPack/SDK Manager. This has the correct release of the correct driver for the version of L4T that is flashed to the Jetson. I am guessing you’ll have to flash the system again.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.