My Jetson AGX Orin 64GB Developer Kit GPU Don't Dectected and Nvidia-smi can't dectected GPU and jetpack not dectectd

Robotics & Edge Computing Jetson Systems Jetson AGX Orin
,
1

I stuck installed everything I can think of that might be needed but i can’t solving using GPU and i think something wrong I try Flash 3 times for refresh and one thing i try i install nvidia-driver after i reboot my jetson don’t detected Display OS: Ubuntu 22.04
because i want to run model LLM i try ollama and docker nvidia-container-runtime error in docker run –gpus
my capture screen
nvidia-smi:

Screenshot from 2025-10-29 14-06-09
Screenshot from 2025-10-29 14-06-09807×356 29.2 KB

jtop:

Screenshot from 2025-10-29 14-14-53
Screenshot from 2025-10-29 14-14-531152×631 87.7 KB

nvcc:
NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Wed_Aug_14_10:14:07_PDT_2024
Cuda compilation tools, release 12.6, V12.6.68
Build cuda_12.6.r12.6/compiler.34714021_0

Thing I do:

  • I try install nvidia-driver (after reboot my Display out of range)
  • I try solution /usr/local/cuda/samples/1_Utilities/deviceQuery but i don’t have in my path ( same cuda12.6 dir )
  • I try install pytorch and pytorch to available devices and false and not dectected gpu and cuda
  • i try change cuda path export in .bashcr and cannot dectected

Thank you in advance for your advice and help.

5

Hi,

  1. Jetson’s GPU driver is integrated in the OS so you don’t need to manually install it.

  2. CUDA sample can be found in the link below, please check out the v12.5 branch for JetPack 6.2.1:
    GitHub - NVIDIA/cuda-samples: Samples for CUDA Developers which demonstrates features in CUDA Toolkit · GitHub

  3. Packages with CUDA support are available in the link below:
    jp6/cu126 index

  4. Below are the environment commands:

$ export LD_LIBRARY_PATH=/usr/local/cuda-12.6/lib64:$LD_LIBRARY_PATH
$ export PATH=/usr/local/cuda-12.6/bin:$PATH

Please use --runtime nvidia to enable GPU functionality in Docker.
Then you will also need to add a /etc/docker/daemon.json file.

Please refer to the document below for the details:

Thanks.

6

Thanks for the head-up. I have know about problem i install everything not support for Jetson
and I use --runtime nvidia then add a /etc/docker/daemon.json file. That work read and used
jetson-containers in repo: https://github.com/dusty-nv/jetson-containers?tab=readme-ov-file
Now In container have gpu and running ollama with gpu right now.