Jetson Orin NX CUDA initialization failure with error: 35

I set up my jetson orin nx using jetson linux R35.3.1 aarch64 deb package but after installation the jetson jetpack does not come with cudnn and cuda therefore i went to install cudnn-local-repo-ubuntu2004- and cuda-tegra-repo-ubuntu2004-12-0-local_12.0-1_arm64.deb.
i also install nv-tensorrt-local-repo-ubuntu2004-8.6.1-cuda-12.0_1.0-1_arm64.deb with tensorflow=2.12.0

for context i am trying to run a machine learning program using onnx model. when i ./main to run the program
this error occur

WARNING: Unable to determine GPU memory usage
WARNING: Unable to determine GPU memory usage
WARNING: CUDA initialization failure with error: 35. Please check your CUDA installation: NVIDIA CUDA Installation Guide for Linux
Segmentation fault (core dumped)

i could not find ways to fix it.
anyone can help me to fix this error?
thanks in advance.


Welcome to the forums! I’m moving this topic to the Jetson category for better visibility.


The package shared on the cuDNN and TensorRT website is not for Jetson.
You can find it from SDKmanager or this link:

More, how do you flash the device?
If you use SDKmanager to flash, you can install all the components directly.


thanks for the advice. i have install it using sdk manager and everything is installed properly.
I have another problem which is that when i connect the board to some monitors there is no connection between the board and the monitor. Is there a certain limitation on the resolution of the monitor that the board can support? I am using a DP to hdmi connector however DP to DP works.

I have a orin nano board which can be plug into any monitor.


Thanks for the feedback. Glad to know re-flashing fix the issue.
For the monitor’s problem, would you mind filing a new topic?


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