How to use pytorch 2.3 + torchaudio in driver os

Autonomous Vehicles DRIVE AGX Orin DRIVE AGX Orin General
1

Please provide the following info (tick the boxes after creating this topic):
Software Version
DRIVE OS 6.0.10.0
DRIVE OS 6.0.8.1
DRIVE OS 6.0.6
DRIVE OS 6.0.5
DRIVE OS 6.0.4 (rev. 1)
DRIVE OS 6.0.4 SDK
other

Target Operating System
Linux
QNX
other

Hardware Platform
DRIVE AGX Orin Developer Kit (940-63710-0010-300)
DRIVE AGX Orin Developer Kit (940-63710-0010-200)
DRIVE AGX Orin Developer Kit (940-63710-0010-100)
DRIVE AGX Orin Developer Kit (940-63710-0010-D00)
DRIVE AGX Orin Developer Kit (940-63710-0010-C00)
DRIVE AGX Orin Developer Kit (not sure its number)
other

SDK Manager Version
2.1.0
other

Host Machine Version
native Ubuntu Linux 20.04 Host installed with SDK Manager
native Ubuntu Linux 20.04 Host installed with DRIVE OS Docker Containers
native Ubuntu Linux 18.04 Host installed with DRIVE OS Docker Containers
other

Issue Description
I want to install torch 2.3 and torchaudio 2.3 on Orin Drive, with CUDA driver 12.1 and CUDA runtime 11.4. The official precompiled package that supports CUDA runtime 11.4 is torch 2.0 - python 3.8. Is there any solution to use python 3.10 with torch 2.3 and torchaudio 2.3?

Error String
< Please provide just the error messages here >

Logs
Provide logs in text box instead of image

Please paste the complete application log here. If there are multiple logs, please use multiple text box

3

Dear @wenhui.yang ,
We don’t officially support PyTorch installation on DRIVE. For development you can use host and on DRIVE you can use PyTorch → ONNX → TensorRT for efficient deployment.
You can check with Jetpack prebuilt libraries from https://developer.download.nvidia.com/compute/redist/jp/v60dp/pytorch/ to see if it helps?

1 Like
8

Hi there，

Thank you for your rely.

I got a Python project that I want to quickly deploy on a GPU. If I can’t use Torch, does Orin Drive have an official Python TensorRT installation package?

Additionally, my CUDA version is as follows. Can I upgrade the CUDA runtime to 12.1? How do I upgrade? Because when I try to install and import Torch 2.3, I get the following error: ”OSError: libcudart.so.12: cannot open shared object file: No such file or directory“

CUDA Driver Version / Runtime Version 12.1 / 11.4

9

does Orin Drive have an official Python TensorRT installation package?

Yes. 
nvidia@tegra-ubuntu:/usr/src/tensorrt/samples$ dpkg -l | grep TensorRT
ii  libnvinfer-bin                         8.6.13.3-1+cuda11.4                     arm64        TensorRT binaries
ii  libnvinfer-dispatch8                   8.6.13.3-1+cuda11.4                     arm64        TensorRT dispatch runtime library
ii  libnvinfer-lean8                       8.6.13.3-1+cuda11.4                     arm64        TensorRT lean runtime library
ii  libnvinfer-plugin8                     8.6.13.3-1+cuda11.4                     arm64        TensorRT plugin libraries
ii  libnvinfer-vc-plugin8                  8.6.13.3-1+cuda11.4                     arm64        TensorRT vc-plugin library
ii  libnvinfer8                            8.6.13.3-1+cuda11.4                     arm64        TensorRT runtime libraries
ii  libnvonnxparsers8                      8.6.13.3-1+cuda11.4                     arm64        TensorRT ONNX libraries
ii  libnvparsers8                          8.6.13.3-1+cuda11.4                     arm64        TensorRT parsers libraries
ii  python3-libnvinfer                     8.6.13.3-1+cuda11.4                     arm64        Python 3 bindings for TensorRT standard runtime
nvidia@tegra-ubuntu:/usr/src/tensorrt/samples$ python
Python 3.8.10 (default, Mar 17 2025, 19:35:05)
[GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorrt as trt
>>>
Please check samples at /usr/src/tensorrt/samples/python/. Please follow the ReadMe file for each sample to setup the needed files to run the sample and file a new topic incase of any issues?