How to use pytorch with Jetpack 5.1.2 in docker container

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin Nano
,
1

Hello.
I use Jetson Orin Nano which is installed JP6.2, and I use docker container JP5.1.2
I already tried to install pytorch 2.1.0, but it was printed “print(torch.cuda.is_available())”.
Another community mentioned that pytorch2.1.0 and torchaudio0.1.6 are compatible with jetpack5.1.2, so I installed the same ones
I will attach the result of “pip3 list”.
If anyone has solved this problem, I would appreciate a reply.
Thanks for taking the time to read this.

3

Hi,

Due to some driver dependencies, we recommended using the JetPack 6 container on the JetPack 6 environment.
If you prefer the JetPack 5.1.2 environment, please reflash your system into JetPack 5 as well.

Thanks.

1 Like
4

Thank you for your reply.
I reinstalled JP5.1.2, and I was able to successfully create the Docker image within that environment.
I appreciate your help in resolving the issue.

6

Do you now have cuda support in the docker container Jetpack 5.1.2?
I am using Jetpack 6.1 and am looking for a pytorch docker with cuda support.

I tried this “nvcr.io/nvidia/pytorch:21.10-py3” but I am getting this

root@c8e3f7a7321b:/workspace# python
Python 3.8.12 | packaged by conda-forge | (default, Sep 29 2021, 20:54:09)
[GCC 9.4.0] on linux
Type “help”, “copyright”, “credits” or “license” for more information.

import torch
print(torch.cuda.is_available())
False

7

Hi,

For JetPack 5 environment, please find l4t-pytorch container instead:

Thanks.