search

dusty-nv / jetson-containers

Machine Learning Containers for NVIDIA Jetson and JetPack-L4T
MIT License
1.93k stars 422 forks source link

Guidanced Needed in Understanding and Downloading Correct Version of ONNXRUNTIME-GPU and CUDA #395

Closed HShamimGEHC closed 4 months ago

HShamimGEHC commented 5 months ago

I am still a beginner with jetson containers and am trying to understand things conceptually.

I have docker image that is my own image that I want to run as a container utilizing the GPU on my Jetson AGX Xavier. For that, I need to modify the Dockerfile for the image to install Cuda and onnxruntime-gpu, whose versions should match and be compatible with the nvidia driver I have.

It is my understanding that I cannot build the onnxruntime and cuda containers in this repo and use them for my image. So how can I ensure that I am downloading the right versions in the dockerfile for my image?

Guidance would greatly be appreciated. Thank you!

nvidia@ubuntu:~$ cat /etc/nv_tegra_release R35 (release), REVISION: 4.1, GCID: 33958178, BOARD: t186ref, EABI: aarch64, DATE: Tue Aug 1 19:57:35 UTC 2023

nvidia@ubuntu:~/uswebapp/jetsonserver$ dpkg-query --show nvidia-l4t-core nvidia-l4t-core 35.4.1-20230801124926

dusty-nv commented 4 months ago

@HShamimGEHC IIRC the onnxruntime dockerfile from this repo currently builds onnxruntime 1.16, however if it's easier for you the appropriate wheels for your environment can just be installed from https://elinux.org/Jetson_Zoo#ONNX_Runtime

HShamimGEHC commented 4 months ago

Thank you - I ended up using https://elinux.org/Jetson_Zoo#ONNX_Runtime as I was running out of storage when trying to install it via the repo.