Jetson 4.2.1 Tx2 support for nvidia-docker tensorrt image and docker caffe image

I have flashed the latest version of jetpack 4.2.1 on tx2 board. nvidia-docker is running fine on the tx2 board. However I am facing issue w.r.t the running of nvidia caffe docker images as well as the tensorrt images.

I used the following commands for the docker to install caffe image.
docker pull nvcr.io/nvidia/caffe:19.06-py2

The above image throws error “exec format error” while executing on the tx2 board.

The issue persists for the tensorrt image as well.
docker pull nvcr.io/nvidia/tensorrt:19.06-py3

Can someone confirm whether we have images available for arm architecture for the caffe in ngc.nvidia website?

Hi,

We are checking this issue with our internal team.
Will update information with you once we got any feedback.

Thanks.

Hi,

Sorry for keeping you waiting.
You can follow this page for using the container on Jetson:
https://github.com/NVIDIA/nvidia-docker/wiki/NVIDIA-Container-Runtime-on-Jetson

The exec format error indicates that you are using an x86 image on the ARM system.
However, we don’t have an immediate plan to enable these x86-based images on Jetson.
But you can use an Jetson image on an x86-based machine with the instructions shared in the wiki above.

Thanks.