My goal is to use OpenGL 4.6 in docker image based on nvcr.io/nvidia/l4t-base:r35.1.0.
The host Jetson NX has Jetpack 4.4.1 installed.
Is it possible?
(I’ve read a lot and have tried a lot, and currently fully confused, so would like to start from simple questions).
Even, simpler )
Is it possible to run container based on nvcr.io/nvidia/l4t-base:r35.1.0 (Jetpack 5.0.2) on a host with Jetpack 4.4.1 ?
There are some thoughts regarding this here: Is it possible to run Jetpack 5.* images on Jetpack 4.* host? · Issue #169 · dusty-nv/jetson-containers · GitHub
But I would ask to get more formal answer.
Though,it is possible to install CUDA 11.4 into container. It failed to run and complained on:
CUDA error at …/…/common/inc/helper_cuda.h:777 code=35(cudaErrorInsufficientDriver) "cudaGetDeviceCount(&device_count)
Your topic was posted in the wrong category. I am moving this to the Jetson Xavier NX category for visibility.
There might be some issue between the CUDA library and GPU driver which is integrated into the OS.
Is it possible to upgrade your environment with JetPack 5 first?
Thank you for support
Actually, the problem is that we cannot upgrade Jetpack 4.4.1 on our Jetson NX boards for some reasons (additional HW, updated BSP and etc.). We also are interested in using ROS 2, which requires Ubuntu 20.04. We thought, that using the container based on Jetpack 5.x will be a good solution for us.
Can you please confirm, that the only supported configuration for Docker on Jetson boards is when container is based on image, which has a full match with Jetpack on a host?
This is correct - you should use Docker images that match your JetPack-L4T version. You won’t be able to run JetPack 5.x container images on JetPack 4.x
Note that I have ROS2 Foxy/Galactic container images built for JetPack 4.4.1 / L4T R32.4.4 here:
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.