This is a follow up to this question about there being no l4t-base:r36.3.0 for Jetpack 6.0.
I have an Orin AGX (although I also have an Orin NX 16GB I will use) with Jetpack 5.1.2 on it and am trying to build a container based on JP 6.0. Per the above forum post since there is no l4t-base:r36.3.0 I tried nvcr.io/nvidia/l4t-cuda:12.2.12-runtime as that seemed to be the equivalent of l4t-base with the cuda version matched up to jetpack 6.0 (although I guess there doesn’t seem to be a way to tell if it actually is jetpack 6.0.)
When trying to do anything with that container, however, I get this error:
$ docker run -it --rm nvcr.io/nvidia/l4t-cuda:12.2.12-runtime bash
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: nvidia-container-runtime did not terminate successfully: exit status 1: time="2024-10-14T14:52:35-04:00" level=error msg="failed to create NVIDIA Container Runtime: failed to construct OCI spec modifier: requirements not met: unsatisfied condition: cuda>=12.2 (cuda=11.4)"
: unknown.
For the particular container I need, I don’t actually need cuda, I am just looking for an equivalent to l4t-base. Strangely, the image nvcr.io/nvidia/l4t-jetpack:r36.3.0 appears to work fine, but it’s a bit bigger than I would like for this application.
There’s no reason in particular I’m trying to run JP 6.0 from Jetpack 5.1.2 other than I’m trying to get ahead on moving to jetpack 6.0 but not ready to actually update my board. Is there another l4t-base docker image alternative for jetpack 6.0?