CVE Patching and Release Cycle for nvidia/cuda Ubuntu-based Images

Accelerated Computing CUDA CUDA Setup and Installation
, ,
1

I’m trying to understand the release and patching strategy for NVIDIA’s Ubuntu-based CUDA images, specifically the cudnn-runtime variants.

From reviewing the Docker Hub tags (nvidia/cuda), it appears that new image versions are typically released every couple of months. However, I would like to confirm if there’s a defined release cadence for these updates.

In particular, I’m interested in how vulnerabilities in the Ubuntu base image are handled. For example, the image nvidia/cuda:12.9.0-cudnn-runtime-ubuntu24.04 is currently available, but when scanned with common vulnerability scanning tools, it shows multiple medium and low severity issues inherited from the base image.

Could you please clarify:

  1. How frequently are base image vulnerabilities addressed in the cudnn-runtime Ubuntu images?
  2. Is there a schedule or trigger for rebuilding and republishing images with patched base layers?
  3. What is the best practice for consumers to stay updated on security fixes in these images?

This information would help us make informed decisions about using and maintaining Ubuntu-based CUDA images in our environments.

Thanks in advance!