Hi Team,
I am setting up a new DGX Spark system to run NVIDIA Aerial (AI‑RAN) simulations for a proof‑of‑concept project. This is my first time configuring DGX Spark and NGC access, so I am following the official documentation.
Steps I have completed so far
-
Logged into the DGX Spark machine (Ubuntu-based system).
-
Tried verifying GPU access with:
docker run --rm --gpus all nvidia/cuda:12.2.0-base nvidia-smi→ The image was not found locally, so it failed.
-
Created an NGC account and generated an API key/token.
-
Logged into NGC using:
docker login nvcr.io Username: $oauthtoken Password: <my NGC API key>→ Docker reported “Login Succeeded”.
-
But
~/.docker/config.jsonremains empty (no auth token stored). -
When attempting to pull Aerial:
docker pull nvcr.io/nvidia/aerial/aerial:latestI get this error: denied: you are not authorized to access this resource
Current Issue
-
I created the NGC account myself, and I have not been added to any specific NVIDIA enterprise organisation or Aerial entitlement group.
-
This DGX Spark is new, and I want to run Aerial simulations for demonstration and testing.
-
It is unclear whether DGX Spark automatically includes Aerial access or if additional entitlement is required.
Request for NVIDIA Support
Could you please clarify:
-
Do I need to be added to any special NVIDIA Aerial entitlement group to access:
nvcr.io/nvidia/aerial/aerial ? -
Is Aerial available for DGX Spark users by default, or does an admin need to provide enterprise access?
-
Why does
docker login nvcr.ionot write credentials intoconfig.json? -
What are the correct steps to authenticate and pull Aerial containers on DGX Spark?
Any guidance or documentation links would be greatly appreciated.
Thank you!