Skip to content

NVIDIA NIM

The table below lists the topics included in the NVIDIA NIM section:

Topic Description
NVIDIA NIM air-gap deployment guide Describes how to deploy NVIDIA Inference Microservice (NIM) in environments without direct internet connectivity ("air-gapped" clusters)
Air-gap cache processing script Python utility to stage a local NIM cache and optionally upload model profiles to S3-compatible object storage.
Air-gap misc configuration Optional tooling and cluster configuration for air-gapped NIM, including the Crane CLI and Trust Manager public CA bundles.
Air-gap troubleshooting Common errors when running NIM without internet access, with symptoms, causes, and resolutions.
NVIDIA NIM configuration Describes the NVIDIA NIM microservices included in NVIDIA AI Enterprise, which are a set of easy-to-use microservices for accelerating the deployment of foundation models.
GPU support matrix NVIDIA-published minimum and optional GPU configurations validated for each NIM container image and version.
GPU resource bundles How bundle labels map to GPU node types, counts, FP16 throughput, and memory on major managed Kubernetes platforms.
GPU bundle validation Which NIM models are validated or estimated on standard GPU bundles for EKS, AKS, OpenShift, restricted network EKS, and OKE.