NVIDIA Holoscan for Media

NVIDIA Holoscan for Media is the most advanced AI platform for live media, allowing live video pipelines to run on AI infrastructure. It revolutionizes application development for this rapidly evolving market by enabling NVIDIA-accelerated container orchestration for multi-vendor live production and AI. Developers can take advantage of templates and reference services for building and deploying applications.

What's Included

This collection contains Holoscan for Media install files and the information needed to use them. These resources include our reference deployment guide, automation scripts, reference applications (Helm charts and containers), as well as links to information about the NVIDIA SDKs and application frameworks this platform utilizes.

See what's new in the latest Release Notes.

Getting Started

Install Files

• Platform Setup – Ansible playbooks to install and configure a Red Hat OpenShift compact cluster for Holoscan for Media in production.
• Local Developer Setup with Kubernetes – Ansible playbooks to install and configure Kubernetes for Holoscan for Media application developers using NVIDIA Cloud Native Stack.

Prerequisites

• NVIDIA Rivermax license
  • Rivermax is an optimized networking SDK that leverages ConnectX and BlueField DPU hardware-streaming acceleration technology and supports GPUDirect, in bare metal and virtualized environments. It is used by Holoscan for Media applications to comply with the stringent timing and traffic flow requirements of ST 2110. A Rivermax license is required to use this functionality.
  • Request a developer license at Rivermax Download.
  • To run a Rivermax application container on a Kubernetes cluster, follow the instructions in the Media Gateway Next Helm chart documentation to mount the file via a Secret.
  • To run a Rivermax application container with Docker, follow the instructions in the Local Developer Setup with Docker guide to mount the file from the host.

Reference Applications

• Helm Dashboard and Helm chart – Helm Dashboard is an open-source project which offers a UI-driven way to view, manage and install Helm charts.
• NMOS Registry and Helm chart – Container implementation of the AMWA Networked Media Open Specifications (NMOS) providing a complete NMOS Registry, a JavaScript-based NMOS Client/Controller and an embedded NMOS Virtual Node. For more information please refer to http://amwa.tv/.
• NMOS Controller and Helm chart – This container provides a browser-based application that uses NMOS APIs to discover, connect and manage resources within a networked media system. For more information on the Networked Media Open Specifications, see https://www.amwa.tv/nmos-overview.
• Media Gateway Next and Helm chart – Media Gateway Next is a reference container for Holoscan for Media, built on DeepStream. It provides NMOS registration and control of ST 2110 sinks and sources in a flexible GStreamer pipeline passed to the container. The source code can be downloaded from the associated Media Gateway Next resource.
• Chrome Remote Desktop Helm chart - Set up remote access to a desktop environment running on the cluster.

Documentation

• Platform Reference Deployment Guide - Detailed instructions to install and configure an OpenShift cluster. The source files included in the document can be downloaded from the associated Platform Setup resource.
• Local Developer Setup with Kubernetes - Get all the Holoscan for Media containerized applications running under Kubernetes on a single Linux host* with ConnectX network adapter. The source files included in the document can be downloaded from the associated Local Developer Setup resource.
• Local Developer Setup with Docker - Get all the Holoscan for Media containerized applications running under Docker on a single Linux host* with ConnectX network adapter. The source files included in the document can be downloaded from the associated Local Developer Setup resource.
• Developer Starting Guide - Covers Holoscan for Media application development, and includes useful links to get you started.
• Answers to some other Frequently Asked Questions.

*Note: Local developer setups do not provide an NVIDIA-supported Holoscan for Media configuration for production. Such a system can be used for some development and debugging but it is not recommended for performance and stream compliance testing, which should be undertaken using several machines, or with a full 3-node compact cluster as described in the Holoscan for Media Platform reference deployment guide. Capabilities of the Holoscan for Media platform such as PTP, resource slicing, Helm orchestration, and other functions are not provided in the local dev setup with Docker.

Release Notes

Production-Ready Platform Release

Announcement:

• Live Media Reimagined: NVIDIA Holoscan for Media Now Available for Production
• Transform Live Media Pipelines with NVIDIA Holoscan for Media

Expanded Developer Availability Release

Announcement: Next-Generation Live Media Apps on Repurposable Clusters with NVIDIA Holoscan for Media

Early Access Update 2

Announcement: Next-Generation Live Media Apps on Repurposable Clusters with NVIDIA Holoscan for Media

Early Access Update 1

Announcement: Streamline Live Media Application Development with New Features in NVIDIA Holoscan for Media

Resources

Support

• Ask the community or get in touch with NVIDIA specialists via the Welcome Page.

Additional Information

• Network Operator and Helm chart - The NVIDIA Network Operator simplifies the provisioning and management of NVIDIA networking resources in a Kubernetes cluster. The operator automatically installs the required host networking software - bringing together all the needed components to provide high-speed network connectivity. These components include the NVIDIA networking driver, Kubernetes device plugin, CNI plugins, IP address management (IPAM) plugin and others. The NVIDIA Network Operator works in conjunction with the NVIDIA GPU Operator to deliver high-throughput, low-latency networking for scale-out, GPU computing clusters.
• GPU Operator and Helm chart - The NVIDIA GPU Operator manages NVIDIA GPU resources in a Kubernetes cluster and automates tasks related to bootstrapping GPU nodes. Since the GPU is a special resource in the cluster, it requires a few components to be installed before application workloads can be deployed onto the GPU. These components include the NVIDIA drivers (to enable CUDA), Kubernetes device plugin, container runtime and others such as automatic node labeling, monitoring and more.

License

By downloading and using this software, you accept the terms and conditions of the NVIDIA AI Product Agreement.