Software Stack Installation
The Clara Holoscan SDK requires a specific software stack to build and run. The SDK can either run on a Development Stack for Clara Developer Kits based on Holopack and for x86 Linux compute platform or on a Deployment Stack for Clara Developer Kits based on OpenEmbedded.
The Deployment Stack is recommended if you want to include just software components that are actually needed for your application. Furthermore, the runtime BSP can be easily optimized with respect to memory usage, speed, security and power requirements.
Development Software Stack with Holopack on Clara Developer Kits
Installation of the Clara Holoscan SDK is automated via the NVIDIA SDK Manager. See the SDK Manager documentation for details on how to [install and use the SDK Manager to install the Clara Holoscan SDK] (https://docs.nvidia.com/sdk-manager/install-with-sdkm-clara). Make sure you have joined the Clara Holoscan SDK Program and, if needed, the RiverMax SDK Program before using the NVIDIA SDK Manager.
Development Software Stack on x86
Since v0.3.0, the Clara Holoscan SDK is also available for x86_64 (amd64) Linux platform and has been exclusively tested on Ubuntu 20.04. Please refer to the prerequisites located in the Clara Holoscan SDK source code
Deployment Software Stack with OpenEmbedded on Clara Developer Kits
NVIDIA Clara Holoscan accelerates deployment of production-quality applications by providing a set of OpenEmbedded build recipes and reference configurations that can be leveraged to customize and build Holoscan-compatible Linux4Tegra (L4T) embedded board support packages (BSP).
Holoscan OpenEmbedded/Yocto recipes add OpenEmbedded recipes and sample build configurations to build BSPs for NVIDIA Clara Developer Kits that feature support for discrete GPUs (dGPU), AJA Video Systems I/O boards, and the Clara Holoscan SDK. These BSPs are built on a developer’s host machine and are then flashed onto a Clara Developer Kit using provided scripts.
There are two options available to set up a build environment and start building Holoscan BSP images using OpenEmbedded.
The first sets up a local build environment in which all dependencies are fetched and installed manually by the developer directly on their host machine. Please refer to the Holoscan OpenEmbedded/Yocto recipes README for more information on how to use the local build environment.
The second uses a Holoscan OpenEmbedded/Yocto Build Container that is provided by NVIDIA on NGC which contains all of the dependencies and configuration scripts such that the entire process of building and flashing a BSP can be done with just a few simple commands.
Once the software stack has been installed on your device, Clara Holoscan SDK can be installed from source or using a pre-built container runtime.
Using the container from NGC
The Clara Holoscan Sample Applications multi-arch container (for arm64 and amd64/x86_64) is the simplest way to run the sample applications as it includes all necessary binaries and datasets, and allows for some customization of the application graph and its parameters.
Refer to the overview of the container on NGC for prerequisites, setup, and run instructions.
The sample applications container from NGC does not include build dependencies to update or generate new extensions, or to build new applications with other extensions. Refer to the section below to do this from source.
The Clara Holoscan SDK source code provides reference implementations for the GXF extensions and the sample applications, as well as infrastructure for building the current extensions and applications or your own.
Refer to the top-level
README.md in the open-source repository on Github for
prerequisites, setup, and run instructions.