NVIDIA Unified Compute Framework (UCF)

UCF is a fully accelerated framework for developing real-time edge AI applications

Available Late 2022

Edge applications must process some combination of high-speed IO, data and signal processing, AI modalities, and computer graphics.

UCF allows developers to create low latency edge applications that process those pipelines in real-time.

Microservices

The Unified Compute Framework enables developers to combine optimized and accelerated microservices into real-time AI applications. Every microservice has a bounded domain context (vision AI, conversational AI, data analytics, graphics rendering) and can be independently managed and deployed within the application. The abstraction of each domain from the application alleviates the need for low-level domain and platform knowledge.

UCF includes validated deployment ready microservices to accelerate application development. New and custom microservices can be created using NVIDIA SDKs.

No-Code Design Tools

The Unified Compute Framework includes an IDE and a set of no-code design tools to create, manage, and deploy applications.

Designs are visualized as a combination of data processing pipelines. Using drag and drop operations, users can quickly create and combine these pipelines to build powerful applications that incorporate different AI modalities, graphics, and other processing functions. Built-in design rules and verification built into the IDE ensures that UCF applications are correct-by-construction.

Once complete, you can package UCF applications into containers and deploy easily using HELM charts.