Arm Neural Graphics SDK for Game Engines: Bringing AI-Powered Graphics to Mobile

The SDK is an open-source, engine-agnostic library. It’s part of the Arm Neural Graphics Progress Kit.

This SDK helps developers integrate Arm neural graphics technologies into their preferred game engines – whether they’re in-house, customized, or commercially available – to enable AI-powered graphics in mobile gaming.

The SDK is derived from the AMD FidelityFX SDK, ensuring seamless compatibility with its API design. This similarity lowers the adoption barrier for developers. Its layered and modular framework enables smooth implementation and integration of Arm neural graphics technologies.

These technologies include Neural Super Sampling (NSS), and future use cases like Neural Frame Rate Upscaling (NFRU) and Neural Super Sampling Denoising (NSSD). This allows developers to integrate only the components they need.

In the Arm Neural Graphics SDK for Game Engines (Fig 1), use case providers reside in the API Layer. they act as the primary interface between the SDK and game engines.The providers expose the functionality of specific use cases, such as neural super sampling or neural frame rate upscaling, by linking the upper-level API with the underlying component layer.

The component layer contains the algorithmic implementations for each use case, including neural network inference, shader logic, and runtime orchestration.

Built on the Vulkan backend – a modern, cross-platform graphics API – the SDK provides a stable interface across supported environments. This design enables seamless integration, simplifies upgrades, and promotes long-term maintainability for developers.

NSS is Arm’s first open neural graphics technology. It uses machine learning to upscale lower-resolution frames into high-quality visuals.

It goes beyond conventional shader-based upscaling with neural networks, enabling better image fidelity and computational efficiency across mobile platforms.