Neural Dawn Showcases the Future of Gaming on Android: Ray Tracing, AI, and Advanced Lighting on Future Arm Mali GPUs
Arm
Arm has showcased one of the first concrete applications of its neural rendering technologies for mobile devices. In collaboration with the British studio Sumo Digital, the company presented Neural Dawn, a video game developed for Android that will serve as a technological showcase for the next generation of Mali GPUs equipped with dedicated neural accelerators.
The title will exclusively arrive on future Android smartphones equipped with the upcoming Mali GPUs and represents the first mobile game to utilize Unreal Engine MegaLights, the new lighting system introduced by Epic Games with Unreal Engine 5.6.1. This announcement fits into Arm's strategy to introduce neural graphics, an approach that combines traditional rendering with neural networks to enhance image quality and energy efficiency. The stated goal is to overcome one of the main limitations of mobile gaming: the need to maintain high graphic quality without compromising battery life, operational temperatures, and consumption.
At the center of this evolution will be dedicated neural accelerators that will debut in the upcoming Mali GPUs. These units will be used to run technologies such as Neural Frame Rate Upscaling (NFRU), a form of frame generation for mobile devices, and Neural Super Sampling and Denoising (NSSD), a solution that combines upscaling and denoising through artificial intelligence to make advanced ray tracing and path tracing effects feasible in real-time.
According to Arm, these technologies help reduce the computational load required by the graphics subsystem, freeing up resources for more complex visual effects. The company also claims that Neural Super Sampling (NSS), an evolution of the current Arm Accuracy Super Resolution, can halve the load on the GPU through the use of trained neural networks instead of traditional algorithms.
The Neural Dawn project was created as a concrete demonstration of these capabilities within a regular production workflow. The game offers about two hours of content spread over four levels and follows the story of a scientist engaged in exploring a vast network of caves in search of the causes of a civilization's collapse. Light is a central element for both the artistic direction and gameplay, guiding the player in exploration and interaction with the environment.
For this reason, the development team chose to base much of the experience on the Unreal Engine MegaLights system. The technology allows managing a very high number of dynamic light sources with soft shadows and ray-traced lighting while maintaining a relatively predictable impact on performance. Unlike traditional approaches, MegaLights uses a shared lighting path based on ray tracing rather than processing each light source separately. This approach offers greater creative freedom to developers and makes it possible to utilize lighting as a narrative and gameplay element, rather than just a backdrop component.
However, some technical compromises remain: the system calculates a fixed number of light samples per pixel and favors the most important sources. In particularly complex scenes, it may therefore increase reliance on denoising systems, leading to possible image softening or ghosting effects.
According to Sumo Digital, the use of Arm's neural technologies has enabled significant energy savings, making it possible to activate MegaLights and ray tracing simultaneously in a mobile context. The company describes this shift as a significant change for smartphone game development, as it would reduce some of the constraints that have historically limited the technical and artistic ambition of mobile titles.
Another element highlighted by Arm pertains to integration into existing development workflows. Neural graphics capabilities are accessible through plugins for Unreal Engine and do not require proprietary rendering pipelines or deep modifications to the tools used by studios. Neural Dawn itself was developed by a team of 17 people over about 18 months.
Alongside the project, Arm announced an update for the Neural Graphics Development Kit expected in July, which will include support for NFRU and new resources dedicated to NSS. The company also released a practical guide called Neural Technology Playbook and opened an Early Access program for developers interested in experimenting with these technologies in advance.