Innovation

Exclusive: AMD Reveals Shocking Plans for the Future of Graphics

15 July 2024

|

Zaker Adham

Last week in Los Angeles, at AMD’s Ryzen 9000 Tech Day, I delved into the Zen 5 architecture and upcoming CPUs. However, what stood out the most wasn’t the hardware but a comment about AI's future role in gaming. Sebastien Nussbaum, AMD’s chief architect for computing and graphics, suggested that AI could one day “dream up” the lighting in PC games. This bold vision of AI-driven game design immediately captured my interest.

To understand this concept better, I sat down with Chris Hall, AMD’s senior director of software development. He explained that although it sounds far-fetched, the integration of AI into game design is progressing faster than many realize.

Hall mentioned OpenAI’s video generator, Sora, which can create video from text prompts and understand 3D worlds despite not being trained for it. While we’re not there yet for games, the potential is clear. AMD and the gaming industry are already researching technologies that could lead to such advancements, though it requires a significant shift in the traditional graphics pipeline.

Hall highlighted the importance of AI in mundane tasks, such as Unreal Engine’s Neural Network Engine (NNE) for efficient cloth simulation. This application of machine learning frees up GPU resources for more impressive visuals, as seen with Nvidia’s Ray Reconstruction in games like Alan Wake 2. The real innovation is in these practical uses, which gradually push the boundaries of visual quality.

Despite the hype around AI, the challenge lies in the resources needed to run these models. Current frameworks, like NNE, are limited to CPUs and GPUs, and high-end features like Nvidia’s DLSS require expensive hardware. Nevertheless, companies like AMD, Nvidia, and Intel are laying the foundation for more efficient AI integration in games.

Hall emphasized that AI in gaming isn’t about flashy features but solving real problems in game development. It’s about enabling developers to achieve more with the same resources, ultimately pushing the medium forward without exorbitant costs. This gradual evolution could lead to AI-generated lighting and other advancements, providing a major leap in visual quality.