In the latest development of GPU technology, AMD is stepping up to challenge NVIDIA’s DLSS with a new AI-powered approach to ray tracing. NVIDIA’s DLSS 3.5 has already introduced Ray Reconstruction technology, designed to rebuild the complex lighting system in ray-traced games using artificial intelligence. This feature enhances both performance and visual quality by reconstructing the light from a limited number of rays.
Today, even top-tier GPUs, like NVIDIA’s RTX 4090, face challenges in rendering games with a high number of ray-traced rays. Since this limitation results in a noisy, incomplete image, advanced techniques like DLSS 3.5 are crucial to clean up the noise and improve the quality. This is exactly what DLSS 3.5 achieves with its AI-driven denoising, making games look more refined and lifelike.
AMD’s Response
Until now, AMD has not had an equivalent technology to compete with DLSS. However, through its GPUOpen platform, AMD is introducing a concept it calls “Neural Supersampling and Denoising for real-time path tracing.” This new approach leverages AI to enhance the visual quality of images generated by a limited number of rays. It applies AI to denoise and rescale images, creating a smoother and higher-quality gaming experience, much like NVIDIA’s DLSS 3.5 but with AMD’s own twist.
AMD’s technology is designed to be compatible with RDNA architecture GPUs, which could make it an accessible option for AMD users. Unlike NVIDIA’s DLSS, AMD’s solution is free and open-source. The company suggests this technology could play a pivotal role in improving real-time path tracing.
While this breakthrough is exciting, gamers and developers alike may need to wait until the release of FSR 4 or another advanced version to see AMD’s denoising in action. Until then, AMD’s announcement via GPUOpen gives a promising look at what’s to come.
