Modder enables Nvidia DLSS 5 Neural Rendering on AMD RDNA 4 GPUs
A new GitHub project leverages FP8 math support to run Nvidia's AI rendering on Radeon RX 9000 series hardware.
Nvidia's proprietary DLSS 5 Neural Rendering (NR) has successfully run on AMD hardware for the first time. Modder Daniel Blanco has developed a tool that enables the AI-driven rendering technology on AMD's RDNA 4 architecture, specifically demonstrating the feat on a Radeon RX 9070 XT.
The project, hosted on GitHub as 'DLSS-NR on AMD,' allows the technology to function in supported titles—such as Cyberpunk 2077 and GTA V: Enhanced—provided the user supplies a legitimately obtained DLSS 5 DLL. According to Blanco, current performance remains a work in progress, with the RX 9070 XT achieving roughly 28 frames per second at 1080p. The modder stated his goal is to continue optimizing the tool until it performs on par with an Nvidia RTX 5070 Ti.
The Technical Bridge
DLSS 5 Neural Rendering was designed for the RTX 50 series, though previous community efforts have already ported it to older Nvidia architectures. The leap to AMD hardware is possible because DLSS 5 NR relies on FP8 math support for its AI compute requirements. Because AMD's RDNA 4 architecture (found in the RX 9000 series) natively supports FP8 math, the hardware is capable of executing the necessary calculations despite the lack of Nvidia-specific Tensor cores.
Industry Implications
This development proves that the core functionality of DLSS 5's neural rendering does not depend on proprietary, Nvidia-only hardware features, but rather on standardized AI compute capabilities. By breaking the vendor lock, the mod demonstrates that the gap between Nvidia's closed ecosystem and AMD's open approach is narrower than previously thought. If the mod reaches a stable, high-performance state, it could provide AMD users with access to Nvidia's superior upscaling and rendering techniques, potentially pressuring AMD to accelerate the development of its own FidelityFX Super Resolution (FSR) suite.
What to Watch
While the proof-of-concept is successful, the mod remains unoptimized and requires manual installation of system files. Future updates to the 'DLSS-NR on AMD' repository will likely focus on performance gains to make the tool viable for daily gaming. It remains to be seen if Nvidia will attempt to block such ports through software updates or if this will encourage a shift toward more cross-vendor AI standards in the GPU market.