A New Frontier for Radeon Users
In a surprising turn of events, enthusiasts have discovered a way to leverage Nvidia’s DLSS 5 Neural Rendering technology on AMD Radeon graphics cards. A recently released homebrew application on GitHub now enables this feature on non-Nvidia hardware. While the capability to run this tech on Radeon silicon is an impressive technical achievement, early testing suggests it places a massive burden on existing AMD hardware.
The Installation Process
The solution, provided by a user named danielblnc, involves placing specific files into a game's directory—such as the bin/x64 folder for Cyberpunk 2077—alongside the necessary nvngx_dlssnr.dll. While the setup may trigger security warnings, initial reports suggest these are false positives. However, getting the software to run consistently remains a challenge; users may encounter black screens, stability issues, or the need to bypass launchers like Steam to execute the game directly from the executable file.
Performance Realities: The Hardware Hurdle
The primary issue lies in the fundamental architecture of modern GPUs. DLSS 5 is engineered specifically for the specialized Tensor Cores found in GeForce RTX series cards. As one tester noted regarding their experience on an RX 9070 XT:
«Running at 4K RT Ultra settings was already demanding, yielding 61 fps with FSR 4 set to Balanced. Enabling the DLSS Neural Rendering toggle caused that figure to plummet to just 9 fps.»
This results in an approximately 85% drop in performance, a figure that is effectively unplayable for most gaming scenarios. Even with some optimization, the current implementation struggles due to a lack of dedicated matrix calculation silicon on RDNA 4 chips compared to Nvidia's advanced hardware.
Future Outlook
Despite the current performance deficit, there is incremental progress. Observers have noted that the software has already seen a 12% performance improvement since its initial release. However, most experts suggest that while the community-driven port is a fascinating experiment for taking high-quality screenshots, the hardware limitations of the current generation of Radeon cards likely prevent Neural Rendering from becoming a viable, everyday feature for AMD users.