Modder enables DLSS Neural Rendering on AMD RDNA 4

Jackson Turner 09/08/2026

The Cross-Platform Quest: How a Modder Brought Nvidia‘s DLSS 5 to AMD GPUs

In the rapidly evolving world of graphics technology, where hardware rivalry often dictates feature exclusivity, a remarkable feat has been achieved. A dedicated modder has successfully cracked the code, porting Nvidia’s cutting-edge DLSS 5 Neural Rendering technology to AMD GPUs for the very first time. This achievement proves that the power of AI-enhanced image upscaling is no longer locked behind a single manufacturer, opening up exciting new possibilities for the entire PC gaming community.

The magic behind this cross-platform compatibility was unlocked by developer danielblnc, who created a custom modification. This tool, known as DLSS-NR for AMD, utilizes Nvidia’s core neural rendering model, integrated via a custom installer and the nvngxdlssnr.dll file, allowing RDNA 4 graphics cards to leverage this advanced technology in supported DirectX 12 games.

How does this technical trick work? The process involves placing the necessary files directly into a game’s bin directory. Neural rendering is then toggled on using either a FSR 3 or FSR 4 profile, seamlessly blending the power of Nvidia‘s AI with AMD‘s architecture.

Testing confirmed that this integration works effectively with demanding titles like Cyberpunk 2077 and GTA V: Enhanced. However, the journey isn’t without its challenges. Initial performance benchmarks showed significant strain on AMD hardware, with some tests revealing frame rate drops—for instance, a Radeon RX 9070 XT running Cyberpunk 2077 at 1080p fell from over 80FPS down to a mere 11 to 12FPS.

Fortunately, the modder swiftly responded to the performance concerns, releasing an update that dramatically improved efficiency. This refinement allowed the system to increase performance by approximately 12%, demonstrating that the integration, while complex, is entirely feasible.

The technical breakthrough hinges on a crucial architectural detail: DLSS 5 utilizes FP8 math. This mathematical standard is natively supported by AMD’s RDNA 4 architecture, placing it on par with Nvidia’s RTX 40-series and RTX 50-series GPUs. This capability bypasses the previous restrictions where DLSS 5 was positioned as exclusive to Blackwell hardware.

While the process remains somewhat impractical for casual, everyday gameplay—due to restrictions imposed by anti-cheat software limiting usage strictly to single-player titles—the overall message is undeniably powerful. As one industry observer noted, while the unofficial implementation may not be ready for mass adoption, it definitively demonstrates that the capabilities of DLSS can be realized across the entire AMD ecosystem.