A new GitHub-based prjoect known as DLSS-NR is enabling AMD Radeon 6000, 7000, and 9000 series GPUs to utilize NVIDIA's DLSS 5 upscaling technology. While the mod offers a way to access high-end AI rendering,it currently imposes a massive performance penalty due to the absence of specialized hardware.
The 52 fps drop on the Radeon RX 9070 XT
The performance trade-off for using this mod is currently extreme. In testing conducted within Cyberpunk 2077, enabling DLSS Neural Rendering on an AMD Radeon RX 9070 XT caused frame rates to plummet from 61 fps at 4K RT ultra settings to a nearly unplayable 9 fps. As the report indicates, this massive bottleneck occurs because AMD's architecture lacks the dedicated Tensor Cores that NVIDIA uses to handle these specific AI workloads.
While the 4K experience is currently crippled, there is some light at the end of the tunnel for lower resolutions. Recent builds of the DLSS-NR mod have shown that at 1080p, the same Radeon RX 9070 XT can reach between 90 and 100 fps. This suggests that while the mod is far from a seamless replacement for native features, the optimization process is actively targeting the efficiency gap between different resolutions and hardware capabilities.
Rewriting the neural network runtime for 6000 and 7000 series cards
The DLSS-NR mod functions by essentially rewriiting the DLSS neural-network runtime to be compatible with AMD's GPU architecture. This community-driven effort allows users with older or current-gen Radeon cards—specifically the 6000, 7000, and 9000 series—to tap into an AI-driven upscaling ecosystem that was previously a closed NVIDIA feature.
However, this is not a plug-and-play commercial solution. The project is an experimental demonstration of what is theoretically possible when emulating specialized silicon. Because the mod is essentially forcing a workload designed for Tensor Cores onto general-purpose hardware, it turns a high-speed AI feature into a significant computational bottleneck.
AMD’s RDNA 5 "Neural Lighting" and the FSR rivalry
This experimental mod arrives at a time of intense competition between the two major GPU manufacturers. AMD currently relies on its FidelityFX Super Resolution (FSR) suite, which is designed to be more universal and does not require specialized AI cores to function effectively across a wide range of hardware.
Looking toward the future , AMD is reportedly developing a proprietary solution called "Neural Lighting" for its upcoming RDNA 5 platform. According to the report, this feature would provide a hardware-optimized equivalent to DLSS 5, but it would remain exclusive to AMD silicon . this highlights the ongoing industry tension between NVIDIA's closed, AI-centric framework and AMD's more open, broadly compatible approach to upscaling.
The hurdle of providing legitimate DLSS DLL files
Beyond the performance issues , the DLSS-NR mod presents a significant barrier to entry for the average user. To use the mod, players are required to provide their own legitimate copies of the DLSS DLL files from a game that already supports the technology. This means the mod cannot create DLSS support where none exists; it can only "hijack" it for AMD users in titles that already ship with NVIDIA's files.
This limitation leaves several critical questions unanswered for the gaming community. Will the developer be able to optimize the runtime enough to make 4K gaming viable on high-end Radeon cards? Furthermore , since the mod relies on existing game files, how many modern AAA titles will actually be compatible with this workaround? For now, the project remains a fascinating testbed for cross-vendor AI rendering rather than a practical tool for the everyday gamer.
Comments 0