{"slug": "dlss-nr-mod-ports-dlss-5-to-3-amd-gpu-generations-2026", "title": "DLSS-NR Mod Ports DLSS 5 to 3 AMD GPU Generations [2026]", "summary": "Independent developer Daniel Blanco released Alpha 0.8.0 of DLSS-NR on AMD on October 17, 2026, a loader-based project that runs Nvidia's DLSS Super Resolution 4.5, Frame Generation, Ray Reconstruction and DLSS 5 Neural Rendering on three generations of AMD Radeon GPUs, according to Tom's Hardware's October 18 report by Jhet Borja. Blanco, who first shipped the project on September 3, 2026 with Neural Rendering support only, states his goal is to match native Nvidia GPU performance on AMD cards. The project is distributed outside both vendors' driver stacks, with no warranty or official support, and Tom's Hardware reported mixed results across games.", "body_md": "A solo developer has done something Nvidia and AMD have spent years insisting was effectively impossible: run Nvidia’s full DLSS 5 feature stack, including Neural Rendering, on AMD Radeon graphics cards. The project, called DLSS-NR on AMD and built by developer Daniel Blanco, shipped its Alpha 0.8.0 update on October 17, 2026, and by October 19 it was the top hardware story on [Tom’s Hardware](https://www.tomshardware.com/pc-components/gpus/amd-cards-now-run-all-dlss-features-thanks-to-one-developers-project-plans-to-match-native-nvidia-gpu-performance), with coverage spreading fast through outlets including Newswav and TechPowerUp.\n\nThe pitch is simple and, for Radeon owners, almost too good to be true: install a loader called NeuralHub, point it at a DirectX 12 game that already supports FSR, DLSS, or XeSS, and it swaps in Nvidia’s DLSS Super Resolution 4.5, Frame Generation, Ray Reconstruction, and DLSS 5 Neural Rendering on hardware that was never supposed to run any of them. Blanco’s stated goal, according to Tom’s Hardware, is to match native Nvidia GPU performance on AMD cards. The results so far are promising in places and genuinely messy in others, which is itself the real story.\n\nDon't miss new tech stories on Google\n\nAdd Shattered once in the Google app and our stories show up in your news suggestions.\n\n[Add now](https://www.google.com/preferences/source?hl=en&q=https%3A%2F%2Fshattered.io%2F)\n\n## What Happened: AMD GPUs Just Gained Nvidia’s DLSS 5 Stack\n\nDLSS-NR on AMD did not appear out of nowhere. Tom’s Hardware reports that Blanco first released the project on September 3, 2026, and at that point it only brought DLSS 5 Neural Rendering, Nvidia’s newest frame-interpretation and denoising technology, to Radeon cards. That alone was notable, since Neural Rendering ships as a feature exclusive to Nvidia’s RTX hardware. The October 17 Alpha 0.8.0 update is a much bigger leap: it adds DLSS Super Resolution 4.5, Frame Generation, and Ray Reconstruction on top of Neural Rendering, which together make up nearly the entire modern DLSS feature set, a jump [Newswav’s republished coverage](https://newswav.com/article/modder-brings-full-nvidia-dlss-5-suite-to-three-generations-of-amd-gaming-g-A2610_Hyafqj) of the Tom’s Hardware report also flagged as the project’s biggest update yet.\n\nFor context on why that matters, DLSS and AMD’s own FSR upscaling technology have been positioned as rival ecosystems for years, each tied tightly to its own silicon. Shattered.io has covered that rivalry extensively, including the gap in blind-test results in [DLSS 4 vs FSR 4: Nvidia Leads, AMD Wins by 40%](https://shattered.io/dlss-4-vs-fsr-4-2026/) and the rollout of AMD’s answer in handheld hardware in [AMD FSR 4 Hits Handhelds as Ryzen Z3 Leaks 12 CUs](https://shattered.io/amd-fsr-4-handhelds-ryzen-z3-leak-2026/). DLSS-NR does not change which company makes which chip. What it attempts to change is which software features that chip can run, by inserting itself into the upscaling pipeline before a game ever talks to the GPU driver’s native path.\n\n## Who Is Daniel Blanco and What Is DLSS-NR\n\nDaniel Blanco is the independent developer behind the project, and according to Tom’s Hardware’s report, published by Jhet Borja on October 18 and updated October 19, DLSS-NR on AMD is explicitly not an official AMD or Nvidia release. It is a solo project distributed outside of either company’s driver stack or software ecosystem. That distinction matters for everything that follows: there is no warranty, no official support channel, and no guarantee that a future driver update from either vendor won’t break it outright.\n\nThe project’s name, DLSS-NR, originally referred narrowly to Neural Rendering support, which is reflected in the abbreviation. As the scope expanded in the October update to cover the full DLSS 5 suite, the name stuck even though the feature set now reaches well beyond Neural Rendering alone. Reporting from Tom’s Hardware frames this as a case of one developer doing, in a solo capacity, what neither GPU vendor has attempted: cross-pollinating a flagship feature set across a competitor’s hardware line.\n\n## Inside the NeuralHub Loader: How the Mod Actually Works\n\nAccording to the available reporting, DLSS-NR does not work by swapping AMD shader binaries or replacing driver-level code wholesale. Instead, the mechanism is described as an upscaler-pipeline interception approach, built around a loader called NeuralHub. The practical flow, as described in coverage of the release, works like this: a player enables FSR, DLSS, or XeSS inside a compatible game’s graphics settings, since the game needs to already expose one of those three upscaling paths. NeuralHub then inserts itself into that pipeline and applies the user’s chosen DLSS configuration instead of the upscaler the game thinks it is running.\n\nReports indicate the project uses FSR 4.1.1 as a consistent fallback and baseline upscaling implementation within its own configuration, even as it layers Nvidia’s DLSS models on top. The mod is reported to work specifically with DirectX 12 titles running on Windows 11, which narrows its practical footprint but still covers the large majority of recent AAA releases. This is not simply a reskinned version of OptiScaler, the open-source DLL-swapping tool that has circulated in modding communities for similar purposes; available sources specifically identify NeuralHub and the DLSS-NR project as the operative loader and injection mechanism, distinct from that established tool.\n\n### Why This Differs From a Simple DLL Swap\n\nOlder cross-vendor upscaler hacks have typically relied on tricking a game into loading a different vendor’s DLL file in place of the one it expects, a technique that works for some upscalers but breaks easily and offers little control over which specific sub-features activate. DLSS-NR’s pipeline-interception approach, by contrast, is built to expose configuration options and an in-game overlay through NeuralHub, giving users more granular control over which DLSS 5 components run and how. That added complexity is also where much of the current instability comes from, since every additional feature toggle is another point where a given game, driver version, or rendering path can misbehave.\n\n## Which AMD Radeon GPUs Are Supported\n\nTom’s Hardware describes DLSS-NR as targeting three full generations of Radeon hardware, spanning RDNA 2 through RDNA 4. Support is not uniform across all three, however. The newest cards get the deepest feature coverage, while the oldest generation is explicitly limited.\n\n| AMD Radeon Family | Architecture | DLSS-NR 0.8.0 Support Level | Notes | \n|---|---|---|---|\n| Radeon RX 9000 series | RDNA 4 | Full feature set | Primary target hardware; best-performing configuration reported | \n| Radeon RX 7000 series | RDNA 3 | Full feature set | Reported to match RX 9000 feature coverage | \n| Radeon RX 6000 series | RDNA 2 | Experimental | Largely limited to DLSS 5 Neural Rendering only | \n\nThat tiered rollout tracks a familiar pattern in GPU feature support more broadly: newer silicon tends to have more compute headroom and more modern instruction support to absorb an extra software layer without collapsing frame rates. It also mirrors how AMD itself staggered FSR 4’s rollout, which initially favored its newest RDNA 4 cards before expanding backward, a pattern Shattered.io detailed in [RTX 5070 vs RX 9070 XT: AMD Wins by Up to 18%](https://shattered.io/rtx-5070-vs-rx-9070-xt-2026/), where the RX 9070 XT already stood out as AMD’s strongest current-generation card relative to Nvidia’s competing tier.\n\n### Why OptiScaler Comparisons Keep Coming Up\n\nBecause OptiScaler has been the best-known tool for swapping upscalers between vendors in recent years, early discussion of DLSS-NR understandably drew comparisons to it. The distinction reporting draws is that OptiScaler primarily translates calls between existing upscaler APIs, letting a game that expects one upscaler feed its data to a different one, without necessarily carrying Nvidia’s actual DLSS 5 models and Neural Rendering pipeline across to AMD hardware. DLSS-NR’s NeuralHub loader is described as a separate, purpose-built project rather than a variant or fork of OptiScaler, built specifically to carry Nvidia’s newest model-based features onto Radeon silicon rather than just redirecting an API call.\n\n## Performance Claims: 90 FPS and a Run at the RTX 5070 Ti\n\nThe headline performance figure circulating in coverage of the release comes directly from Blanco: a reported 90 to 100 frames per second at 1080p on an RX 9070 XT while running the DLSS-NR implementation. Blanco’s stated target, per Tom’s Hardware and TechPowerUp’s reporting, is performance comparable to an RTX 5070 Ti, Nvidia’s own upper-midrange card, which would be a striking result if it held consistently across games.\n\nThere is also a reference point from the project’s earlier, narrower build: a prior Neural-Rendering-only version reportedly achieved a 74% performance increase in a single day of development work, according to coverage of the project’s history. That figure describes an internal optimization jump during development rather than a result players can expect to reproduce today, but it illustrates how actively the project has been iterating since its September debut.\n\nIt is worth being precise about what these numbers are and are not. They are developer-reported figures and user-submitted results, not outputs from a standardized, independently run benchmark suite across a controlled set of titles. Reporting is explicit that the claim of matching native Nvidia performance should be read as the project’s goal, not as an established, repeatable result spanning the current game library.\n\n## The Catch: Inconsistent Results and Up to 25% FPS Losses\n\nUser reports collected in the coverage paint a far less tidy picture than the headline frame-rate figures suggest. Some users reported no measurable performance loss compared with running native FSR on the same hardware. Others reported losses ranging from a handful of frames per second up to as much as 25%, with results varying by game, specific configuration, and driver version. One Reddit-sourced test, cited in secondary reporting, described a Cyberpunk 2077 path-tracing run at 1440p on an RX 9070 XT showing no performance loss at all, though that result is explicitly user-reported rather than independently verified in a lab setting.\n\nRay Reconstruction in particular shows the widest spread of outcomes, with some configurations showing no measurable hit and others showing the full 25% drop. That inconsistency is the clearest signal that DLSS-NR remains genuinely alpha-stage software: Blanco has acknowledged, per the available reporting, that image quality is close to native Nvidia DLSS but that some effects may not work correctly yet. For a feature set this complex, built entirely outside of either vendor’s validation pipeline, uneven results across hundreds of possible game and driver combinations are close to unavoidable.\n\n## DLSS-NR vs Native Nvidia DLSS 5 vs AMD’s Own FSR 4\n\nThe most useful way to size up what DLSS-NR actually delivers is to line it up against both ends of the rivalry it straddles: Nvidia’s own native DLSS 5 implementation, and AMD’s official, vendor-supported FSR 4.\n\n| Feature | Native Nvidia DLSS 5 (RTX hardware) | DLSS-NR on AMD (Alpha 0.8.0) | AMD FSR 4 (official) | \n|---|---|---|---|\n| DLSS Super Resolution 4.5 | Yes, native | Yes, via mod injection | No; uses FSR’s own upscaling algorithm instead | \n| Frame Generation | Yes, native | Yes, via mod injection | FSR 4 has its own separate frame generation tech | \n| Ray Reconstruction | Yes, native | Mixed: 0% to -25% FPS depending on config | No direct equivalent | \n| DLSS 5 Neural Rendering | Yes, native | Yes, added first in the September 2026 build | No equivalent feature | \n| Vendor support and warranty | Yes, Nvidia-backed | None; community alpha software | Yes, AMD-backed | \n| Anti-cheat compatibility | Yes | No; reported to be blocked or unstable | Yes | \n\nThe table makes the trade-off plain. DLSS-NR offers feature parity on paper that AMD’s own FSR 4 simply does not attempt to replicate, since FSR 4 is built around its own upscaling and frame generation techniques rather than cloning Nvidia’s. But the mod trades away the two things vendor-backed software reliably provides: consistency across titles and compatibility with the anti-cheat systems that gate most competitive multiplayer games. Shattered.io’s earlier look at the two ecosystems in [DLSS vs FSR: Nvidia Wins 48% Blind Test](https://shattered.io/dlss-vs-fsr/) found Nvidia’s native implementation still ahead in controlled testing conditions, which is the bar DLSS-NR is now trying to clear on hardware that was never designed to run it.\n\n## Why Anti-Cheat Software Is the Mod’s Biggest Problem\n\nOf all the caveats attached to DLSS-NR, the anti-cheat issue is the one most likely to limit real-world adoption. Coverage of the release is consistent on this point: anti-cheat software either blocks the mod outright or is flatly incompatible with how it operates, and reporting recommends limiting any testing to single-player games as a result. That is a meaningful restriction given how much of the PC gaming audience spends its time in competitive, anti-cheat-protected titles.\n\nThe underlying reason is structural rather than incidental. Modern anti-cheat systems are designed to flag exactly the kind of behavior NeuralHub performs: injecting into a running game process and intercepting its rendering pipeline before the game’s expected code path executes. That is indistinguishable, from an anti-cheat engine’s point of view, from a wide category of actual cheating tools, regardless of the mod’s benign intent. Until or unless a tool like this is explicitly allow-listed by individual anti-cheat vendors, which none have done according to available reporting, the safest assumption for any player is that using DLSS-NR in a competitive online game risks a ban.\n\n## AMD and Nvidia’s Silence Speaks Volumes\n\nNeither GPU maker has issued an endorsement, validation, or formal response to DLSS-NR, according to the available coverage. That silence is probably strategic on both sides, for different reasons. For Nvidia, a mod that ports its flagship software feature onto a competitor’s hardware undercuts one of the clearest reasons buyers have historically chosen Nvidia over AMD at a given price point: feature access, not just raw frame rates. Acknowledging the project at all, even to criticize it, risks drawing more attention to the fact that its biggest differentiator can be partially replicated on rival silicon by one independent developer.\n\nFor AMD, the calculus is more complicated. A tool that makes Radeon cards more attractive by giving them access to Nvidia’s feature set could, in theory, help AMD sell GPUs. But it also implicitly concedes that AMD’s own FSR 4 isn’t yet compelling enough on its own to keep every buyer in-house, and it creates a support headache if players start blaming AMD drivers for stability issues that actually originate in an unsupported third-party mod. Staying quiet lets both companies avoid taking a position until the project either stabilizes into something with real staying power or fades the way many ambitious cross-vendor mods have before it.\n\n## Historical Context: Cross-Vendor GPU Modding Isn’t New\n\nDLSS-NR is the highest-profile attempt yet to cross the Nvidia-AMD software divide, but it is not the first. Shattered.io covered an earlier, narrower case of boundary-pushing in [DLSS 5 Modded to Run on RTX 4090, No Nvidia Fix](https://shattered.io/dlss-5-rtx-4090-ada-lovelace-mod-2026/), which involved forcing Nvidia’s newest DLSS version onto an older Nvidia architecture that was not officially supported, rather than crossing between vendors entirely. That earlier effort showed there was already demand, within Nvidia’s own hardware lineup, for unlocking software features tied artificially to specific silicon generations.\n\nDLSS-NR goes a full step further by crossing the vendor line itself rather than just a generational one. The broader pattern it fits into is decades old in PC gaming: enthusiasts reverse-engineering locked software features onto hardware a vendor never intended to support, whether that was unofficial driver hacks enabling SLI on unsupported motherboards in past GPU generations, or community ports of proprietary upscaling technology. What makes this case notable is the scope: a full suite spanning Super Resolution, Frame Generation, Ray Reconstruction, and Neural Rendering, working across three separate Radeon architectures rather than a single narrow hack.\n\n## Market Impact: What This Means for GPU Buying Decisions\n\nFor the typical GPU buyer weighing a new purchase today, DLSS-NR should not yet factor into a buying decision, and the available reporting does not suggest otherwise. Alpha software with inconsistent performance, no anti-cheat compatibility, and no vendor backing is not a reliable substitute for choosing hardware based on its official, supported feature set. But the project’s existence still has a real market effect, independent of whether any individual buyer ever installs it.\n\nIt chips away, even if only modestly for now, at one of the stickiest arguments for paying a premium for Nvidia silicon: exclusive access to the newest upscaling and frame generation technology. If a community project can bring a meaningful fraction of that feature set to Radeon cards, even in alpha form, it changes the conversation among price-sensitive buyers deciding between, for example, an RX 9070 XT and a similarly priced RTX card. It also puts a small amount of indirect pressure on AMD to keep investing in FSR’s own trajectory, since a thriving unofficial DLSS port is itself a signal that FSR alone hasn’t fully closed the perceived feature gap for every buyer.\n\n## Competitive Landscape: FSR 4, FSR 5, and the Neural Rendering Race\n\nDLSS-NR’s emergence lands at an active moment in the upscaling arms race. AMD has been steadily pushing FSR 4 into new hardware tiers, including handheld gaming devices, while Nvidia continues rolling out DLSS 5 across its own lineup and new hardware platforms, including the rack-scale and PC deployments detailed in [NVIDIA RTX Spark Launch Week: DLSS 5 Doubles Speed](https://shattered.io/nvidia-rtx-spark-launch-week-dlss-5-2026/). Both companies are racing to make neural, AI-driven rendering techniques the default way games render frames, not an optional add-on.\n\nDLSS-NR sits awkwardly, and interestingly, outside that official race entirely. It does not represent AMD catching up to Nvidia on its own terms, nor does it represent Nvidia extending its lead through official channels. It is a third path: an unofficial bridge that borrows Nvidia’s technology wholesale and forces it onto competing hardware, rather than racing to replicate the underlying technique independently the way FSR’s own neural upscaling work has tried to do. That makes it a wildcard in the competitive picture rather than a data point either company can claim credit for.\n\n## Predictions: Where DLSS-NR and Cross-Vendor Modding Go Next\n\nA handful of outcomes look plausible from here, based on how similar mod efforts have played out in the past and the specific pressures already visible in this case.\n\n- Expect continued rapid iteration from Blanco in the near term, given the pace shown between the September Neural Rendering-only debut and the far broader October 17 Alpha 0.8.0 release just six weeks later.\n- Anti-cheat vendors are likely to start explicitly flagging or blocking NeuralHub’s injection signature as the mod gains visibility, mirroring how other injector-based tools have been treated once they cross a certain adoption threshold.\n- AMD faces indirect pressure to accelerate its own neural rendering roadmap beyond FSR 4, since a credible unofficial DLSS port on Radeon hardware undercuts the argument that FSR alone is sufficient for feature parity.\n- Expect no official statement from Nvidia unless the project’s adoption grows large enough to show up in retail GPU purchasing patterns; quiet non-engagement remains the lower-risk posture for now.\n- If stability improves significantly in future alpha or beta builds, similar cross-vendor porting attempts targeting Intel Arc GPUs are a plausible next step, following the same interception-based blueprint NeuralHub has established.\n\n## Should You Try DLSS-NR Right Now\n\nFor most Radeon owners, the honest answer right now is: only if you understand you are installing unsupported alpha software with no vendor backing. Single-player gaming on a well-supported title is the lowest-risk use case, consistent with the guidance implied in the available coverage. Multiplayer, anti-cheat-protected games carry a real risk of enforcement action and should be avoided entirely until, and unless, specific anti-cheat vendors explicitly clear the tool. Performance and image quality results remain inconsistent enough, ranging from parity with native FSR to a 25% frame rate loss in some configurations, that treating any single benchmark number as representative of your own setup would be a mistake.\n\nOwners of RX 9000 and RX 7000 series cards get the fullest feature coverage today, while RX 6000 owners should expect a much narrower, Neural-Rendering-only experience given the experimental support tier. Anyone considering it should also budget time for troubleshooting, since alpha-stage, community-built software interacting with proprietary driver stacks from two different vendors is inherently more fragile than a standard driver update.\n\n## Frequently Asked Questions\n\n**What is DLSS-NR on AMD?**\n\nIt is an independent, unofficial project built by developer Daniel Blanco that brings Nvidia’s DLSS 5 feature set, including Neural Rendering, Super Resolution 4.5, Frame Generation, and Ray Reconstruction, to AMD Radeon graphics cards through a loader called NeuralHub.\n\n**Is DLSS-NR an official AMD or Nvidia product?**\n\nNo. According to Tom’s Hardware’s reporting, it is explicitly not an official release from either company. It is a community-built, alpha-stage project distributed independently.\n\n**Which AMD GPUs work with DLSS-NR?**\n\nThe project supports three Radeon generations: RX 9000 series (RDNA 4) and RX 7000 series (RDNA 3) get the full feature set, while RX 6000 series (RDNA 2) support is experimental and largely limited to DLSS 5 Neural Rendering.\n\n**How much performance does DLSS-NR cost compared to native FSR?**\n\nResults vary widely. Some users report no measurable performance loss, while others report losses as high as 25%, depending on the specific game, configuration, and driver version in use.\n\n**Can I use DLSS-NR in online multiplayer games?**\n\nThis is not recommended. Reporting indicates anti-cheat software either blocks the mod or is incompatible with it, and coverage specifically suggests limiting use to single-player titles to avoid enforcement action.\n\n**Have AMD or Nvidia responded to DLSS-NR?**\n\nNo official response, endorsement, or validation has been issued by either company as of this reporting.\n\n**Does DLSS-NR replace AMD’s FSR 4?**\n\nNo. The mod reportedly uses FSR 4.1.1 as a fallback upscaling path within its own configuration, while layering Nvidia’s DLSS 5 models on top. FSR 4 remains AMD’s official, vendor-supported upscaling technology.\n\n**What games does DLSS-NR currently work with?**\n\nThe mod is reported to work with DirectX 12 titles on Windows 11 that already support FSR, DLSS, or XeSS upscaling, since NeuralHub inserts itself into that existing pipeline rather than adding support to games that lack any upscaler integration.", "url": "https://wpnews.pro/news/dlss-nr-mod-ports-dlss-5-to-3-amd-gpu-generations-2026", "canonical_source": "https://shattered.io/dlss-nr-mod-amd-gpus-dlss-5-2026/", "published_at": "2026-10-19 14:33:30+00:00", "updated_at": "2026-10-19 14:44:41.603641+00:00", "lang": "en", "topics": ["ai-products", "generative-ai", "neural-networks"], "entities": ["Daniel Blanco", "DLSS-NR on AMD", "Nvidia", "AMD", "Radeon", "DLSS 5", "NeuralHub", "Tom's Hardware"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/dlss-nr-mod-ports-dlss-5-to-3-amd-gpu-generations-2026", "markdown": "https://wpnews.pro/news/dlss-nr-mod-ports-dlss-5-to-3-amd-gpu-generations-2026.md", "text": "https://wpnews.pro/news/dlss-nr-mod-ports-dlss-5-to-3-amd-gpu-generations-2026.txt", "jsonld": "https://wpnews.pro/news/dlss-nr-mod-ports-dlss-5-to-3-amd-gpu-generations-2026.jsonld"}}