AMD has joined forces with Microsoft to push FSR into a new era where breaking past 100 FPS is meant to feel effortless. Under the name Redstone, the company is launching a full-on assault in the graphics and performance race, aiming to stand toe-to-toe with NVIDIA. By weaving AI-based scaling, lighting, and frame generation into one platform, AMD is making a long-term promise to PC and console gamers alike.

For several years now, the real battle in PC gaming – and increasingly on consoles – has shifted away from raw hardware specs and onto the software battlefield. Over the last five years, image upscaling, frame-interpolation tools, and frame rate boosting technologies have advanced dramatically, even if many of them produce numbers that feel somewhat artificial and push players to lean on external tech instead of pure horsepower. Driving this entire arms race is AI, which has quickly become the engine powering graphical evolution across the industry.

NVIDIA blazed the trail with its AI and machine-learning solutions, steadily pulling ahead of AMD. Now the red team is hitting back with FSR Redstone, a complete reinvention of its visual upscaling and reconstruction stack. This is not just another iteration of FidelityFX Super Resolution (FSR); it is a fundamental change in direction, a commitment to baking machine learning and new reconstruction methods deep into the pipeline in order to compete directly in a space long dominated by its rival.

What Is FSR Redstone, And How Does Microsoft Fit Into The Picture?

FSR Redstone arrives as a modular system that touches several critical parts of the rendering process: AI-based scaling, frame generation, better ray tracing, and lighting enhancements powered by machine learning. AMD presents it as a generational leap and, at least on paper, that description fits. The company needed a genuine game-changer, and Redstone is designed to be exactly that – a platform built to match, and in some cases try to surpass, what DLSS offers in its latest iterations.

In this case, that leap forward has been made with the support of Microsoft. As AMD – Mark Papermaster’s company – confirms, Redstone was developed in collaboration with Microsoft to push upscaling further and, perhaps more surprisingly, to position Xbox a step ahead for the next generation of consoles and their full adoption of these technologies. What started as small experiments and cautious tests is expected to become an everyday reality over the coming years, helped along by a likely new generation of hardware built on RDNA 4.

Everything You Need To Know About AMD FSR Redstone

“For many games, ray tracing is prohibitively expensive for players, and Redstone is here to change all that,” says Matt Booty, President of Content and Studios at Xbox. The idea is to tighten the partnership and go all in on upscaling and so-called “fake FPS” in a landscape where graphics engines and studios regularly demand hardware that either does not yet exist for most players or is owned by only a tiny fraction of them. “We have co-developed FSR Ray Regeneration technology with AMD to achieve a real and tangible improvement in ray tracing,” the executive adds in AMD’s media presentation video, which you can watch on YouTube right now.

As a result, FSR Ray Regeneration stands out as one of the headline features in AMD’s latest wave of technologies, though it is far from the only one. It is the main selling point of Redstone, a technique built to clean up noise and improve the quality of shadows, reflections, and lighting through ray tracing – the notoriously expensive technology that has been a convenient excuse for the current generation of home consoles. A comparison with NVIDIA is unavoidable here: Ray Regeneration – a method that uses Artificial Intelligence to enhance ray tracing without computing actual light rays – has already shown that AI can work wonders with physically based lighting, delivering results that previously demanded raw power alone.

“Realistic lighting at a fraction of the current cost.”

AMD has effectively taken that idea, adapted it, and extended it with its own reconstruction system, which is still in its early days but already looks promising. According to the AMD team, it can deliver superior sharpness in static scenes, even if it still shows some instability in motion that will be ironed out over time. Combined with Radiance Caching – a technology that will not arrive until next year – Redstone improves global illumination using ray tracing without the need to physically calculate every single light bounce. “Realistic lighting at a fraction of the current cost,” summarizes Jack Huynh, AMD’s senior vice president and general manager of Computing and Graphics.

ML Frame Generation

The other major new feature, available starting today in Call of Duty: Black Ops 7, is ML Frame Generation, AMD’s direct answer to the frame-generation tech introduced by NVIDIA. Many players dismiss it as “fake FPS”, but in practice it leaves a very positive impression because it can push frame rates to extremely high levels, depending on your hardware. In essence, it boosts smoothness by using AI to interpolate new frames, lowering the load on the GPU and allowing newer GPU generations to shine. While AMD’s implementation is still in its early stages, it is a hugely anticipated step forward for those who felt that FSR “lagged behind” in fast-paced scenes or games that live and die by high frame rates.

ML Super Resolution

Finally, another pillar of Redstone is ML Super Resolution, the missing piece of the puzzle and the main pain point for many players. In simple terms, it brings machine learning directly into FSR’s image scaling: the algorithm now learns visual patterns, reconstructs fine details more accurately, and cuts down on visual artifacts that used to be obvious in motion. This is not black magic, but the same philosophy that has turned DLSS into a de facto standard – let AI step in wherever raw hardware starts hitting a wall.

The catch? ML Super Resolution replaces the traditional scaling methods used in FSR 1, 2, and 3, which makes FSR Redstone exclusive to the new generation of AMD GPUs that can actually leverage this tech. While AMD has traditionally been a pioneer in not locking away technologies that improve gaming for everyday players – allowing anyone to use its FSR upscaling and even frame rate generation – the details of this leap limit Redstone to the RX 9000 series and above. The reason is straightforward: the new GPUs from the red team have dedicated Artificial Intelligence cores, something previous generations lacked, which is exactly why earlier FSR versions relied on algorithm-based scaling instead of true AI.

Can AMD Compete With NVIDIA? Not Yet, But It’s Clearly Getting Closer

So the obvious question is: is Redstone as good as what NVIDIA currently offers? The most honest answer is: not yet, but it can be, and it clearly intends to get there. Early benchmarks are promising enough to suggest this from the outset. According to AMD – though these numbers should be taken with a grain of salt for two reasons: the comparisons are not against FSR 4.0, and they are measured against native 4K, a notorious FPS killer – we are looking at roughly triple the performance of native 4K in games like Cyberpunk 2077, which hits 123 FPS, or Hell is Us and F1 2025, which both average above 130 FPS.

Even so, we should not be naïve or assume that every comparison will magically favor the red team. NVIDIA has years of experience in this space, particularly thanks to its dedicated AI cores and generation-specific trained models. However, AMD is doubling down on a more open and flexible philosophy, with the idea of allowing parts of this technology to run on other platforms in the future – just as it has done so far, which has been key to FSR’s success. That support has already started in practice: Black Ops 7 is the first game to debut FSR Ray Reconstruction for ray tracing, and another 40 titles are set to receive ML Frame Generation before the end of the year.

FSR Redstone Wants To Be A Real Alternative – With One Clear Promise: Accessibility

This is why Redstone’s greatest strength is its ambition to become a genuine alternative. Yes, it leans heavily on AI. Yes, it embraces machine-learning based scaling and reconstruction. But it does so while holding on to one crucial promise: to remain as accessible as possible, and to ensure that once you buy an RDNA 4-based card you can expect years of better image quality and higher performance thanks to these technologies. In the current narrative of the ongoing “AMD vs. NVIDIA war,” Redstone marks the beginning of a new chapter for AMD. With Redstone, the story starts to change. It may not be the final knockout punch yet, but it is a giant step toward a future in which AMD competes head-to-head on every front.

Source: 3djuegos