AMD FSR Redstone: AI Features, Performance & Compatibility – Everything You Need to Know

AMD is set to launch its highly anticipated FSR “Redstone” update this afternoon. Ahead of the official presentation, here’s a summary of the information, leaks, and speculation surrounding what is being described as the biggest leap yet for AMD’s FidelityFX Super Resolution. The update promises to significantly enhance image quality and performance, and represents a growing trend of AI-powered upscaling technologies reshaping the gaming landscape.

New AI Features Expected in FSR Redstone

At the core of the excitement are four new, AI-powered graphics features bundled under the codename Redstone. AMD intends these modules to work together as a streamlined pipeline, delivering substantial improvements in both performance and image quality compared to traditional rendering methods.

• Neural Radiance Caching: This feature utilizes AI-powered global illumination, predicting indirect lighting using trained models and storing the results in a cache. This approach aims to dramatically accelerate computationally intensive path-tracing techniques.

• ML Ray Regeneration: This is an AI-based raytracing denoiser designed to reconstruct missing or incomplete details in images. It functions as a neural network replacement for conventional denoisers, potentially restoring lost details in reflections and other complex scenes.

• ML Super Resolution: This employs machine learning for upscaling, using trained models to enhance the resolution of lower-resolution frames. Instead of relying on traditional filtering methods, Redstone leverages AI models to predict a sharper image from fewer pixels, aiming to deliver comparable image quality to native resolution with significantly improved performance.

• ML Frame Generation: This generative frame-creation technology inserts synthetically generated frames into the rendering process to noticeably increase frame rates. Similar to Nvidia’s DLSS Frame Generation, it aims to achieve smoother animations with moderate latency. Redstone marks AMD’s first foray into AI-powered frame generation, building on previous solutions that didn’t utilize machine learning.

Expected Performance Gains and Compatibility

AMD anticipates significant performance improvements over native rendering through the combination of these technologies. The company expects particularly noticeable gains in GPU-intensive scenarios involving path-tracing and raytracing, delivering higher frame rates without sacrificing visual fidelity. Early tests of the Ray Regeneration component in Call of Duty: Black Ops 7 showed virtually no FPS loss compared to the standard denoiser, suggesting Redstone could operate more efficiently than traditional rendering methods – potentially combining upscaling and frame generation to multiply effective framerates compared to native 4K resolution.

However, initial compatibility appears limited. According to official statements, FSR Redstone will initially be exclusive to the Radeon RX 9000-Series (RDNA 4). Older Radeon GPUs based on the RDNA 3 or RDNA 2 architectures will not be supported at launch, a move that appears designed to highlight the latest hardware generation. While reports previously suggested a vendor-agnostic approach – with Redstone potentially implementable via shader units and theoretically compatible with Nvidia and Intel graphics cards – this has not been confirmed. In fact, AMD’s teasers emphasize support only for its RX-9000 GPUs, mirroring the limited launch of FSR 4, which is also officially only available on RDNA 4. (Unofficially, modders have already managed to run FSR 4 on older GPUs, though often with performance compromises.)

Game engine collaboration is also a key factor. For the new features to deliver their full potential, game developers must integrate them directly into their titles. Redstone may be ready on the driver and API level, but implementation in actual games is just beginning. The first demo for Redstone came from Black Ops 7 – though under controlled conditions and with specifically optimized code. Whether similar results can be achieved in complex open-world games or less optimized titles remains to be seen.

Support in Initial Games and Outlook

Call of Duty: Black Ops 7 is leading the way as the first title to incorporate an element of FSR Redstone: AMD’s Ray Regeneration was recently implemented via an update. A forthcoming game update is also expected to add ML Frame Generation to BO7. Observers anticipate limited game support at launch beyond Call of Duty, with some speculating that games co-developed by AMD and Sony – including those related to “Project Amethyst” – could see early adoption of Redstone technologies. Overall, however, it will likely take some time for the new FSR feature set to become widely established in the gaming landscape, with community estimates suggesting it could be over a year before more than 50 games support Redstone.

AMD is expected to reveal all official details about FSR Redstone later today. Our team is preparing an in-depth article that will be published this afternoon, where we will examine Redstone’s real-world performance and assess whether it lives up to the hype. Until then, this summary provides a concise overview of what tech enthusiasts are hoping for from AMD’s major AI update.

While expectations for AMD’s FSR-Redstone update have been fueled by numerous previews and technical speculation, the critical question remains how much of this will translate into practical results. A comprehensive analysis comparing all known leaks, performance predictions, and purported features with real-world outcomes will be necessary to determine which assumptions hold true, where official implementations diverge from prior information, and which aspects of the new ML technologies truly deliver the anticipated advancements.

Sources and Link Table for FSR-Redstone Leaks