AMD FSR 4 Redstone: The Ultimate Performance Review
▼ Summary
– AMD FSR “Redstone” introduces three new machine learning-based features: frame generation, ray tracing reconstruction (Ray Regeneration), and lighting acceleration (Radiance Caching).
– These features are part of the FSR 4 ecosystem and are designed to close the technological gap with NVIDIA’s competing DLSS 4 technology.
– The new Redstone features require RDNA 4 architecture (Radeon RX 9000-series GPUs) for full AI-based operation, though frame generation and upscaling have shader-based fallbacks for older hardware.
– Radiance Caching is available to developers via an SDK, with game integrations planned for 2026, meaning it is not yet available for testing.
– Game developers can implement FSR Redstone features directly or rely on the AMD Radeon Settings control panel to automatically override older FSR versions in games.
The latest evolution in AMD’s graphics technology, FSR 4 Redstone, introduces a suite of machine learning-powered features designed to significantly enhance gaming performance and visual fidelity. This expansion of the FidelityFX Super Resolution ecosystem focuses on three core innovations: frame generation, ray tracing reconstruction, and advanced lighting acceleration. These components aim to elevate the gaming experience on compatible hardware, marking a substantial step forward in AMD’s competitive offerings.
At the heart of Redstone are three distinct machine learning systems. FSR Frame Generation utilizes a sophisticated frame interpolation technique to create additional frames, effectively boosting perceived smoothness and frame rates. FSR Ray Regeneration acts as a neural network-based denoiser, specifically engineered to clean up and refine the image quality in ray-traced scenes, resulting in sharper, more accurate lighting and reflections. Finally, FSR Radiance Caching is a neural radiance cache that intelligently reduces the computational cost of ray-traced global illumination, allowing for more complex and realistic lighting without a crippling performance penalty.
Hardware compatibility is a critical factor for accessing the full potential of these new features. The advanced machine learning inference required for frame generation, ray regeneration, and radiance caching is executed on dedicated AI hardware units found exclusively in GPUs based on the RDNA 4 graphics architecture, namely the Radeon RX 9000-series. For users with older RDNA 2 or RDNA 3 based cards, the experience differs. While shader-based fallback modes exist for the frame generation and upscaling components, these versions lack the performance and quality enhancements delivered by the dedicated AI accelerators in the newest hardware.
From an implementation standpoint, game developers have flexible options. They can integrate FSR Redstone features directly into their games, which would then expose the relevant settings within the in-game menus. Alternatively, support can be managed through the AMD Radeon Settings control panel. This software can automatically override a game’s older FSR implementation with the newest compatible version available for the user’s GPU, eliminating the need for manual file modifications. This system intelligently provides the full FSR 4 Redstone suite for RX 9000-series owners, while seamlessly offering FSR 3.1 for systems running on previous generation RDNA 2 and RDNA 3 architectures.
It’s important to note the availability timeline for these technologies. While the upscaling component of FSR 4 launched earlier, the Redstone additions are now becoming available. Notably, Radiance Caching is being offered to developers via the Redstone SDK, with the first game integrations expected to arrive in 2026. This forward-looking approach provides developers with advanced tools well ahead of their public deployment in finished products.
(Source: Tech Power Up)
