We’re almost four months out from the Computex 2025 announcement of FSR Redstone, which AMD maintains will launch in 2H 2025. While this release will primarily interest owners of RDNA 4 GPUs, it turns out that owners of other cards, both Radeon and rival, could stand to benefit.
One of the key advancements FSR Redstone offers is support for neural rendering technologies, such as Neural Radiance Caching (NRC). This technology uses machine learning to more efficiently render light bounces in a scene, boosting ray and path tracing fidelity and performance.
AMD is developing NRC, Ray Regeneration, and ML (Machine Learning) Frame Generation using its in-house ML2CODE (Machine Learning to Code) translation layer. This transforms the neural rendering core into compute shaders.
Chris Hall, AMD senior director of software development, explained the above in a recent interview with 4Gamer. Importantly, they highlight these compute shaders are universal, meaning “FSR Redstone’s neural rendering core can also run on GPUs made by other companies.”
While FSR 4 Super Resolution remains exclusive to RDNA 4 GPUs, such as Radeon RX 9060 XT 16GB, Redstone could serve as a turning point for the technology. After all, modders are already enabling the feature on older hardware.
The question then becomes how useful is FSR Redstone to Intel and Nvidia GPUs? There doesn’t seem to be much for GeForce RTX 40 Series or Arc owners to benefit from here, as each family has their own set of neural rendering technologies.
Going further back in time, GeForce RTX 30 Series cards would benefit from ML Frame Generation but that depends on whether the feature is usable in tandem with FSR 4 upscaling or DLSS. Here’s hoping AMD can provide greater clarity of compatibility as FSR Redstone draws closer.
You can expect further coverage of FSR Redstone, both leaks and official words from AMD, over the coming months. To make sure you don’t miss a single beat, visit the Club386 Google News feed and make sure you’re following along.
