In a recent interview with 4Gamer, AMD announced that its forthcoming Redstone update for the FSR upscaling technology is compatible not only with AMD GPUs but also with those from rival manufacturers, including Nvidia.

Chris Hall, the Senior Director of Software Development and head of AMD’s ROCm project (which is akin to Nvidia’s CUDA platform), mentioned that FSR Redstone, potentially dubbed FSR 5, was developed utilizing technology from a ROCm initiative called AMD ML2CODE (Machine Learning to Code).

This new FSR version converts its neural rendering capabilities into compute Shader code. Hall emphasized that “this means that FSR Redstone’s neural rendering core can also operate on GPUs made by other brands.”

The implications of this are significant; it affirms AMD’s intent to make its technology more accessible across various platforms, unlike Nvidia’s current approach, which maintains a more exclusive ecosystem.

While the existing Radeon RX 9700 XT and other RDNA 4 GPUs do not boast dedicated AI cores like the ones in Nvidia’s Tensor cores, there’s a possibility that future AMD graphics architectures could incorporate specialized AI logic compatible with Microsoft’s DirectX Cooperative Vector technology. Hall pointed out that while this current structure allows seamless integration into DirectX or Vulkan frameworks, AMD is keen on exploring avenues to enhance AI capabilities further in upcoming models.

This once again opens the floor to discussions on AMD’s strategy regarding AI in gaming and graphics enhancements.