FidelityFX Super Resolution, AMD’s answer to Nvidia’s DLSS, has been around for over three years now and has been a real boon for all kinds of GPUs, including Nvidia’s GeForce models. The latest version, FSR 3.1, offers an improved system and architecture but the world of technology is hyped with AI, AMD has decided that it can no longer ignore it and announced that the next update will will be AI-based, at least in the old generation.
In an interview with Tom’s Hardware, AMD’s senior vice president Jack Huynh was talking about some of the frustrations about playing on handheld gaming PCs and enjoying a high-end format that limits how much you can play.
“On the handheld side, my first priority is battery life. If you look at the Asus ROG Ally or the Lenovo Legion Go, it’s just that the battery life isn’t there. I need more hours. I need to play. [Black Myth] Wukong three hours, not 60 minutes.”
It’s not wrong. Even with the ROG Ally X’s large battery, I rarely get more than an hour of gaming on the architecture and while you don’t need to have the internal chip running at full speed for every game, you certainly do for everything new. high-end graphics. The solution, Huynh reckons, will be the next version of FideltyFX Super Resolution.
“This is where the creation and interaction is [come in]so this is FSR 4 we are adding. [W]e fully preferred the group 9-12 months ago to enter AI based. So now we’re on the next generation of AI-based architecture, architecture integration, and concept enhanced efficiency to maximize battery life. And then we can lock the frames per second, maybe it’s 30 frames per second, or 35.”
That’s all Huynh had to say about it all so we’re left with a lot to ponder and speculate. Let’s start with what seems to be the primary purpose of making FSR 4 AI-based: improving battery life. Strange but I prefer that the use of AI also helps solve the weakness of the FSR feature (see video below).
However, none of the current mobile gaming PCs or Radeon-equipped PCs have special hardware in the GPU to accelerate neural network calculations, unlike Intel Arc and Nvidia RTX chips. But the new Ryzen AI 300 series has an NPU (neural processing unit) that can handle these workloads, all with less power.
So does that mean FSR 4 will require access to one or will it be entirely shadow based, like the current version? I don’t do one in need Matrix/tensor cores to handle AI algorithms, but offloading those functions to them frees up the standard ALUs for shadow processing.
Watch it
If FSR 4 do Use NPU or AMD next generation GPUs, using the next RDNA 4 design, it has a modicum of internal matrix cells, does this mean that FSR 4 will not be available to everyone with a different system? The strongest feature of FSR is the fact that it can be run on all types of GPUs: it only needs to support a certain level of Direct3D or Vulkan shader support.
I would be very surprised if AMD plans to leave that process which leads me to suspect that the nature of AI in FSR 4 will not rely on hardware support such as XeSS and DLSS. That said, Intel has two models for its upgrade, one open to all types of GPUs such as FSR and another specific to Arc GPUs, using the matrix cells of the GPU.
So here’s what I think: FSR 4 will still be shadow-based boosting and will employ an AI system based on model generation that will be able to take advantage of all kinds of tools to speed up the process. – processing (GPU, NPU, maybe even CPU) .
It may be similar to a driver based on the Fluid Motion Frames system that works with RDNA 2 or a newer GPU. In fact, it can even be AFMF and then use a neural network to refine the composite model, like DLSS does for upscaling.
AMD has already said that it plans to roll out AI enhancements to all gaming devices at some point, so perhaps every aspect of FSR 4 will be 100% AI-powered. Now, your imagination is the same as mine.
We probably won’t hear much about the next version of FSR until AMD is ready to talk about RDNA 4. So until then, let’s all have fun guessing what it will actually be. Anyone want to bet it’s a combination of ChatGPT and Stable Spread? “Create a picture similar to these two frames but in the same frame in the future.”
#AMD #set #finally #join #Nvidias #FSR #boost #generation #word