AMD may consider going in a different direction with FSR in the future, and if a prominent hint on Twitter is anything to go by, Team Red's intention is to develop frame rate boosting technology along AI-driven lines, by same as Nvidia DLSS.
This theory is based on a suggestion tweeted by well-known hardware leaker Greymon55, who spotted a new commit in the LLVM repository that refers to the introduction of WMMA (Wave Matrix Multi-Accumulate) instructions in GFX11.
AI is speeding up! It should work with FSR3.0 and other features. https://t.co/Wh31UxdoaIJune 29, 2022
Read more
Alright, that sounds like a lot of gibberish, so how about a translation? Essentially, GFX11 refers to the next generation RDNA 3 GPUs (and Radeon Pro cards) from AMD, and the WMMA instructions are a way to really boost machine learning (AI) operations. So this could indicate that FSR, possibly in 3.0, is using such AI hacks to get higher quality scaling results.
One theory that has been floating around in the background is that perhaps RDNA 3 graphics cards will offer some kind of integrated hardware functionality modeled after Nvidia's Tensor Cores - dedicated AI processors on RTX graphics cards, used as muscle for boost DLSS.
Analysis: Will AMD take the next turn down the AI avenue?
Obviously, we have to be very careful before drawing big conclusions from a single commitment. And that certainly doesn't mean next-gen RDNA 3 graphics cards will necessarily ship with specific hardware designed to speed up AI workloads (like Nvidia's Tensor Cores).
There are a few key things to consider here. Let's not forget that AMD just released FSR 2.0, and with this technology there have been many changes, including the transition to Time Scale, which offers huge improvements over the updated Space Scale (as used in FSR 1.0). We've seen this make a big difference in supported games, and comparisons point to some pretty impressive picture quality results for FSR 2.0 versus DLSS 2.0, with Team Red catching up considerably.
Furthermore, AMD has argued that FSR 2.0 is an effective rival to DLSS without the need for AI-driven acceleration (or even dedicated hardware on the GPU to improve those operations, which takes time). GPU performance). In fact, recently, AMD has largely downplayed the influence and importance of AI in scaling technologies...
So for AMD to consider equipping its RX 7000 graphics cards with some kind of equivalent for Tensor Cores seems like a good stretch, especially since we haven't heard anything about rumors before, and neither have those next-gen GPUs. far from the launch (maybe they will arrive in October).
Granted, RDNA 3 boards can use these WMMA instructions in some way to further optimize FSR 2.0, perhaps, but any implementation of FSR 3.0 that uses machine learning and associated dedicated hardware with that is certainly a good solution. What this seems to indicate is that AI is, in fact, the way AMD envisions for the future (despite these recent comments and doubts expressed about exactly what difference it actually makes to DLSS).
In this case, the plan may be to continue using FSR 2.0 after the next-gen implementation (FSR 3.0) arrives, as with the debut of FSR 2.0, games still have support for FSR 1.0 (because the latter may not be as good, but it provides broader coverage for less capable GPUs).
Who knows, in the end, but what is pretty clear is that AMD created and pushed FSR to be an open standard and can be used not only on their own Radeon graphics cards (even slightly older models), but also on Nvidia and Intel GPUs. So whatever happens, AMD probably doesn't want to turn its back on this core element of its framerate boosting philosophy going forward.
Today's best graphics card deals
Via Wccftech (Opens in a new tab)