Steam Deck FSR 4 Performance 800p target

Switch 2 Handheld mode Quality

Yeah, very impressive - FSR4 is really something that AMD needed for a long time.
What is performance hit for it on Steam Deck vs FSR3 and XeSS?

Somebody made a video about it. 

Drops aren't too huge. As far as I know this is an Int 8 version of FSR 4 that was accidentally leaked by AMD, which RDNA 2 should support fine. Probably why performance is decent. 

Thanks for the link.

It seems that FPS goes 20-25% lower in some cases, but I'd say, for visual quality you get, which is miles better than all FSR3 crap, it's really worth it.

Honestly why hasn't FSR4 been downed ported to PS5 then? As have no intention to get a Pro so fingers crossed this does happen.

What happened to FSR4 being as good as it is because it's "Machine Learning" based? Does this have a higher performance penalty compared to its ML version? Is it still better than PSSR? (what even is the point of PSSR? lmao).

It is an ML model. Just quantized to INT8 instead of FP8 like the officially released FSR4.

Think of these models as a bunch of multiplications of collections of numbers (matrix multiplication.)

You can encode more information by using floating points (decimal numbers of various precision) than integers (whole numbers), but sometimes that doesn't really matter too much for the perceivable performance or conversely you might benefit from having a lower parameter model with more precise primitives (which seems to be why AMD went with FP8 in the end.) 

What makes FSR4 (in its final FP8 form) unrealistic on pre-RDNA 4 hardware is the fact that native FP8 acceleration was implemented with RDNA4. RDNA2 supports INT8 arithmetic natively. 

But in layman's terms, what purpose was in the PS5 Pro including a ML hardware accelerator and exclusively supporting PSSR if all of this can be done to arguably a better effect on older/weaker hardware with less features, no dedicated ML block, tensor cores, etc?  What is it that DLSS2 or PSSR did that made them better? I'm assuming a lower performance penalty but can you confirm this?

What's the point of PSSR? You reckon it will be replaced with FSR4 or will both models evolve separately? Are you able to detect any real world differences between INT8 and FP8 FSR4 yet?

I apologize for spamming questions, but my curiosity is nagging on me.

So there is a lot going on here that influences decisions like this. 

To start, you can run ML workloads on regular general-purposed/graphics-focused cores. Tensor Cores/Matrix Cores are just specialized cores to accelerate parallel low-precision workloads, which deep-learning falls under. The problem is that you'll get a bigger performance penalty than when you have specialized hardware, as 1. you are utilizing resources that could be used more efficiently to render the raster pipeline (not an issue if you have specialized cores for this) and 2. you are likely not natively computing but rather casting into a more precise primitive which is more computationally expensive and memory hungry, if you don't have native ALU support for the less precise data-primitive (i.e no FP8 support until RDNA4 means FP8 needs to be computed using FP16 arithmetic units.) 

From what I gather FSR4 in its final version is a sort of hybrid architecture that incorporates features of ViTs and CNNs. There are many ways you can hybridize ViTS and CNNs at many steps of the model training process, so I don't know what they specifically did (and people are mostly speculating anyway), but for at least the ViT component you do see major benefits from FP8 precision. All we know about this leaked "FSR4" is that inference is done with INT8, which might mean it is a pure CNN model, a quantization of the ViT-CNN hybrid model, one that has transfer learned or distilled from a better ViT one, or it can just be an old model they had before they moved to a VIT-CNN hybrid, with that move mainly being to sell RDNA4 GPUs with an exclusive feature. 

As far as the final result is concerned, it is hard to say. Somebody will have to do an actual experiment using an RDNA4 GPU and released FSR4. It's likely better than the (assumed to be non-lite DLSS 3) implementation in the Switch 2 version of the game, because it is giving a similar result at lower internal and target resolutions when compared to Switch 2 handheld mode. I notice that text is cleaner and the image is sharper on the Steam Deck using FSR4 Performance, but the Switch 2 image is a bit more stable in motion and there is less aliasing overall on Switch 2. 

PSSR probably could work on the base PS5, just with a much higher performance impact that might not make it worth it compared to FSR 2/3 in most cases. Still, I think you probably could get better results than FSR2/3 with some lightweight ML model on base PS5.  

There is also the matter of the fact that as premier models get better (for architectural reasons), they can help smaller models get better (again through transfer learning and distillation) and it could be possible that hardware which is weaker can utilize some very lightweight, but good DL models as time goes on when it couldn't in the past. 

I appreciate the detailed and precise reply 👍