With regards to the implementation of upscaling into a rendering pipeline, there's just one version of DLSS -- Frame Generation and Ray Tracing Reconstruction are separate APIs/plugins. Ignore the fact that Nvidia uses labels like DLSS 3.0 or DLSS 3.5, the upscaling algorithm (in so much as what's required by a developer to implement it) is the same (apart from v1.0 but that's not used any more).
Nvidia repeatedly got a pass, or better said got itself a pass through its army of reviewers and PR paid supporters. Some horrible examples - Nvidia failed design of 16pin power connector for 4090 videocards, Nvidia Display port 1.4 bug which was kept hidden and silenced by Nvidia for more than 1 year (their videocards have randomly black screen and sometimes did not even post). If you will look up, you will find more.
They're bloody everywhere, I legit don't know how you missed it, and continue to miss it. It's brought up virtually every time AMD puts a foot slightly wrong. You won't need to look hard.
Sorry mate, gotta disagree, people are massively vocally outraged at literally every point you brought up, including you right now lol, so no, they're not 'getting a pass'
Because it looks like sh*t compared to DLSS?
FSR2 definitely looks better in more games than the horrible images with DLSS2+RT which Nvidia used to prove how good will be their next DLSS3.5 on their most supported game CP2077, or this.
I ignored the rest of your comment as I'm aware you just hate Nvidia for the sake of hating.
DLSS 3.5 is publicly available -- it's Ray Tracing Reconstruction that isn't, though if one has the right credentials with Nvidia, then it's possible to gain access.
Casual gamers just get on with it and enjoy DLSS for what it's doing, but this type of forum or social media attracts people that are, by definition, far more vocal and passionate.
This is where your suppositions are WRONG. Antigamer moves are bad regardless if they come from Nvidia or AMD and I call them out. It seems that you are upset that I focus more on Nvidia wrongdoings. That's because Nvidia did and continue to do more and worse.
I'm all for competitition, I'm not for fragmented standards so nVidia can try and sell new GPUs. If you own a 3090 then nVidia F***'d you with DLSS3. I loved DLSS when it came out, I had no idea that there were going to be 3 versions, possibly more, running concurrently. FSR makes more sense for more people. DLSS is better but not where it counts, which is availability. Now nVidia is using DLSS to try and sell under-performing cards. I'm tired of nVidia's S***. Their Linux drivers are garbage, they stop supporting old cards as soon as a new series is released and then they expect us to pay $500 for 128bit GPU. I've had nearly every nVidia GPU from the 6800GT to the 1080Ti, I'm done with their brand. The 6700XT was my first AMD GPU and I don't think I'll own another nVidia GPU.
I can be more precise, no game has officially implemented DLSS3.5 in the present.
so you're against AMD locking the driver level fsr3 and anti lag+ to RX 7000 only , even though rx6000 could have it too ? fragmenting standards with hardware compability is still more understandable than locking driver level fsr3 to rdna3.
They all require the same inputs from the game engine, all of them, to confirm, XeSS, FSR and DLSS, if you put the work into the game engine to work with one of them, they all can easily work. Hense a random dude in his spare time got DLSS working in Starfield. It's not fragmentation, it's competition, AMD seemingly doesn't like it.
Explain? Is Frame Generation really that big of a deal?
Again, this has been debunked in the comment section already, DLSS has a single version, then Frame Gen / Ray Reconstruction are plug-ins / API's. We're ignoring DLSS v1 because it's no longer used.
Nvidia absolutely dominate in sales in the PC gaming market, any company that releases a PC game with FSR and NOT DLSS is the actual issue here, when the work to add it is absolutely minimal, there isn't any excuse, as per the rumours, it appears AMD are actively trying to get DLSS sidelined, I wonder why that is...
And AMD aren't doing that? right...
Now this, is all true, everyone's been bored of Nvidia for a longtime, AMD aren't exactly lighting a fire under their arse though...
Really? Cos my old 970 seems to still be getting driver updates?
I was a 1080Ti owner until recently, my experience with the 5700XT has left a scar, so I went Nvidia again personally, I'll attempt to sell another AMD card to someone who wants me to build them a PC, purely out of curiosity to see if it was a 5700XT was a bad egg and actually, newer cards are far less prone to crashing.
yeah, r9 390/fury x are both legacy and are only getting critical system updates every few months. If a game won't start, you're on your own.
AMD is not paying people to not implement DLSS. While there are more PC gamers with nVidia cards, there are more AMD gamers. Both the XBOX and the PS4 use AMD graphics so developers already are implementing FSR by default. And it is fragmentation. A developer can implment FSR and it just works with everything. If they want to implement DLSS they have to make sure it works on the 20 series, 30 series and 40 series. They have to debug each series individually. When they already have to implement FSR because of consoles it doesn't make sense to implement DLSS on top of that. This is because DLSS works on a hardware level. The RT cores that DLSS works on are different across all RTX cards. While this is a good thing for the latest and greatest, those 4090's are soon to be made obsolete when the 50 series comes out and nVidia releases yet another new instruction set on their RT cores making the obsolete.
AMD is not charging $500 for what should be a x050 series card "because DLSS". It is funny that AMD still thinks their 16GB 4060Ti is worth $450 compared to the 7700xt.
your 970 is in long term support mode. I doubt you've received a driver update that has increased performance since the RTX series has come out.
This might sound like a conspiracy theory but I'm inclined to believe it. Keep that in mind if you choose to reply. I have no evidence to back up what I'm about to say and I also feel it is more incidental than intentional. With nVidia's dominance in the PC space and the use of ASICs, not off-the-shelf architectures, when AMD's 5000 series was out developers didn't really have much motivation to debug for AMD cards. AMD's drivers have been opensource and as someone who daily drives Linux, I am using third party drivers for my 6700XT. If it simply was AMD's fault for bad drivers then the opensource community would have fixed it. The opensource community often releases bug fixes for AMD GPUs before an official AMD release. Now that consoles use AMD GPUs, developers have a very large incentive to optimize and code for them. I largely attribute that to whole reason AMD's drivers have "gotten better". It has little to do with the drivers and almost everything to do with market share.
No, I'm not for that. That said, I'm not certain that it is a driver lockout and am curious if AMD pulled an nVidia and added instruction sets to RDNA3 that FSR3 leverages. However, we need to be careful with "what if-ism's". I do not see FSR2 in the open very often, it's FSR1.0 even in new releases like Baulders Gate 3.
The fact remains he did it, there is no excuse for the actual devs to not do it unless they were told otherwise. The vast majority of PC gamers out there are using Nvidia. I don't see Sony blocking its developers from using all three options...
In the case of using Super Resolution, the only difference between any of the RTX models is how fast it can process the algorithm and the variance isn't particularly broad -- a factor of 5 between something like an RTX 2060 and an RTX 4090. The same is true with implementing FSR -- as long as the hardware supports the required shader model, the rest is down to performance.
DLSS uses the tensor cores and from what I can dissect from the SDK, it uses standard FP16 matrix operations -- which are the same across all RTX generations.
what did I just read ? rt stands for ray tracing.
I think it's not free in starfield only, and while he did that, another dude made one free for anyone to use.
what did I just read ? rt stands for ray tracing.
dlss works on tensor cores (fp16/8 accelerators), and they are different indeed, cause newer generations have half of old gen tensor core number per sm but they're twice as fast.
2060 Super has the same number of tensor cores as 3080 and more than 4070Ti.
You think that implementing dlss is a separate process for each generation, which is just wrong. Each new generation of tensor cores serves the same purpose and requires no individual tuning, it just saves on die space by making them more efficient.
the only cards that will need optimizing for different versions of same upscaler will be rdna2 when future versions of fsr use rdna 3's ai hardware for training the algorithm, as the spatial upscaler in fsr2.x probably hit a wall in what it can do.
it was kinda short sighted of amd not to put those units on earlier rdna (2/1) as they're clearly planning on using them for something, but the development of a true image reconstruction technique will be hindered by backwards compability.
same will be true about fsr3, which runs on async compute now, while on nvidia it runs on beefed up OFA on rtx40.I don't have a problem with nvidia saying "our cards from rtx40 onwards will have better OFA for doing FG", I do have a big problem with counting FG as part of generational leap, but that's another issue, unrelated to the technical side of how frame generation is handled on nvidia and amd. Using async compute for frame gen already had amd admitting the results will vary during fsr3 presentation, namely the more a game uses async compute already, the slower it will run. It's not the only limitation of fsr3, as not using OFA will provide less information to the gpu too, to the extent where fluid motion frames tech will be disabled when mouse movement is too fast.
frs3 is intended to run on framerates that are high already (+60 at least), while I've seen people saying that nvidia's frame gen does quite well at 30-40fps already, just not 20s as the lag will be awful even with reflex on, and there will be gameplay-breaking visual inconsistency between frames.
To sum up, fsr3 needs 1. async compute headroom 2. high fps to run well, while dlss3 doesn't really care that much, as long as the framerate is above 30.
Sorry for going a bit OT, but I hope fsr3 vs dlss3 will need no further clarification as far as their technical side goes.
I think it's not free in starfield only, and while he did that, another dude made one free for anyone to use.
Okay, well then I abundantly misunderstood what part of the cards are used for DLSS and how.
That's a good reason. I can totally stand behind that too.
And rightly so, as FSR 2.0 is a really solid piece of algorithm engineering by AMD. As a pure shader-based system, it produces better results than Intel's HLSL-version of XeSS and Nvida's NIS (both of which work in a similar manner to FSR).
Sound business strategies do not equit to a positive outcome for consumers. We shouldn't NEED upscaling tech. nVidia's approach is, frankly, offensive. The very idea that upscaling tech is nessecary and now a selling point is unacceptable.
