How so?
Most likely. Nvidia tends to screw over their previous generation customers. I have a 4000 series card and I fully expect it to be left behind once the 5000 series launches.
Why do you consider it a Band Aid and a crutch for developers?
no it will work on all RTX cards as it just improves regular DLSS 2
Because your logic is that more tools allow the developers to be lazy and not have to maximize thier optimization...
might light some fire under amd’s ass
Don't.
"Idiotic" okay I guess we're off to a great start with this discussion.
threadreaderapp.com
It's open source, so developers use it. AMD has such a tiny marketshare that exclusive tech won't have any traction if made exclusive to their GPU's. Intel were clever about it, XeSS is open source but the tech has better quality/performance on their gpu's thanks to the silicon features their architecture has.
Idiotic was appropriate based on the generalization you made. Then followed up with by providing 2 instances where you may be correct out of how many developers? Grow some thicker skin my dude, if you are going to generalize expect your opinion to be criticized.
I need to grow thicker skin but you're the one getting emotional over a graphics discussion and throwing insults around.
Well then their naming scheme is pretty dumb. From what I had seen, DLSS3 was exclusive to 4000 cards and older 2000/3000 cards were still using DLSS2 tech.
Idiotic is not an insult its an observation..... if you took it as an insult than i suppose you do indeed need to grow thicker skin.I need to grow thicker skin but you're the one getting emotional over a graphics discussion and throwing insults around.
This technology will be available to every rtx card, as per the article.
Lol, yeah. NVIDIA is trembling in their boots because of a feature that was announced a year ago and is still nowhere to be found.
You could probably drop down a quality level in dlss 3.5 get similar image quality to the higher dlss 3 level and get more performance and use less vram if it's an issue. The real issue going forward is going to be apples to apples benchmarks when dlss 3.5 could look like it's at higher detail RT settings if they are just matching base resolutions.
Nope. Works on 2000 and 3000 series. But the frame gen part doesn’t. so confusing
"NVIDIA has confirmed that DLSS 3.5 will make its debut in the fall, featuring in titles such as Cyberpunk 2077: Phantom Liberty, Portal with RTX, and Alan Wake 2. Additionally, it will be available in the NVIDIA Omniverse Platform, Chaos Vantage, and D5 Renderer. The “RR” tech will work across all RTX GPUs (unlike Frame Generation)."
That...depends. Cuda is the least-worst of all the GPGPU frameworks, yes. But it's still shit to use, imho. Their effort to port the C++ STL to a unified library that works seamlessly on device and host side is cool, I admit. Some papers published by Nvidia's research team are really impressive, others not so much (to the point where I question the reviewers). Nvidias drivers are shit, unfortunately. It got better over the years but they are still a pain in the ass, especially their linux drivers. Their profiler for evaluating your CUDA-Kernels is also not as easy and straightforward to use as I'd like. And let's face it, they effectively hold the entire research community hostage with their closed-off CUDA ecosystem. You're vendor locked, which is why they can charge what they charge. I still do hope that Intel and AMD push either OpenCL or some alternative framework for real.
I feel like people aren't reading the source. It specifically says ray reconstruction will work across all RTX GPU's unlike frame generation.
FSR3 will be bad, compared to AI reconstruction techniques. You can't do interpolation with motion vectors only. AI is only breakthrough that can really improve picture quality with interpolation and upscaling. There is zero chance that FSR3 will be good without AI.
Game Ready driver for Starfield and ICARUS :
