Man of Honour
Well one thing is for sure, I won't be getting a new GPU any time in the distant future, the 4090 has to last me as long as the OG GTX 980 did back in the day, 5 years
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
With AI now being the top priority in GPU development I expect future GPUs will be less focused on graphical horsepower and will instead have to rely more on AI features for performance gains.Got quite a long wait, Q1 2025 going by Nvidia's own info and other info coming out, Best case scenario is October next year.
Imagine the 6000 series with DLSS5, The game will play itself so you can then go outside and experience real world ray tracing which is DLSS 9000+ !
With AI now being the top priority in GPU development I expect future GPUs will be less focused on graphical horsepower and will instead have to rely more on AI features for performance gains.
Well one thing is for sure, I won't be getting a new GPU any time in the distant future, the 4090 has to last me as long as the OG GTX 980 did back in the day, 5 years
The more you buy the more you save?We're now at a stage where it's longer between each gen and when it does come you're expected to drop £1500 for the card that offers the most value for money.
What if it is locked to the 5000 series and you are stuck on a 4090
You're probably correct and what with Nvidia making something ludicrous like $80,000 per highest end AI accelerator card they are in no rush to get a new gen of game cards out.
AI can be amazing but going forward I can see it doing a lot of harm via greed in the PC industry.
By the time the 5090 comes out, i'll be ready to not learn my lesson and YOLO again.
And you'll still have the same sig too for the ***** and giggles
The new tech is exciting but I'm still concerned about how much they'll use it to manipulate every GPU series going forward.
We're now at a stage where it's longer between each gen and when it does come you're expected to drop £1500 for the card that offers the most value for money.
If you step down a model, you're still paying over a grand and the performance drop off is gigantic. Hopefully the 5 series doesn't follow the same pattern but...
By the time the 5090 comes out, i'll be ready to not learn my lesson and YOLO again.
Only a year and a half to go and then Jensen is going to play with your bungbung using RT cores !
Aslong as he's wearing his leather jacket and spanking me with a 5090, i'm all in baby.
I remain skeptical - The next advancement will be some form of new rendering (see the Gaussian Splatter thread) - But that runs effectively on modern hardware and all that's needed is lots of VRAM, so that's covered anyway once devs get working on it properly since it's more of a recent innovation on the technology.Nvidia will release a feature either in the 5000 or 6000 series that will force your hand
Jensen: And what's the magic safety phrase ?
You: Stream processors are silly and CUDA is cooler.
Jensen: Good boy, Now prepare your thermal exhaust port for some Jensen tracing !
I remain skeptical - The next advancement will be some form of new rendering (see the Gaussian Splatter thread) - But that runs effectively on modern hardware and all that's needed is lots of VRAM, so that's covered anyway once devs get working on it properly since it's more of a recent innovation on the technology.
We still have neural radiance caching to be enabled in the drivers so that ray tracing can be even faster on existing cards, so plenty of things to come still that will improve RT/PT on the current cards
We still have neural radiance caching to be enabled in the drivers so that ray tracing can be even faster on existing cards, so plenty of things to come still that will improve RT/PT on the current cards
While this can efficiently boost fps counts and visual fidelity, as well as free up VRAM, there is a slight downside, nonetheless. Each frame would see a 2.6 ms render time increase for FHD resolution, possibly more than 3 ms for 4K. Presumably this can still be combined with frame generation.
I'm not so sure it's the same thing, ray reconstruction is new and forms the extension to become DLSS3.5 - Also the article notes that there is a caveat, increased render latency:
Ray Reconstruction has no render latency increase, in fact rendering time is decreased hence the near real-time light latency response from bounced surfaces, on top of that the noise-free RT image quality. And then all RTX cards support ray reconstruction, neural radiance will be a 40 series and above feature only as the increase in latency relies on FG+Reflex to bring it back down again