NVIDIA GeForce RTX 3090: Initial reviews show that native 8K gaming is a step too far the US$1,499 card in many triple-A games

[Redaktion]

The RTX 3090 may be an exceptional card, but it is not powerful enough for native 8K gaming across the board according to multiple reviews. The US$1,499 card can handle 8K gaming using DLSS 2.1, but that is not 8K gaming in the truest of senses.

https://www.notebookcheck.net/NVIDIA-GeForce-RTX-3090-Initial-reviews-show-that-native-8K-gaming-is-a-step-too-far-the-US-1-499-card-in-many-triple-A-games.495684.0.html

[8&8]

RTX 3 series is for 4k stop. We are far away from 8k.

To play in 8k for the next decade we must change to think hardware concept. is not increasing number of cores or power consuption or adding better drivers, our tech is showing its limits. Sylicon is dead like Moore's law.

[Michaelo]

As always when chasing the ever higher frame sizes and frame rates, there's little to no mention of how high(objectively) parameters are needed for native support.

As an example - HDMI 2.1, heralded as the future for GPUs for last few weeks -barely caught up to displays like oled 4k TVs - 120fps 4k is the top for HDMI 2.1. 8k is native only for 30fps.

If you'll go higher than 4k/144 or above 40-ish FPS(assuming VRR working this way; otherwise over 30fps) at 8k, there's compression applied, that with such content (high paced, with no time for more complex algorithms to fix artifacts or latency on games will increase) may result in worse experience.

NVidia DSLL is upscaling that shows its flaws (and always will, there's no magic algorithm that would make out exact details from blurry pixels - if there was one, NASA wouldn't spend the fortune for Hubble telescope, would just snap photos with iPhone and then upscale indefinitely). In Linus clip, there was shown(but not mentioned by him) light artifacting/flickering in DLSS example, not seen in native renders.

When it comes to Graphics RAM - single final 8k frame is 33MPix. 8bit RGB colour(so not HDR) makes it a 100MB still. To make such thing, GPU requires much more memory to store textures for a lot of different objects, often duplicated for more streamlined processing. If single face is i.e. a measly 1024x1024 pixels, it's already a 3MB - with ever higher draw distance (and i.e. 50 characters in FOV) it's up to 150MB in GDDR for faces alone.
If current games (that aimed at 4k tops) can eat up 11GB of GDDR(from my experience), optimal solution for 8K is actually 4 times as much -even if ingesting textures directly from SSD will be implemented.

[Dima1114]

I'll be happy with 2k res. maintaining constant 144fps in AAA games with ultra graphics, no anti-aliasing GPU.

[Klaus88]

Starting from the point that I don't mind at all if the CPU business is in favor of Intel or AMD or NVidia (ARM in future?), as for the GPU business, it is more and more evident, day by day, that this website sides with AMD practically on every aspect. This news is an example, there are various technological issues (cable bandwidth, quadriuplication of rendered pixels, necessity of a lot higher ram quantity) that make this kind of article quite nonesense.
And that is an obvious take: last generation was not able to run AAA 4K games without some issues, it would have been impossible for this generation to run AAA 8K games without any issue.
Anyway, I am quite sad about the AMD side trend, this website has lost  its impartiality.

[Spunjji]

Anyway, I am quite sad about the AMD side trend, this website has lost  its impartiality.

What the hell even is this 😂 "You're not biased enough in favour of my favourite company, that must mean you're biased against them". Smells like projection.