-
Competitor rules
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.
NVIDIA RTX 6000 Rubin Predictions
- Thread starter darkrage
- Start date
I predict I won't be able to afford one.
Honestly? Was hoping for something radical like a multi-chiplet design, but I think it will actually be more-of-the-same but just shrunk to 2nm with the extra trannies used for more AI and Shader cores.
Oh, and driver updates will push fake-frames to around 8-12 per real-frame.
It'll be on 4nm so 20% more transistors in the same space as Blackwell, at the very least it will be a bigger jump than 4000 series to 5000 series, both of which are on 5nm which is incredibly unusual as every new gen typically goes on a smaller node.
I think Nvidia will probably try some kind of extrapolation this time around for their framegen (which currently uses interpolation), they are giving it a go with reflex 2 after all, so research and development is obviously being done in this area. If they pull it off, we could see significantly better input latency compared to how it is now.
I think they will probably want to move on from the cursed 12VHPWR connection and onto a more sturdy one.
I think Nvidia will probably try some kind of extrapolation this time around for their framegen (which currently uses interpolation), they are giving it a go with reflex 2 after all, so research and development is obviously being done in this area. If they pull it off, we could see significantly better input latency compared to how it is now.
I think they will probably want to move on from the cursed 12VHPWR connection and onto a more sturdy one.
Soldato
- Joined
- 2 Aug 2016
- Posts
- 4,242
- Location
- Third Earth
Last edited:
Yes it is
450W TDP thanks to new process node efficiency, 2x powerful than 4090, 32GB-48GB (SUPER) GDDR7 options. dual-slot cooler retained as this makes sense now given the new process node running cooler and better anyway, Jensen laughs on stage saying 12VHPWR was just a beta test and that he thanks everyone for taking part and that a new connector with 3x thicker pins and normal plugs is now being used.
He then ushers the words
He then ushers the words
I can't take these connectors seriously, if someone was trying to follow the nvidia logic when they bought a 4090 they'd have also bought a new PSU that has a 12hwpr port and now if they bought a 5090 they'd also have bought a new psu with a 12+6 port. And then to make sure the GPU aint bottlenecked, they'd have bought a new CPU and RAM, then again a new CPU with the 5090. I can't keep up with all the jank from them, it feels like nothing is drop in anymore, may as well just buy a whole new PC every time Nvidia release a new GPU
You don't need to upgrade the rest of the system when buying a GPU, especially at 4K output as any game that's using a relevant engine is majorly GPU bound not CPU. You just need a half-0decent CPU as a baseline, which most people will have anyway. DDR4 vs 5 is irrelevant within this context.They might bump up the 1% lows a bit in some games, but again the upper end is what counts the most, and ^ lows mean nothing in a game that has crap CPU/GPU utilisation issues, which is essentially everything using UE5 today, with a handful of in-house engines that then get patches after launch to fix.
5090 is bottlenecked even at 4k by anything other than a 9800x3d, which did not exist when the 4090 was released. And to stop the connectors burning the recommendation was a new PSU and cables for each launch as well. Only thing that stays the same is RAM and motherboard if you already had an AM5 board, if its intel then you're due for a new board
Yes it's bottlenecked when factoring in MFG, which is again, not relearnt unless you also have a 480Hz monitor that isn't 1080p (lol) and you absolutely want to use MFG. With any future card that is a decent gain over a 4090 (2x), MFG becomes irrelevant to more people at this point as raw performance is already so high, which means highest quality upscaled 4K is possible now or DLAA for 60fps path traced - All of which is purely GPU bound at these levels.
Anything else hardware wise is not contextually relevant to the exact point I'm making.
Anything else hardware wise is not contextually relevant to the exact point I'm making.
Last edited:
I swear there's already a 6000 series thread somewhere...
But yeah, the 6090 will cost no less than $2.5k and at best be 30-40% better than a 5090. And the only stock at launch will be just the review samples.
I honestly couldn't care less about Nvidia GPUs at this point, it's clear they don't want to offer us anything worthwhile at this point.
But yeah, the 6090 will cost no less than $2.5k and at best be 30-40% better than a 5090. And the only stock at launch will be just the review samples.
I honestly couldn't care less about Nvidia GPUs at this point, it's clear they don't want to offer us anything worthwhile at this point.