It's not always as simple as "just offer higher VRAM" though. On the 4070 and 4070 Ti the only option would be to double the number of VRAM chips since they're already using the largest 2GB chips, and that would come with a substantial power draw increase. Whatever it costs Nvidia to buy, you can double or even triple that for the cost to the end user because they're not going to lower their margins specifically for the higher VRAM cards.
Personally I would much prefer a power efficient 12GB 4070/4070 Ti than a power hungry 24GB version for an extra £100. I can quite happily skip the games with lazy devs that can't be bothered to optimize their console ports.
They could offer more VRAM,by not selling an RTX3060 replacement at the price of higher end dGPUs tiers.The RTX4070TI should be using an AD103 based dGPU with 16GB of VRAM. The reality is this is just another Turing,and if Nvidia does another refresh like they did with Turing,it would involve pushing tiers down again. So a lot of the FOMO people who defended Turing V1 pricing and tiers looked very foolish when Nvidia re-released it. The ones who waited got a much better deal.
PCMR needs to look at how dGPU sales are declining and PC sales in general. It's a buyers market.
If you are spending £600~£800 on dGPUs,and are running a full ATX system,power draw is less of problem and you are more likely to want to turn up settings or use higher resolutions. People struggling with energy bills won't be spending £800 on dGPUs and gaming for dozens of hours a week 24/7. How many are running overclocked CPUs,inefficient motherboards,oversized PSUs,huge monitors and so on. I could even understand if people are building tiny mini-ITX rigs(like I do).
If people are that concerned with power draw,the RX6600XT could be undervolted to run at 95W and you would turn all the settings down to tax the dGPU even less. Or how many desktops can compare to a gaming laptop in efficiency?
We have had high energy costs since late 2021,and yet many "concerned people" were quite happy not to buy more efficient RDNA2 cards which were cheaper. This was the same argument when people used power draw arguments to buy a GTX970(which really only had 3.5GB of fast VRAM) over an R9 390 8GB. Or the same lot who ran full ATX rigs(not even SFF systems) and would rather get a "high end" £190 GTX960 4GB over an AIB partner R9 290 4GB.
It was the same over a decade ago. When ATI was a bit less efficient with the HD4000 series,it was all about Nvidia being better with power. Then when the HD5000 series hammered Fermi with better efficiency,suddenly it wasn't important and all about overclocking and tessellation. In the end this is Apple level marketing really.
In the end,Nvidia wants people to buy their higher end dGPUs. These have 16GB to 24GB of VRAM. So they have no problem the devs will optimise to that level.
So don't think it's all about lazy devs - if the higher end Nvidia dGPUs can handle it,and the consoles can then Nvidia will be fine with it. If AMD then decides to price it's higher VRAM dGPUs cheaper then that it is a good business move from them.
