That only works if you can get everyone to benchmark with those settings. As soon as anyone using an existing method where the magic tricks don't help, the numbers don't look as good.
Maybe we'll eventually get fortunate and have history repeat where AMD comes out with a nice lean chip that punches above its weight class like the 4850 and 4870 back in the day.
I don't think that happens until after Navi unless Navi is itself a new architecture. AMD's GCN is a lot like Intel's Core in that it's long in the tooth and in need of replacement. Even if NVidia blunders with RTX, AMD isn't in a position to capitalize like they did with Ryzen.
This has been debunked so many times. Just because AMD calls it GCN, doesn't mean its the same architecture. Just because Nvidia has been calling it different names doesn't mean its a different architecture.
Even today with Turing, Nvidia's architecture holds roots from Fermi, their GTX 400 series. Okay? And even that had so many similarities with their 200 series.
On the other hand the difference between GCN architectures from 1 to 4 is huge. You can't even compare GCN 2 to 3 or 3 to 4, the differences are huge, so its not the same architecture, though obviously like Nvidia its got roots since GCN 1.
But even GCN 1 has roots since their HD 2000 era. In fact their HD 2000 era graphics were the biggest change in architecture, as they switched to unified architecture.
Vega on the other hand is a new design for them, its a massively redesigned architecture from their GCN ones, though obviously as with any architecture it has to draw roots from somewhere. I mean cars from 100 years ago had an engine, cars today still have an engine, are they the same? Cars 100 years ago had wheels, cars today still have wheels, are they the same?
So no, you are completely wrong in your premise. AMD's architectures are not old, its not the same architecture, in fact they are all massively different from one another, if anything its AMD that had innovated a lot more and added a lot more new features and stuff, while Nvidia played it safe. I'd say that was AMD's error, they innovated too fast, they though low level API's would come faster and become more popular and the standard faster, but that didn't happen.
Even GPU's like Fury were designed to take full advantage of low level api's, if back then low level api's took off, like say Vulkan which is based off of AMD's Mantle, then the Fury X would have swiped the floor with the GTX 980ti.
Nvidia on the other hand were reiterating more, rather than innovating. They more closely designed new architectures to the previous one, deciding to reiterate more and focused on DX11 and high level API's, which bear fruit for them. Even Pascal was much more DX11 focused, rather than DX12/Vulkan. In fact even Volta didn't change that, even Volta is essentially 100% Pascal for DX12/Vulkan. Only Turing is a much more refined DX12/Vulkan architecture and we see that in games like Doom and Wolfenstein 2. In fact Turing cards gain over 50% more performance when compared to Pascal cards in Wolfenstein 2 according to all review sites. The biggest gain Turing has over Pascal is in Wolfenstein 2.
We see it in games like Sniper Elite 4 as well, Nvidia's Vulkan/DX12 performance is only matching AMD's performance only now with Turing.
Navi will be different though as AMD learned their lesson, Vega was supposed to be this jack of all trades monster, good at workstations, good at AI, good at gaming, good at encoding and it ended up being an amazing professional card, beating the Titan X pascal, but it sucked in games.
Now they are taking a different approach, designing 7nm Vega purely for professional workloads and designing a new architecture purely for gaming workloads.