I am transferring this remark from the article for further analysis.
... the bottleneck remains with the "CPU processing power".
CPU processing power will be wasted on gaming only by what ever this is run in software, the most known software this is Microsoft DirectX.
Now I will made the wild assumption that when most DirectX commands those executed by the GPU, them the CPU enjoying less stress.
And now I need to ask of whom responsibility is the fact that CPU this gets stressed unimaginably instead of the GPU? some one did a mistake here.
From the presented test results the comparison of Battlefield V vs Project Cars 3 this works as proof to me, that Game developers they are responsible at 100% of any CPU caused bottleneck.
Some people forget that CPU this is two engines in one, the second engine this is the mathematical processor.
For example In a game scene with a forest, if you add 1000 trees in to the landscape, and you decide (as game developer) the math processor to assist the GPU, then you are getting responsible of your own choices.
CPU caused bottleneck will be there, and the problem can be solved by transferring the cost of this mistake at the consumer pockets, some one must pay and buy the more powerful CPU.
Therefore what we have this is a pack of bad written games and a pack of better written Game titles, and now we need to award the Game developers whom respect consumers wallet by them making fewer mistakes.
Regarding RAW GPU power, NVIDIA about the 3000 series it did make more noise than delivery of performance.
I will predict that when this will use lithography down to six microns this will deliver above 120 fps at 4K and 300fps at 1920x1080.
Personally I would care to buy such a dream card so to get old with it.
At my eyes the 3000 series this is another hot pan this using too much of electrical energy.
PCIe bandwidth load this gets increased when some one decides to share GPU load to CPU.