Some of you don't seem to appreciate all the current RT games are made with Turing in mind. Once more and more multiplatform games with RT are made,it will mean all those games will have to be coded with the RT method AMD is using,and it should improve relative to Nvidia,even if Nvidia is still faster overall. For example RDNA2 is developed for inline RT(DXR1.1),so we need to wait and see in the 12 months how actual RT performance pans out.
Also,I remember when Fermi was ahead in tessellation you had all these threads about it,which glossed over all the other weaknesses it has. Yet,when tessellation was really being pushed years later,Fermi didn't do so well either.
Even HUB mentioned that only a handful of games using RT,whereas 100s of games don't. So much of the current market is still made up of GPUs which can't do RT or can't do it that well.For example most of the current GPU market is under RTX2080/GTX1080TI in rasterised performance,and certainly under RTX2080 level in RT performance. Hardware enthusiasts are still stuck in an echo chamber - most of the market does not even have Turing GPUs,let alone anything around RTX2080 level and above.
It will only be when we have RDNA3 and the Turing successor with another bump up in RT performance that we might see more useful usage of RT,as opposed to a few extra shiny settings in the Ultra submenu.