I won't pretend to be an expert in that demo, but I remember talk pre-RTX launch that the Vega and upcoming 7nm Vega could be capable of ray-tracing through sheer compute power in a way the old GTX cards could not. When we see the same thing running on a GTX 1070 then we can officially debunk that tech demo
Obviously Vega is a different beast to our current Navi and the RDNA 2.0 Navi coming to consoles with ray tracing, though somehow I doubt these upcoming cards copied NVidia RTX cores and suspect they will be approaching the same challenge in a new very different way.
Firstly that demo is running an a GCN card, we don’t yet know how GCN compares to RDNA for ray tracing - we don’t know if on a per core basis it’s worse, same or better yet. It’s pre mature to suggest otherwise. What we do know is that AMD is keeping GCN for compute and RDNA for gaming which suggests GCN would still be the compute king.
You won’t see GTX cards or any cards until Crytek Releases that demo publicly. We have seen GTX cards run actual games though with similiar performance to that demo albeit games are more demanding than that demo.
The point I was making is that we should look at Ray Tracing like Tesselation, not Physx. That’s the mistake many people are making.
They think Nvidias way of games utilising ray tracing is like Phsyx and therefore it will die out.
They don’t realise that it’s actually more like tesselation in that both amd an nvidia run the same tesselation in games and the actual performance just comes down to how it’s handled by the driver and gpu architecture, it’s not locked down. If ray tracing was locked down it wouldn’t even run on GTX cards just like physx did not run if you didn’t have a physx add on card.
And then it’s again to pre mature to say one side is better than the other - we just don’t know who will end up faster yet, but I believe both will be running the same ray tracing in the same games and developers won’t need to do anything special for it, it’s just down to how calculations are done by the card.
I’d like to believe that console code translates into dominant gains for AMD on PC but in the past it just hasn’t happened. When the PS4 came out with its 8 core x86 cpu, many people expected that Bulldozer cpus would get a big performance boost thanks to all its cores - it never happened. People thought AMD GPUs would get a big boost thanks to developers using a 7000 series amd gpu - it never happened