This. One of the major advantages to realtime raytracing is achieving realism with less effort than existing methods.
Performance is only truly a concern right now because weāre on first generation hardware. First generation hardware is always heavily restricted compared to subsequent generations. NVIDIAās first graphics cards came out in 1995 but they didnāt gain hardware texturing and lighting until 1999.
In fact because software rendering was both very good and nearly feature identical to a dedicated card they had to make the cards valuable in other ways to ensure sales. NVIDIAās NV1 didnāt just rendering 3D. It handled sound playback and joystick controllers too. And even then it was just barely āacceptableā in reviews because it wasnāt that good at any of it.
It has already happened, Unity demonstrated GI baking, RTX accelerated at recent Unite. Itās the easiest way to use the hardware, to accelerate existing workflows.
Substance is accelerated by it too, in fact I think most industries use it more for that still, that any serious gaming, at least until prices become sane and the next consoles pop up.
NVIDIAās RTX hardware only has one purpose and that is to trace the path the rays take in the scene. Anything beyond that like the way the light appears is handled with normal hardware.
RTX is the trademark for the hardware acceleration implemented by NVIDIA. DXR is the trademark for the software APIs implemented by Microsoft. Neither is dependent on the other. DXR will support any hardware assuming the manufacturer has implemented support for it in their drivers.