About a week ago, while the RTX GI SDK was announced... (although as a note access is only "By Request", and is NVIDIA 'Open Source' like GameWorks is right now, not how GPUOpen API/SDK/Tools are), something that seemed to go under the radar was that Vulkan has added their own version of DXR to Vulkan with Ray-Tracing Support as part of the Beta Features.
This as a note launched with an NVIDIA Beta Driver that enabled RTX to support it.
Unfortunately, there is presently no AMD Driver that support it... and I'd say that's a mistake.
While the lack of Drivers support for DXR did (and still does to a degree) make sense, at least until RDNA 2.0 is launched with Hardware Ray-Tracing Support; not providing us with a Vulkan Ray-Tracing (Beta) Driver doesn't.
As noted this is ONLY available in the Experimental Vulkan Path., which no Developers are going to use for Retail.
That means that unlike DXR which effectively launched and was available for Retail Games at the same time as RTX, where supporting it would've resulted in an unfair comparison between GCN (at the time) and RTX Architectures ... with Vulkan, while sure _some_ might think to try to make similar comparisons; I think actually it would be a bigger boon in terms of Ray-Tracing Adoption and Development on AMD Hardware by Developers.
Ultimately such a Driver will be needed at some point., and I think being the first in this case would be quite beneficial.
Don't get me wrong here... Radeon Rays is a "Good" Solution, but it's always been a bit difficult to integrate into Games because it's such a complex pipeline.
Vulkan Ray-Tracing is more-or-less a handful of new GLSL Commands (from a Development perspective) that much more easily allows you to more easily integrate it into existing Pipelines and Rendering Engines.
As opposed to Radeon Rays, where you're essentially having a Dedicated "Ray Tracing" Phase in the Graphics Pipeline...
Getting this into Developer hands earlier, will allow them not only to actually get more used to Ray-Tracing Integration... but also will provide more time they can dedicate to optimising them for usage on GPUs that don't have Hardware Ray-Tracing Support to actually be able to utilise said features as well in Real-Time.
This will further also mean they'll begin working on projects that they'll evolve onto Hardware Ray-Tracing Hardware when it becomes available.
Frankly my argument comes down to not allowing NVIDIA to be the First-to-Market and thus ultimately being able to Dictate Terms of how you compete. And the thing is I have a feeling NVIDIA is going to have a harder time actually getting the Parallel Compute Version to play nice with their Unified Shader Architecture.
