09-14-2015, 03:14 PM
This talk about nVidia hardware lacking support for async compute is truly disturbing. It must be the absolute truth or nVidia would have done something to refute it by now. At this point, the only thing that I can think of that would give any credibility to their stance would be for them to travel back in time, several years, and build something into their GPU that utilized async compute in a manner that would truly stand out to the end users, given the API limitations of the time this would be tough, maybe they could do.... I don't know..... how about physics. They could come up with some dumb marketing name, we'll call it PhysX for the sake of argument- and it would use async compute to do things like generate large amounts of particles, handle things like cloth and fog.... stuff like that. They could do this on older DX11/10/9 games and add physics based calculations that ran concurrently with the graphics engine on the same GPU to prove that their parts were built from the ground up, for many years, for async compute.
Of course, something like that is way outside of the realm of possibility.
Of course, something like that is way outside of the realm of possibility.