Joined
·
17,145 Posts
That is a good question, but time will tell. DLSS 2 is Nvidia's implementation, and they have a long time in development, MS is about to release their own version on consoles with the XSX. Sony's implementation is just a patent at the moment, which may or may not see the light of day, but at least there is something. In the end we will have 3 different implementations with probably 3 different results and performance levels. However, the problem I see is that Sony may be a bit late, as the other 2 may be far in development and optimizations. Maybe I'm wrong, but it's almost impossible to be on the same level from the beginning when you do your own implementation.I find that tech to be excellent, but I'm still curious to see how AMD's version will perform. Nvidia benefit from having polished it over time (the early versions sucked) and from running it on dedicated hardware (helps avoid the performance impact).
Im still sold on it, and I see myself getting a new AMD card in the near future, but I do wonder how itll compare to actual DLSS2.
AMD on Windows will most probably use DirectML and down the road probably an alternative will appear.
I'm just reading through the samples and maybe I will toy myself with the code as is something that interest me.
A sample that uses DirectML to execute a basic super-resolution model to upscale video from 540p to 1080p in real time.