Nvidia has a lot work to do to change the minds of the unimpressed. Personally still impressed by the raw potential of Dlss and will wait for the teething issues to improve with maturity and hard work to buy rtx.
What's impressive about DLSS? I mean, when the 3000 comes out, specifically the ti version, will it require 8k to use it, as it will be to fast to use DLSS at 4k? As it will render the native 4k image faster than DLSS per Nvidia's explanation of why lower resolutions are not available on the 2080Ti.
Then you add in that just simply upscaling the image from a lower resolution is a better, cleaner, sharper image. In other words DLSS isn't really doing crap but seriously blurring the image.
Nvidia's best move going forward is to ditch DLSS and focus only on Ray Tracing/TAA. The other option is to put all their R&D into focusing on the tensor cores, to improve DLSS image quality to be superior to even be better than a simple 1440P image upscale, as well as getting tensor core speed improvements so the cards can do lower resolutions as well, which is where AA is needed most. If that is even possible. The only option there might only be to increase the number of tensor cores, which means costs will sky rocket. If they choose to do that, then that means that raw performance of the card will most likely not improve since no R&D will be going into the actual core of the GPU, and that will be a non starter for most people.
Now we know why DLSS can't be performed on a native resolution frame (4k native), and has to do it on a lower resolution frame (1440p) and be upscaled (to 4k). Because it is way to slow to do it at the native resolution and speed would be inferior to TAA giving us lower frame rates. DLSS is appearing to be just trickery. Kind of like where smoke and mirrors are used to get people to believe in magic..