NVIDIA’s DLSS 2.0 Is Getting Added to More Games as Mechwarrior 5 and Control to Get Updated with It

[Peter_Brosdahl]

Image Credit: The FPS Review



NVIDIA introduced a new super sampling technique in September 2019. DLSS launched with their Turing based cards. Using AI and deep learning the new process essentially samples images at a lower resolution and then scales them to the desired resolution. This is accomplished thru the Tensor cores of the RTX line. In theory it can offer increased frame rates by decreasing the workload on the GPU at higher resolutions. The initial launch was less than spectacular and actually cast a shadow of doubt for DLSS as a long-term solution. Users quickly noticed significant blurring on one launch title in particular, Metro Exodus. As many other sites, and users, began to compare, and question, DLSS NVIDIA has worked on improvements.



This next iteration of DLSS is currently being referred to DLSS 2.0. It should be noted that this new SDK is not compatible with the first...

Continue reading...

 

[Brian_B]

I think devs are better off investing in (non-proprietary) tech like variable rate shading. I'm sure these companies are getting well compensated to implement these features, but these vendor-specific tricks almost always turn into dead ends, either it just falls off the vine and dies, or best case it gets picked up and changed by a major API but the original proprietary implementation is still left to rot.

 

[Peter_Brosdahl]

I mostly agree. Still got sour memories from what happened with PhysX. I think the idea of this is great but MS should incorporate something into DX. Since both AMD, and NVIDIA, both have a huge background in AI tech they each could easily have tech to continue DLSS without it being proprietery.