haxxiy said:
You can just emulate all of DLSS's FP8/INT8 instructions on FP16/INT16, which every CPU/GPU supports. It's not any harder than emulating any other aspect of the hardware, it'll just come at a higher cost than native... as it is always the case. 480p to 1080p and 720p to 4K is waaay too expensive to be practical on Switch 2-level hardware. The precious milliseconds you need to do it and still run a game at 30 fps+ are simply not there. You're far more likely to see 720p to 1080/1152p or something. |
It's not as simple as that.
When you halve the precision, you tend to double the amount of computations you can perform (Depending on hardware support. I.E. Rapid Packed Math)... And that means the reverse is true. - It's rare on GPU's where you halve the precision you retain the same level of output.
So in general... FP16 will be twice as slow as FP8.
So it is actually harder to emulate.
For the most part AMD and nVidia have supported low-precision INT/FP for 5+ years now and more recently started to adopt bfloat.
As for DLSS itself, it does consume rendering budget, it's not a free lunch.
dharh said: Just to be clear. I am strictly talking about docked most performance. I don't think the Switch 2 can do PS5 level compute. Frankly, after listening to some people with more knowledge on this kind of stuff that likely docked mode is closer to PS4 to PS4 Pro compute with up-scaling to reach 4k. |
The Switch 2 is more like a mobile Playstation 4, but with extra tricks up it sleeve.
Just like how the Switch 1 was more like a mobile Xbox 360 with extra tricks up it sleeve.
It allows it to punch above it's perceived paper-specifications, but it's weighted in those classes of hardware.
4k upscaling is going to be a waste truth be told, remember it likely only has a tiny 12GB of Ram.
--::{PC Gaming Master Race}::--