Reflex 2 is said to be used only in very fast online games where latency matters more than anything, as it creates too many issues/artefacts to use in normal games where quality of image matters more. At least that's what I've seen/read in few different places, but time will tell.
I'd suspect you have both versions implemented and it will be up to the player to use whatever works for him.
It's only going to be in 2 games so far - since it's not in their beloved CyberPunk yet, I doubt it's easy to implement, and they already have "100+" games with Reflex 1 that aren't automatically getting upgraded.
I'd also imagine that it probably won't be possible to implement for games that tie their physics engine to FPS, so that'll be a huge amount of games disqualified.
Implementation will be the main problem.
Physics shouldn't matter, the game logic still ticks at the basic frame rate, meaning physics is still calculated at 60fps even it the game shows 240fps. That extra fps is just extra smoothness, nothing more.
How are you guys testing latency? Maybe I need run some tests as honestly 50ms doesn't sound as bad to me. But I could be completely wrong. Not looked into it a very long time.
50ms is 20th of a second. Maybe I am also just getting old. Lol
Nvidia has it in its own OSD from the app. It doesn't show in all the games, but I works in CB77, for instance. And no, 50ms isn't that much. You can have it with around 60fps or more, especially with v sync.
Using DLDSR, 2880x1620, DLSS Balanced and the latency can jump around 40-60ms +/- at 180-200fps with raster.
Going down to "native" 1080p with same settings, fps around 220 +/-, latency goes to 12-14 ms.
So, if you add effects or techniques, latency will increase regardless of FPS. And while you'll feel the difference between 13ms +/- to 40 ms (probably, but not guaranteed), for most it wouldn't matter.
That, of course, assuming Nvidia's software offers relatively good latency numbers.