Well not really because 50 series can interpolate more. Doubling the interpolation doesn’t double performance just like turning on FG originally didn’t double performance.
It's not just a feeling but fact. NVIDIA is stretching how much they can lie with media parroting the lies before crap hits the fan. So far they've succeeded, sadly. Any scaling, framegen and whatnot should be left out from any comparison which doesn't specifically compare said feature.
Weeeeeell, that's very subjective.
I don’t think FG really adds latency, it just doesn’t remove it. So 40 with FG feels identical to 40 without, but it won’t feel like 80. For many (like myself) this feels weird because we are used to input latency decreasing as frame rates increase.
Isn't that's the point? Look at the Nvidia comparisons, they aren't comparing 5070 to a 4090 there, they are comparing it to a 4070. Which means that they are comparing it to a _slower_ card, and then they are adding MFG on top of that card's result with FG - which means that you've already paid the FG latency penalty on the 4070 and from that you're just getting "free generated frames" on the 5070 without additional latency. So as I've said such comparison seems entirely valid - you're basically comparing the improvements the new FG nets you in framerate over what you would be getting without it. If they'd be comparing FG vs FG (2x mode or whatever its called on 5070) then that's make a lot less sense in fact when you can use a 3x one and get more frames "for free".
It's less subjective than title to title dependent. Some are fine even with 30 fps baseline and for some you'd want 60+. On average though, if it's not a twitch shooter and if the baseline frametime health is good my own estimates show that a 40 fps baseline is fine for DLSS FG.
Sure but there’s going to be an even larger latency disconnect between 2x and 4x. No, there won’t be added latency beyond the usual FG penalty, but going from 40->80 will feel less jarring than 40->160 as now we’re seeing 160 while feeling 40.
Obviously leaps like Samsung 8nm->TSMC 5nm aren't likely to happen again with major GPU releases, but there is definitely still a couple more generations of reasonable improvements to go before I think we're gonna face something of a nightmare scenario for high performance consumer processors.
Well DF just put out a video showing that from a base of about 50ms with original 2x frame generation, going to 4x only incurs another 6.5ms or so of input penalty. That's very acceptable if you can take advantage of the extra visual fluidity.
What's "latency disconnect"? You're not getting worse latency, you're just getting more frames. And as for latency this in general isn't inherently tied to the framerate. You could be getting high latency even with high framerates in some titles so getting the same with MFG would look just like an engine specific. The only way to see that its not would be to run the game at a similar framerate w/o FG - which for MFG modes will likely be impossible to do even by going to the lowest in-game setting.
I see no reason why they would be any worse, and that 1 frame in FG was basically indistinguishable from a "real" frame when viewed at high fps. The only FG artifact which is visible sometimes is the HUD interpolation but it's not a big deal.
But they are not. Again, look at the comparisons. All 50 cards in them are _faster_ than 40 cards which they are being compared to even without FG.
This is true, but it's also one frame at the "new" framerate, not the original framerate. So long as the FG framerate is at least double the original, the latency will be at least equal to the original. There's a strange gray area where the FG rate is less than double the OG framerate, where latency does suffer a bit. What I'm not entirely sure of is if latency continues to drop with FG framerate > 2xOG framerate... I suspect this becomes a property of the app code input sampling.
When the FPS counter says 160 FPS but it feels like 40 FPS, because 1 in every 4 frames is interpolated and your input has no affect on it.
This is just factually incorrect, given the same setup with the same controllers and display a higher frame rate will almost always give you a lower latency unless the display bugs out at higher refresh rates (fortunately not a modern issue).
But they are? Most of their comparisons are with framegen on. The 5070 will almost certainly not match the 4090 without the new 4x FG yet it’s the literal tagline Nvidia used when introducing it lol.
I agree, the extra fluidity will be nice but it will never feel like an actual high frame rate setup.
This is factually correct. Latency is a product of the engine more than the framerate. You can make it smaller by improving the framerate but at any given framerate the latency can be wildly different between different engines and even just titles on the same engine. So just saying that 150 fps with 30ms latency would somehow "look incorrect" is completely wrong because you can easily get the very same result in some title without any frame generation. In other words you will not be able to tell from latency alone that a game is using framegen unless you'll start changing settings.
On both 40 and 50 cards which they are comparing, yes. So it's not a FG vs no FG comparison, it's a FG vs MFG one. And that logically is a lot better one.
The only place where they've made that comparison is the marketing slide in the keynote. You won't find it anywhere on their website.
I think it’s very clear we are talking about within a given game so idk what different engines has to do with this. Inside a game given the same setup, increasing frame rate decreases latency. Why do you think CS pros play at like 400 FPS?
so we agree that Nvidia was deceptive and using 4x FG vs 2x FG to claim the 5070 is faster than a 4090 is wrong?
it’s better but it’s still misleading and incorrect.
If you take a single game and run it at 40 FPS interpolated to 80 FPS it will feel like 40 FPS and not 80 FPS, due to latency. To get 80 FPS yes you’d have to lower settings but I think we’re making this hypothetical more complicated than it needs to be, with FG you get more fluidity at the same latency, with more raw frames you get more fluidity at lower latency. This is the point I am making and that’s why they aren’t directly comparable.
Sure but you're talking about some inherently obvious latency from using FG - which isn't a thing at all. You can use FG or not use FG and get a similar latency at a similar performance which means that the idea that MFG would somehow produce a clear "latency disconnect" is just wrong. The only thing which is important for latency is for it to be low enough for you to be able to play the game without input issues. The actual figure can be whatever as it depends on a game, your own perception, you input method and a bunch of other things.
It's marketing ¯\_(ツ)_/¯
No, it's not. You will be seeing that more frames on a 50 series card. Yes you won't get better latency on it but since you're comparing FG to MFG it is expected already.
