Dictator93
Member
Andy, send this commentary along with some of the other light boost stuff over to the engineering team.
-
Hey Guest. Check out your NeoGAF Wrapped 2025 results here!
G-SYNC - New nVidia monitor tech (continuously variable refresh; no tearing/stutter)
- Thread starter artist
- Start date
Andy, send this commentary along with some of the other light boost stuff over to the engineering team.
One interesting aspect is that with G-sync a monitor wouldn't necessarily have to be capable of displaying e.g. 400 different images per second to benefit from such a high transmission rate. The only thing that counts is how quickly one image can be transferred from the GPU to the monitor. Of course, we are talking about a difference of a few milliseconds compared to 144Hz here.
It's limited to 650Ti Boost and above, the rest of the Displayport 1.2 capable Kepler family doesnt get this.
The % of Kepler based cards above 650Ti Boost. Isnt that what you were implying?
It's 175.
I haven't extensively tested this yet. Since while testufo did show an improvement with lightboost it was not worth the discomfort, so I just gave up. I do remember cranking up the brightness. I want to give this another shot though, taking these factors into account.
Alright. I stand corrected in my belief that this is going to be better than lightboost in every way but this tech does make lightboost a lot less attractive because lightboost demands your games to run a much faster refresh rate while gsync in the future can offer benefits to people on a gaming notebook with an integrated IGP.
Gsync does have you take a hit to colors. As good as you and Vega have worked on addressing this it was only a sacrifice I was willing to tolerate. I'm hoping gsync can be applied to IPS monitors that already exist like the Catleap.
Regardless it is very cool that people who are still interested in the lightboost trick will benefit even more from gsync if they don't care as much about the caveats I mentioned.
Andy said that they are trying to get it to 130.
Thats how technology works.
The confusion in pricing came from Anandtech which said the price of the module was ~130 and Nvidia were trying to get it down to 100.
Sir Abacus
Member
175 is fine. 130 would be better. 100 would be fucking fantastic. At 100 I'd buy two more of them and two extra panels to swap out the VS247H-Ps that I'm currently using along side the VG248QE.
This is what's wrong with the industry. A multinational company like Nvidia spends million of dollars in R&D dreaming up some new outstanding technology and manufacturing it and then they decide to keep it in-house and turn a profit from it? It's ludicrous. Give the tech to AMD and whoever else wants it. As a matter of fact, Nvidia should just stop making GPUs altogether and become a charity, spending all of their money to develop technologies for everyone else. For fuck's sake, Nvidia.
The Hermit
Member
I read a lot about and just watched the live demo. Let me see if I got it right:
So with this my Gpu would be free from V-sync and a 30-45 FPS game would feel (lol) like 60FPS to me. Is that right?
So with this my Gpu would be free from V-sync and a 30-45 FPS game would feel (lol) like 60FPS to me. Is that right?
I doubt that 30-45 FPS would feel like 60 (though some who have seen it claimed that). However, any variable frame rate will feel much better than on any current display.
The more important part is that:
- You will not get any tearing but also not get any additional input lag from traditional v-sync or triple buffering
- There will be no judder associated with fitting a variable frame rate to a fixed display refresh rate
- You will get the absolute minimum possible input lag from a frame being rendered to it being displayed possible, in any situation
This is what's wrong with the industry. A multinational company like Nvidia spends million of dollars in R&D dreaming up some new outstanding technology and manufacturing it and then they decide to keep it in-house and turn a profit from it? It's ludicrous. Give the tech to AMD and whoever else wants it. As a matter of fact, Nvidia should just stop making GPUs altogether and become a charity, spending all of their money to develop technologies for everyone else. For fuck's sake, Nvidia.
Well, there's a decent argument to be had here to counter this. The gaming industry is not so large that it can afford to cut the throats of major players. Just the same, a company should not have a stranglehold on a game changing technology like G-sync just because they got there first. They should reap their first mover benefits, then open the technology up or at least license it for a minimal fee, because that's what is best for the industry in the long run. This is why Mantle won't be 'open' to Nvidia until AMD has used up its sales potential. The same goes for G-sync.
The whole idea is to break free from any one company dictating the future of gaming. All four guys on the panel expressed that desire. How terrible would it be if Valve solely owned the ability to digitally distribute games? The industry would never move forward. It's the same with tech that is important to advancing the art. Both companies placed different bets. It just so happens that G-Sync requires zero effort on the part of devs, and is flat out the more important technology. Sweeney and Carmack more or less implied that Mantle was not a good idea in a world where Microsoft is still heavily invested in gaming and won't relinquish DirectX. They were also excited about SteamOS which is a move by Valve to move gaming to neutral ground.
AMD and Nvidia have every right to reap the benefits of R&D. But similar to sharing tech in the automotive industry, companies should not sit on something that can help the industry as a whole.
This is what's wrong with the industry. A multinational company like Nvidia spends million of dollars in R&D dreaming up some new outstanding technology and manufacturing it and then they decide to keep it in-house and turn a profit from it? It's ludicrous. Give the tech to AMD and whoever else wants it. As a matter of fact, Nvidia should just stop making GPUs altogether and become a charity, spending all of their money to develop technologies for everyone else. For fuck's sake, Nvidia.
People aren't looking for Nvidia not to turn a profit from it, they're looking for Nvidia to charge AMD ten or twenty bucks a card for it for the next 20 years rather than just boosting their turnover for a year or two until someone else creates an incompatible alternative and -does- license it out.
MadScientist
Banned
On the lower end (games running below 30 fps or between 35 and 55 fps) the problem could more effectively be solved by investing $50 - $100 more in a graphics card, which not only solves most of the the sync-issue, but also gives you more frames per second.
That really depends -- sub-60 FPS is not necessarily always on the low end. I often play games at 45-60 FPS, but it's not for lack of (cheap) GPU power. It's because I value the IQ more than a constant 60 FPS. G-sync would allow me to push IQ even further without compromising fluidity and responsiveness too much.
Digital Foundry piece on G-Sync:
http://www.eurogamer.net/articles/digitalfoundry-nvidia-g-sync-the-end-of-screen-tear-in-pc-gaming
So exciting
http://www.eurogamer.net/articles/digitalfoundry-nvidia-g-sync-the-end-of-screen-tear-in-pc-gaming
So exciting
IMO that's one thing that's being overstated - for vast majority of software, the VSync period (16ms or less if you're on high-end monitor) is a small fraction of the input latency, and when frame does stall on VBlank it's almost always less than the whole period. Once you add network latency on top of that the VBlank period is really just miniscule in comparison so the impact on human reaction times is minimal.Durante said:
I'd argue that stutter probably has more impact on your reaction speeds because it fucks with brain processing (not directly changing the input latency).
The Hermit
Member
Since I begun PC gaming I do everything to keep my FPS at 60. First time I saw how fluid everything could be, I couldn't go back, even if that meant lowering my IQ. I never had problem with input lag though,(never used 3overrider thing) but like i said i sacrificed everything.
This tech aparently solves that constant dilema. I am still having a hard time believing a 45FPS can go unonticed or even a 50, but the few videos I've seen show how clear the diference is.
Damn, this is too much. It fixes some of the Oculus Rift problems too.
I have to see it with my own eyes, its too good to be true.
Edit: Reading the DF article, aparently there is some ghosting when its below 60. I can accept that trade, this thing is mine.
Dreams-Visions
Member
That is really cool, can't wait to get one of those.
This is true, particularly compared to network latency (though not every game has that). However, the best we can do is eliminate one source of latency at a time, and that's what G-sync does for V-sync stalls.
A bit of a silly article title (I'd rather call it an end to judder or forced 60 FPS locks, since most people I know always played with V-sync on), but sounds good otherwise.
Not currently. Maybe eventually.
Nvidia's mentioned it right on their G-Sync page;
http://www.geforce.com/whats-new/ar...evolutionary-ultra-smooth-stutter-free-gaming
Sir Abacus
Member
No.
You can make sure you have enough excess power to run 60fps locked at all times. But apart from that, no.
No, it's a genuine solution to a problem that has plagued gaming for a very long time now, and which could only be mitigated by various trade-offs previously.
How important the problems it solves are in the grand scheme of things is debatable, but it being a genuinely innovative and imminently useful technology is not -- just look at the glowing endorsements by industry luminaries such as Sweeney or Carmack.
To benefit from it you should be experiencing problems first. If you already have silk smooth fps, you don't need it.
Getting faster GPU not to have "sometimes FPS drops to 30" problem is (imo, much better) option.
One point about g-sync is rather strange, it doesn't work with some games (I can't quite imagine why)
It's not about dropping to 30 FPS. Its about being able to drop to 56 or 49 FPS for a while and having them feel like 56 or 49 FPS, rather than like 30 FPS -- or perhaps worse due to temporal judder! -- like current technology.
I would assume this applies to those games which try to do their own frame pacing in unusual/weird ways (e.g. double v-sync, sleep timers for the V-blank period or some such sillyness). This is rather rare, but it happens from time to time (and often already causes other issues on some setups).
It'll be nice to finally have a definite way of playing Dead Space
Self_Destructive
Member
Can't you just disable vsync in the game and force it though D3DOverrider and have 60fps vsync'd gameplay?
Yeah, it definitely works, though unless they did something funny this should still make it better (but what game wouldn't that be true for?)
Of course, but jittering and mouse-lag. G-Sync will be the definite solution
. I knew about D3D Overrider trick, fortunately. All those strobe lights demand it.Didn't the ingame Vsync work in 2? Been a longass while since I played. I think I'm due for a replay of the first, actually
.This sounds fucking fantastic. It'll be so great to not have to worry about D3DOverrider, or V-sync lag ever again, just to rid myself of screen tearing. Hahahaha, the feel of 60fps@35-40fps, delicious.
mdrejhon
Member
The latency mathematics is rather simple (to me, as a amateur vision researcher, being Chief Blur Buster). Generally, it will have less input lag than VSYNC ON, less input lag than LightBoost, but slightly more input lag than >144fps during VSYNC OFF (since G-Sync will cap out at 144fps), based on the mathematics I've successfully gleaned from the digrams given already in nVidia's presentations.
The impression I'm getting is that G-Sync doesn't appear to add any framebuffer delays beyond existing electronics. It will still be real-time top-to-bottom scanout of the refresh (example high speed video, non-LightBoosted mode), just no longer forced to wait till the next scheduled refresh.
To clarify this a bit, shouldn't it be >*monitors max refresh rate* during VSync off?
The current numbers are only 144hz because that's the one monitor they have it implemented in thus far but in the future, it'll be whatever monitor a persons using.
Chance Hale
Member
Are they going to jack up the price of monitors with it built in? Not sure whether I should wait for a monitor with it built in or just grab one of the DIY kits. But then they mention crazy 4K models and ugh, so much money!
They said next year for the DIY kits but are the built-on ones coming sooner?
They said next year for the DIY kits but are the built-on ones coming sooner?