G-sync necessary?

[crisium]

As I understand it most pro gamers or wanna be's play at uncapped framerate or native x 2 + 1 (289 limit on a 144hz panel). It's for absolutely minimum delay in input lag, as well as minimum time in the game updating to your machine.

VRR at ~142 fps cap is smoother (usually have to go under native by 2 fps or so). Your minimum in games like CSGO and Overwatch at lower settings will never go below 142 so the argument about needing constant time is bogus: it is constant 142hz. It is "smoother" because tearing doesn't interrupt anything. But pro gamers prefer the game updating as constantly as possible and having absolute minimum input lag. Maybe the word "smooth" is subjective, but I don't think that's why progamers play at 289+ fps cap with tearing.

Personally, I prefer CSGO at 289 fps cap at 144hz with backlight strobing. I have tried this with Overwatch, but for whatever reason I prefer to do this one with VRR with a 130fps cap (used to do 142, but I think I agree with that video linked that Freesync has troubles that high). I notice tearing more in Overwatch and I really can't stand it. For CSGO it's fine though, for some reason. Dunno.

 

[tamz_msc]

Overwatch is really forgiving with aim, maybe that is the reason why it feels better to play it with VRR at sub-144fps. CS:GO, Quake, UT, are completely different.

 

[richaron]

*

Come on mate, don't give up now. There is an important distinction to be made, one which means you could walk away with a "draw".

I hope this isn't a "call out", but I don't think this conversation has reached it's natural conclusion yet. A conclusion which would be informative to many/all interested. I'll finish it myself if needed, but I also like to <insert potentially condescending teacher expression>.

 

[EXCellR8]

ah, these types of "is x necessary" threads never fail to amuse.

i've gamed on everything from bubble CRT to 4K IPS... does the adaptive sync make a difference? yes. is it necessary? not really if you don't mind a bit of tearing.

i love my freesync panel but that doesn't stop me from using it in conjunction with GTX cards as well. i can certainly tell the difference between tearing and not tearing so the answer will vary from person to person.

 

[w3rd]

Come on mate, don't give up now. There is an important distinction to be made, one which means you could walk away with a "draw".

I hope this isn't a "call out", but I don't think this conversation has reached it's natural conclusion yet. A conclusion which would be informative to many/all interested. I'll finish it myself if needed, but I also like to <insert potentially condescending teacher expression>.

If you do not know something, or understand what is stated as fact... then ask.

Don't rally on & on and bring up anecdotal stories and opinions about APUs and your gaming sessions, etc. Science doesn't care about your stories. And no one here has time to teach you about Time & rate of time/constant. And how there is a known constant frequency in which human's pace things, etc.

ergo: Time doesn't speed, up or slow down for normal humans, it remains constant. That is how humans are able to pull off amazing feats of accuracy, or precision. From MLB batters, to Acrobats, Formula1 drivers, to Astronauts.



In contrast to this, and in your disbelief... is the reason why I gave you a very simple (mindful) experiment that would illustrate exactly what phenomena takes place when you have heavy use VRR and trying to use the mouse with precision. It is not that hard to grasp (first search). What the experiment would show you, is exactly what He^ mention's in the posted link.

Imagine if you just take your mouse and moved it in a constant circle (on your screen) for about 25 seconds, & look at the pattern your ghosting makes. Looking at the size & shape. If you mentally logged that a few times (recorded it, or hotkey Hz, or randomize it), then changed your Monitor's Hz up from 60hz, to 144hz, then make the same circular patterns as before...

One would notice, that one series of circles are on average twice as large as the other series of circles. It goes back to science, there is nothing you can do about it. And sadly, You have spent a lot of posts, not arguing with me, but against science...! (ie: spacetime interval)




It's science, learn to live with it.

But as stated many times, only a small percentage of fast paced Gamers are bothered by it.. during gameplay some may even mistake it's effect as studdering and don't even realize it is their g-sync/freesync doing it. (FWIW: It comes down to One's in-game first person proprioception. Tearing doesn't take you off your mark using muscle memory, because you don't need to see the virtual window behind you, your exact memory of where it is, and your ability to make that mouse movement without seeing the screen is there.)

Heavy use of VRR during gameplay is like holding your hand steady & signing your name on a moving piece of paper. Sure you can do it, but that won't be your signature.

 

[Bacon1]

It is not that hard to grasp (first search).

Except most people in that thread are pointing out that the OP is wrong, because that's not how (most) game engines work at all.

Heavy use of VRR during gameplay is like holding your hand steady & signing your name on a moving piece of paper. Sure you can do it, but that won't be your signature.

Thats just wrong...

I've never seen that in any game I've played. Care to provide proof?

 

[Flapdrol1337]

Imagine if you just take your mouse and moved it in a constant circle (on your screen) for about 25 seconds, & look at the pattern your ghosting makes. Looking at the size & shape. If you mentally logged that a few times (recorded it, or hotkey Hz, or randomize it), then changed your Monitor's Hz up from 60hz, to 144hz, then make the same circular patterns as before...

One would notice, that one series of circles are on average twice as large as the other series of circles. It goes back to science, there is nothing you can do about it. And sadly, You have spent a lot of posts, not arguing with me, but against science...! (ie: spacetime interval)

Titanfall 1 might work that way, but pretty much all other games don't. And even in titanfall, it's based on fps, not monitor output.

 

[crisium]

And he is still avoiding discussing a constant frame rate with VRR - which is similar to v sync but without extra input lag. If a G Sync user is playing a game at 142fps in game cap (no extra lag from in game) and the minimums never fall below 142fps, it is constant, it is smooth. Those aren't the reasons pros play at 289fps+ unsynced. It's the search for absolute minimum input lag. And some of them have backlight strobing and prefer its motion clarity.

 

[richaron]

*SNIP*

Dude you are the epitome of both "a little learning is a dangerous thing" and also people who think they win arguments with quantity rather than quality.

Since it seems like you won't get to 'there' any time soon I'll just tell you you solution: You have confused the timing of the game engine with the timing of the frame rates. There is an important distinction here, which it seems you almost grasp.

Let's go back to the simplest games. I remember making an Asteroids clone back in high school. Imagine a few "pages" of code describing a simple game which the system runs from top to bottom, and when it reaches the bottom it starts back up at the top again: This is a game loop.

This simple game loop will run as fast as possible depending on hardware and software limitations, and obviously the game will run at a proportional speed. So my little space ship could zoom around faster with a faster computer. Everything would run faster with a faster computer, the bullets, the asteroids, etc. And equally everything would run slower with a slower computer. Also since the code is full of "ifs" and loops the "timing" for the whole system would vary depending on what's happening in game, and thus the frame rate varies correspondingly. <-This sounds like the level your understanding has reached.

The next basic step is to introduce a timer into your game loop. Since for simple games the system can process the code much faster than necessary, what you do is start a timer at the top tell the system to wait until X time has passed until it can restart the game loop. This means that, whilst the system can process the code within X time (which most should), the ship/bullets/frames are calculated and drawn at a consistent rate independent of hardware. In this situation only the slowest systems will run the game at a different "time rate" when they can't complete the code within the X time window. And in all of these situations the frame is drawn/updated on the same piece of code as the rest of the engine.

Now fast forward a bit, I'm sure you can imagine some smart people could design an engine which 1) would keep the ship and asteroids moving at the same speed on any/all computers and 2) would run at the highest frame rate and minimum input lag that any particular computer can manage. These 2 things are both very important and both very different.

...Bored of typing now so I'm cutting this short. I hope at least others get what I'm talking about and I've helped the world a little. I maintain your arguments are wrong because when you say "constant time" or even just "time" you are using the term with respect to frame time. That is what this thread is all about after all (omg I went on topic).

So your conclusions are useless because what you actually meant, and what your reasoning is all about, about is a consistent engine time (something very different from VRR tech).

 

[w3rd]

You can go on and keep reading your post.... over and over. Has nothing to do with science and known facts.

If you don't get it, or understand what VRR is and does to the monitor (many times per second), then I can't engage with you. You will have to fight someone else over science. Dishearteningly, Anadtech is kinda of an adult site, with heady people, so nobody is going to argue the basic principles of physics with you.

Ramble on bro...

 

[Bacon1]

Has nothing to do with science and known facts.

So where is your proof then? Sorry but posting a thread from some other guy on [H] who was also told that he was wrong isn't proof. Where is the science that shows that VRR monitors will change how your mouse moves in the game engine?

 

[Flapdrol1337]

You can go on and keep reading your post.... over and over. Has nothing to do with science and known facts.

If you don't get it, or understand what VRR is and does to the monitor (many times per second), then I can't engage with you. You will have to fight someone else over science. Dishearteningly, Anadtech is kinda of an adult site, with heady people, so nobody is going to argue the basic principles of physics with you.

Ramble on bro...

you're the one rambling, in some games mouse movement depends on fps, but never on screen refresh rate.

And VRR doesn't do all that much, the monitor doesn't really change frequency, it just doesn't start a new refresh until the new frame is ready or until it really has to.

 

[w3rd]

So where is your proof then? Sorry but posting a thread from some other guy on [H] who was also told that he was wrong isn't proof. Where is the science that shows that VRR monitors will change how your mouse moves in the game engine?

Show you proof of constant vs variable?

Ever use a regular multisync monitor, where you can sync at whatever Hz u want ? When in Windows 10 at 144hz my mouse movements and alacrity is different, then when I am in the exact same environment at 60hz. (It is not 1:1.)

Now, instead of having a set Hz, you have a random variable rate of frequency.. what do you think happens. Specially if your movements are no longer based on a set clock rate(movement=time over distance), but instead based on the the rate in which two hardware devices can sync a frame, randomly.

Thus, the time interval has changed and is no longer constant, there for distance drawn (by mouse), is no longer constant.

 

[richaron]

Show you proof of constant vs variable?

Ever use a regular multisync monitor, where you can sync at whatever Hz u want ? When in Windows 10 at 144hz my mouse movements and alacrity is different, then when I am in the exact same environment at 60hz. (It is not 1:1.)

Now, instead of having a set Hz, you have a random variable rate of frequency.. what do you think happens. Specially if your movements are no longer based on a set clock rate(movement=time over distance), but instead based on the the rate in which two hardware devices can sync a frame, randomly.

Thus, the time interval has changed and is no longer constant, there for distance drawn (by mouse), is no longer constant.

You are confusing yourself trying to sound smart, or you are deliberately being obtuse, either way I hope the readers of this thread don't pay too much attention to what you are saying.

Again you are confusing the "engine" time with the "frame rate" time. Again, these are two different things. So yes some of your arguments make sense, but not from the perspective you are presenting them. And since you've confused these two things your conclusions on this matter are flat out wrong.

Lets try this example: Imagine you are playing a first person shooter and you watch another character run across your screen from side to side. In modern games there is a disconnect between the speed the character runs and the frame rate of your computer. So someone with a 144Hz monitor doesn't see the character run faster than other people on 60Hz monitors. And also if there's a huge explosion or something in the background, which tanks FPS, the character doesn't correspondingly slow down. This is the big point I've been telling you over and over again but you still don't understand.

Your mouse movement examples are about the engine being able to keep a constant rate of time. But the frame rate, whether variable or not, has little to do with your examples. Please try to consider what I'm saying; it's painful for the rest of us here.

 

[Pookums]

I think it depends on what you like. For example some like mouse acceleration and some players do not. Whether Adaptive Sync or otherwise is useful may depend on the player.

In My Experience:

In Overwatch G-sync appears to result in an improvement in accuracy and player tracking due to the consistency of frame reference. Flickshots are ever so more precise. Reacting to changes in direction may be minutely more difficult only against characters such as genji and tracer, though its nearly imperceptible. Input Response time from mouse to screen is either nonexistent or vastly negligible.

Doom in my experience also saw an improvement from Gsync.

I do not play CSGO, but my understanding is that game is coded to respond to FPS gains.

I have not played UT4 with GSYNC thus have no comment. My last experience also involved an older 60hz monitor as well.

Shovel Knight seemed to have an input issue with GSYNC. The delay could be quite severe.

Single Player Refresh/FPS-locked games seemed to diverge in responses to GSYNC. Some work better with it on, others better with it off.

------------------

Gsync in my experience does benefit players in the same way that no acceleration on mice can aid players. If you want to follow or relocate the camera to a location you do not want variable reference mid movement. It maintains visual fidelity. Let me know when we have 360hz monitors and then 1000hz true sub MS monitors. Those two jumps should fix the majority if not all of remaining issues still in existence in regards to human perception vs visual input.

 

[4K_shmoorK]

This conversation is really confusing to me.

I was under the impression that if you want to minimize input lag: (ignoring refresh rate)

G-SYNC Off, V-Sync Off, Framerate Uncapped

Gets you the 'fastest' experience, in terms of lag, possible by eliminating any kind of framebuffer. Or use the 2(Hz) + 1 frame cap to keep GPU from getting to loud/hot

G/V Sync, to me, has always been about providing quality and performance, but not the 'best' performance. Combine it with fast refresh, and the lag contributed from Sync is minimized without sacrificing quality. Achieving low lag is not a new thing, as I'm sure everyone here knows (i.e. 4:3 High Refresh CRTs)

As far as this weirdness about human reaction time as it relates to the game engine and refresh rate. All are independent, no? Engine renders frames. Stop. Monitor displays frames. Stop. Person acts on Frames. Stop.

Game Engine constant for everyone, not dependent on hardware or human capabilities. Right?

 

[Pookums]

I had a long response Typed out to be informative, but realized I was becoming too technical, and it was probably harder to follow than this thread. So I'm going to discard and simplify.

Gsync off, vsync off, Uncapped only benefits you if can maintain FPS above 144 frames(usually well above it), the network/server tick rate is uncapped (or extraordinarily high say 300+) and the game engine determines collision based on visual traces of FRAMES. If a game engine calculates a point to point collision based on Time Precise MOMENT of INPUT, updating constantly to any novel change, or the engine uses weapons that are not hitscan -> it is pointless to have a discussion based upon input lag. Only frame TRACES can be affected.

If the server tick rate is limited any benefits to Traces are rendered moot. If your FPS is below 144fps, benefits are negligible or nonexistent.

Hitting all 3 points in the multiplayer engine are RARE. In a single player game they are absolutely pointless unless like Shovel knight it affects the gameplay detrimentally.

Additionally, Without adaptive sync you will have a variable frame of reference. This just means that visual gaps across a constant velocity between refreshed frames can be shortened or extended depending on graphics output and lag. Adaptive sync in its efforts to limit tearing, will attempt to match frame output and refresh rates resulting in less variable frame reference. GSYNC therefore allows more precise tracking and a smoother visual clarity, plus its initial focus of limited or no tearing.

As a result you have to compare GSYNC's precise visual reference to ALL SYNC OFF - UNCAPPED's Trace input response benefits in viable games. This might depend on the player.

 

[crisium]

Competitive gamers want minimum input lag. That is the only reason, unless they are also combining it with backlight strobing for better motion tracking. They want every millisecond, so even G Sync at 142hz being 3ms slower than uncapped is unacceptable to these hardcore people. There is nothing to be confused about.

 

[Bacon1]

They want every millisecond, so even G Sync at 142hz being 3ms slower than uncapped is unacceptable to these hardcore people. There is nothing to be confused about.

Not sure if you posted the wrong graph, but the first one didn't have gsync on in any test and the 2nd only had less than 1ms difference in worse case, 1ms difference better in best case and .3ms difference avg in overwatch.

There is no input lag from gsync/freesync, only the extra from reducing the FPS.

If you had a 300hz *sync monitor you'd have no additional input lag or issue compared to 300 hz w/o. I doubt you'd notice tearing at 300hz refresh rate though

 

[thehoj]

So i see that the GTX1080 has dropped in price now that the ti version has arrived. I was pondering getting something like the following to go with it:
https://www.amazon.com/gp/product/B0149QBOF0/

That seems to be quite a bit more expensive that non-gsync monitors. Would i be losing out on alot by not bothering with gsync?

I like playing shooters such as titanfall, battlefield and overwatch.

I have some experience with Freesync, not sure if it's exactly the same as GSync, but from my perspective it's barely noticeable. That being said, I don't know if I'm just blind to tearing or what, but I just don't notice tearing at all. I'm running an RX480 and an Asus MG279Q (144Hz Freesync), and running at 144Hz is absolutely amazing, but I honestly don't notice a difference between freesync on and off.. If it's an untrained eye on my part, I don't really care I guess.. Ignorance is bliss! haha.

I play Titanfall 2, Overwatch, Quake Champions, that type of thing, I play on low enough settings that I'm usually in that 90 - 120 fps range in what I play, so maybe that's a factor as well.

 

[Bacon1]

but I honestly don't notice a difference between freesync on and off.. If it's an untrained eye on my part, I don't really care I guess.. Ignorance is bliss! haha.

I play Titanfall 2, Overwatch, Quake Champions, that type of thing, I play on low enough settings that I'm usually in that 90 - 120 fps range in what I play, so maybe that's a factor as well.

You should notice a lot of tearing if you are playing at that FPS. Does your monitor have a hz overlay? I wonder if you aren't using VSync or borderless windowed and not true fullscreen?

Download the gsync pendulum demo. Turn off vsync / freesync and turn it to the fastest animation and the bar (show test pattern bottom left). You should see tearing easily when freesync is turned off and it will smooth out the bar when turned on. Check your OSD to see if there is an option to show the current refresh rate (my Acer shows it top right) as it should match the FPS and constantly go up/down as the fps does.

 

[crisium]

Not sure if you posted the wrong graph, but the first one didn't have gsync on in any test and the 2nd only had less than 1ms difference in worse case, 1ms difference better in best case and .3ms difference avg in overwatch.

There is no input lag from gsync/freesync, only the extra from reducing the FPS.

If you had a 300hz *sync monitor you'd have no additional input lag or issue compared to 300 hz w/o. I doubt you'd notice tearing at 300hz refresh rate though

Huh? I'm not talking about G Sync lag. There is no G Sync lag. I agree with you.

I posted this to help explain why progamers and wannabes prefer uncapped to 142fps G Sync. Mainly so that one troll understands (smoothness has no bearing here). G Sync results in increased input lag compared to uncapped (if you can actually drive the full uncapped). Not from the technology, but from 300fps resulting in less input lag than 142, regardless of the monitor's refresh rate. I do not disagree with you.

The first chart 118FPS @ 120hz is meant to be G Sync. I think I took the screenshot a second before he put in the G Sync result, but there was no difference since G Sync has no lag as you said.

 

[Bacon1]

There is no G Sync lag. I agree with you.

I posted this to help explain why progamers and wannabes prefer uncapped to 142fps G Sync.

Ok so yeah we were saying the same thing then

Yeah I'd never give up *Sync to have slightly (not even noticable tbh) response time. Heck the tearing would probably bother me more than the ms delay. Not to mention many monitors have higher ms delay than advertised.

 

[crisium]

I keep going back and forth with what I prefer. After earlier preferring 142fps Freesync for Overwatch, I have been playing more with uncapped and backlight strobing. The tearing sucks, the strobe cross talk sucks, but it makes it easier to track horizontal movement and I think I play better, most of the time. But it's all due to the backlight strobing for me, I can't notice any input lag difference. If I weren't such a try hard I'd easily prefer 142fps Freesync because of the no tearing and no cross talk, FWIW.

 

[Bacon1]

I keep going back and forth with what I prefer. After earlier preferring 142fps Freesync for Overwatch, I have been playing more with uncapped and backlight strobing. The tearing sucks, the strobe cross talk sucks, but it makes it easier to track horizontal movement and I think I play better, most of the time. But it's all due to the backlight strobing for me, I can't notice any input lag difference. If I weren't such a try hard I'd easily prefer 142fps Freesync because of the no tearing and no cross talk, FWIW.

Well Overwatch is an odd ball as well because the game simulation time is directly tied to FPS as well which is dumb. So yeah you want the highest FPS possible in it. Battlenonsense also covered that here: https://youtu.be/nilBzupE4Cc