It is. I am just not understanding these claims that the SK frame rate limiter is “better” than G-Sync/Freesync, when they do completely different jobs.
Yeah I did notice that, was a little bit weird.
Thanks for your findings but don’t get banned for these things!
This is like… really wrong no matter how one sees it. I can understand that having a stable frame pacing at 30 FPS is more enjoyable than an unstable frame pacing combined with a variable frame rate, because that’s definitely the case, but at no point will 30 FPS “feel like” it’s running at 60 FPS unless one is incapable of recognizing the massively more stuttering experience that 30 FPS is.
The only time a capped 30 FPS is more enjoyable or experienced as less “stuttery” is when the alternative is a variable frame rate between 30 and upwards with uneven frame pacing. This is, for example, something that DigitalFoundry touches upon quite frequently – many console games can feel ‘smoother’ v-synced and locked to 30 FPS.
And even then most of the stuttering experience can be solved either by capping the game to 30 FPS using V-Sync synced to each second VBLANK, or by capping to 30 FPS using some form of limiter – most of which, regardless of their frame render variance, should take care of the majority of the issue.
Special K’s frame pacing is quite good, yes, but it’s not magical – a 30 FPS cap isn’t going to feel like a 60 FPS experience just because the 0.01% (or even less!) frames are rendered within a 0.x ms variation from the rest.
G-Sync or FreSsync by itself has nothing really to do with how “smooth” the image is experienced – unless, again, one is comparing with a scenario with a highly variable frame rate where the FPS goes from one end of the spectrum to the other. But in those cases basically anything would feel the opposite of smooth, as there’s no frame pacing to speak of to begin with.
It wouldn’t matter if you had a VRR monitor or not – both would feel uneven as their frames are displayed at different lengths of time on the screen; some would be displayed on-screen for 16.67ms (60 FPS) while others would be displayed on-screen for 25ms (40 FPS), or even 33.33m (30 FPS). The lack of frame pacing results in a lack of “smoothness” in the overall visual experience.
But again, that has nothing to do with VRR per se and everything to do with what options you use in games, what average FPS you sit around, and whether you use something (anything, really) to cap the FPS to improve the frame pacing (anything is an improvement when comparing to nothing at all).
The only difference between VRR and a non-VRR monitor is whether the user would have screen tearing on top of that, and additional input latency if they went for the V-Sync option.
The key to getting good overall frame pacing with minimal stutters and whatnot is, in general terms, basically:
-
Decide upon a target frame rate. The lower, the more “stuttery” the overall experience will be experienced as simply due to the lower number of individual frames displayed per second.
-
Lower the visual quality settings until you can hit that target frame rate in the vast majority of the scenes you can expect in the game. Preferably add an additional breathing room above the target FPS to give the system some room to “breath” if necessary (sudden resource spikes etc). E.g. if I want to target 60 FPS, I typically ensure that the game can run at above 65 FPS at all time.
-
Cap the FPS to the target frame rate using some optional method such as a FPS limiter or V-Sync.
- Many games have a subpar built-in frame rate limiter – using a third-party FPS limiters (RTSS or SK) or e.g. Nvidia’s built-in
Max Frame Rateoption (aka NVCP FPS limiter v3) is highly recommended in those cases.
- Many games have a subpar built-in frame rate limiter – using a third-party FPS limiters (RTSS or SK) or e.g. Nvidia’s built-in
-
If you target a non-standard refresh rate (e.g. 40 FPS, 70 FPS, etc) then be sure to use a VRR monitor with G-Sync and FreeSync which would allow you to remove screen tearing that would otherwise occur.
-
There’s also the option to force a non-standard refresh rate on the monitor and then use V-Sync, but it usually requires the use of CRU (Custom Resolution Utility).
-
If you want to lower input latency at the desired target frame rate, VRR is a must unless you accept the screen tearing that occurs.
-
-
Enjoy your gaming experience with good frame pacing and minimal stuttering.
Additionally, Special K can improve the situation further by improving the overall state of the swapchain of the game, but those sorts of changes only take you so far – the above five points still dictate the overall approach to improving frame pacing regardless.
No worries, I don’t actually use Special K during regular play. What I also like of Riot Games is their use of “skill-based metrics” in determining whether the use of a third-party tool will be something they ban for or not. Basically if they feel that the third-party tool conflicts with the overall competitive nature of their game as well as artificially influences the supposed “skill” of a player, it becomes a bannable offense.
As an example DXVK might introduce flip model and synced refresh rates on the monitor (and overall lower input latency), but that alone won’t suddenly make me play like… whatever highest skill rank League currently have. Overall the use of it would barely even affect my overall skill in the game – I’d be able to react some milliseconds faster to plays, but that’s about it. I suddenly wouldn’t be able to go 1v1 vs. Lee Sin with an Ezreal myself… (I friggin hate Lee Sin in URF)
On top of that I basically only use the Practice Tool when I’m throwing Special K at the game 
1 Like
I am going to throw a wild guess and say that these guys are confusing G-Sync/Freesync with fast-sync, and then confusing fast-sync with traditional V-Sync?
Otherwise, I struggle to understand how anyone can claim that the VRR implementation and software frame-capping are even closely related to one another.
Basically I don’t get the same blur (Ghosting) or lag with SK but with regular vsync I do get much more lag and blurring. There is a clear difference actually. Basically a slow motion effect takes place when I use conventional Vsync. But with SK that slo motion doesn’t occur.
The game yes plays a little bit slower at 30 but not as much if I was using something other than SK. Its highly noticeable without SK framepacing.
With SK I do tend to get higher fps in some areas and lows a bit higher but highs are lower as well. So tighter in framepacing. WIth Watch Dogs Legion with regular vsync I noticed my fps hits around 28 to 30s. With SK I hit avg of 40 fps for some odd reason. But still hit low 30s. It’s just more smoother and not blurry like when using SK. Though SK hits lower fps in Watch Dogs Legions the jitters / stutters are much less and the laggy type feel of movement of character is much less as well. While with conventional Vsync that is much worse and more noticeable. WIth backbuffer setting and latency setting I can almost do a speed up effect that negates the low fps effects that happens with conventional vsync.
Basically without SK I get massive amount of jitters and stutters at very high level. Just simply using SK and capping at 144 it reduces all that at insane amount. Pretty much butter smooth in comparison. I do have AMD card and Freesync Monitor. Even with freesync it wasn’t as smooth. If I am not mistaken Freesync is supposed to make all fps ranges feel much smoother without too much latency right? At least that’s what I got out of its description.
I actually do this with SK believe it or not. Using SK is the best way to achieve this with best results. Using other limiters not so much at all.
This is a hardpass for me b/c freesync on my monitor causes horrible flickering at levels below 60 fps. And b/c its a VN panel its much worse. Its a known bug actually and something I can’t fix. I have freesync 2 monitor which has range of 48 to 144 but its very bad when goes below 60 fps. The flickering is like a strobe light. So I am forced to turn of VRR / Freesync.
Considering most games I play don’t actually run above 60 fps at all times. So that’s why I use SK over any other method b/c its superior than any other method at those low level fps ranges. B/c I do not get the slow motion blur effect (Ghosting) with SK or get extreme lag. And how I lock most of my games I remove just about all stutters and jitters. Even if I did that with conventional Framelimiters it doesn’t do it nearly as good as SK.
Edit:
Also my monitor has a unique feature on it that when Freesync is off it allows much faster response timing and input latency mode that reduces the issue by a ton. When freesync is on its not as good. I actually noticed latency frames would be much higher when this setting is off. But when response time is at highest it drastically reduces those frames. I mean by like 20 frames or so.
Also do realize Kaldaien himself stated that his framepacing limiter does better job for most part than Freesync or G-sync. Though recent findings when using SK with Freesync / G-Sync it does a better job. But if you compare it by SK alone verse using G-Sync by alone, SK wins against those methods.
Also Kaldaien stated that G-sync by itself actually has much higher latency then conventional Vsync.
Yes, but also no. Freesync does absolutely nothing to frame times; in fact, nothing changes on the computer side of things. The mechanism is monitor-specific, and it reduces tearing/stuttering by ensuring that the monitor can get the next frame at the requested timeslot sent by the GPU instead of having the GPU wait for the monitor to be ready.
I think it’s a well I need to read this more but a mixture of V-Sync, frame rate and frame timing.
They work well together and SpecialK can help but hardware support through adaptive sync works really well.
Most times, some of the dynamic supported ranges particularly earlier FreeSync models can be pretty rough and it looks like G-Sync hit some issues with newer hardware and what appears to be LG having problematic firmware.
Framerate time and smoothing for AMD I think is down to how Chill isn’t a terribly good framerate limiter you can use for that but it overshoots by 1 - 2 FPS and enhanced sync is bug prone though more so for the 5000 series of GPU’s AMD’s still not quite fixing it although they have improved other issues.
(The mouse cursor was problematic for a while even among other weird and strange issues.)
I think some of the AMD AGS implementations also leverage FreeSync 2 advantages but it might just be HDR specific stuff so less of a issue.
(Premium Pro and all the other marketing names AMD felt they had to stick to it instead of a initial decent stricter criteria for manufacturers. 
)
And for G-Sync to jump back to that Wonder if HDMI2.1 here and the new hardware will see a new scalar module with additional advantages plus the newer panel features like black frame inserts which going by his posts Kaldaien is really a fan of even above G-Sync support itself for that display heh.
Well AMD has some work still for their driver side of things, think NVIDIA’s various framerate limiters and stuff is generally acceptable to be pretty good with minimal tweaking but there’s a lot that’s only visible through Inspector as far as I know too for further tuning or their own profile specific tweaks.
EDIT: First thing in the morning let’s attempt to read up and reply to complex issues, what can go wrong ha ha.
Now to wake up and re-read this all more thoroughly again I must have missed a few things.
EDIT: Well of course SpecialK can help especially if games like Watch_Dogs and what not overshoot or do terrible things to input latency and what not, some games or game engines well at least there’s a tool for improving it and then things can go back to being smooth.
Possibly with a few crashes and bugs and save game breaking issues ha ha.
Sigh, well at least GAAS ensures longer patch support and a lovely state of release for the game, yeah some of these titles definitively needed a bit more work.
Funny you mention BFI. My monitor has that if I disable Freesync and use Fastest Response. And that works really well if FPS is above 60 fps otherwise it strobes quite badly like Freesync 2 does. But it also makes my screen much darker too when I use BFI mode.
Again, this is some weird misconception you have – G-Sync doesn’t do anything about the overall framepacing.
Any frame rate limiter will produce better frame pacing than none at all, which is what using solely G-Sync would be like.
This here needs the appropriate context to put in perspective. Kaldaien have made various findings and whatnot, but in general, no, G-Sync doesn’t have higher input latency than conventional V-Sync. It’s all relative to what context, and without knowing Kaldaien’s specific comment/statement, it’s impossible to say what exactly he was talking about.
But no, in general G-Sync does not have higher input latency than conventional V-Sync.
He’s probably talking about this instance Topic-Free Mega Thread - v 1.11.2020 - #2410 by Kaldaien
Which is a reference to this; Topic-Free Mega Thread - v 1.11.2020 - #2381 by Kaldaien
This as well
@Diakou
Thank you that is where i remember seeing some of his talk about G-sync. ANd I quoted another place where he mentioned G-sync verse SK limiter.
Mhm, there’s a ton of various factors involved here that one needs to keep track of, and they’re all at the end of the day simply tools to influence and achieve the desired end result.
-
Conventional V-Sync
-
“Fast Sync” or similar multi-buffering approach to allow an unrestricted frame rate above the maximum frame rate while still syncing one buffer to VBLANK using V-Sync.
-
VRR such as G-Sync / Free-Sync which removed tearing and lowers input latency at frame rates below the maximum refresh rate.
-
Frame rate limiting
-
Frame pacing
-
How the overall swapchain affects the results
Then there’s the various special cases to keep track of as well:
-
G-Sync (possibly even FreeSync) without V-Sync enabled does not allow the technology to make adjustments for minor render time variations, which causes tearing at the bottom of the monitor due to uncontrolled output.
-
V-Sync with an FPS limiter at just below the V-Sync target can decrease the input latency vs. conventional V-Sync.
-
“Fast Sync” can still cause a stuttery experience if the frame rate isn’t a multiplier of the refresh rate (so e.g. 120 FPS on a 60 Hz monitor, etc).
-
etc…
To achieve the best result it’s entirely all about what one’s own target is, and then it’s a matter of using the tools available to the user to attempt to achieve that setup.
-
Kaldaien, for example, makes use of Black Frame Insertion on his OLED to remove the otherwise present motion blur of the monitor entirely. To help him achieve this, he’s using SK’s frame rate limiter which is one of the few FPS limiters producing the best results – minimizing stuttering that might otherwise occur if the frame output is off.
-
I use G-Sync exclusively and so targets a below-maximum-refresh-rate on any game I play. I use SK primarily to help me achieve a capped FPS as well as flip model in games that otherwise does not use it (which allows me to evade fullscreen exclusive as well as Nvidia’s DWM hack). But beyond that, my needs can also be fulfilled by using DXVK alongside RTSS or similar third-party FPS limiter, which is my go-to solution if Special K is incompatible for one reason or another.
-
etc…
1 Like
Ah yes, that one. I was unable to reproduce the exact behavior and difference that Kaldaien spotted on my end when I looked into it, as G-Sync never showcased that massive of a render latency as Kaldaien displayed in those images.
None the less, those posts also have nothing to do with a comparison with conventional V-Sync enabled. They’re comparing the following scenarios:
- G-Sync Off + Waitable On
- G-Sync On
- G-Sync On + V-Sync Off
So basically no, that’s not at all what Kaldaien was saying.
You quoted a message where Kaldaien says the same thing that the rest of us have said: G-Sync/FreeSync has nothing to do with how “smooth” the experience of a game can be experienced – which is exactly because VRR does nothing for overall frame pacing.
You are missing this part, which is basically him stating he rather use SK than G-sync.
Oh and this:
Which still backs up my statement about SK doing better than G-sync or just as good.
Gonna go back to your initial statement here:
Kaldaien is replacing G-sync with his own SK limiter in a sense b/c he gets better use without G-sync and using SK.
No, I’m not missing any part – it’s not relevant to the discussion at hand.
Kaldaien is after an experience without motion blur which to him is the most important of all – to achieve that he needs to use BFI which requires an insanely stable frame pacing and in general isn’t really compatible with G-Sync to begin with (there’s hacky methods to be able to enable both simultaneously, but it’s better to not rely on those at all).
Yes, because VRR does nothing for the overall “smoothness” of an image? What part of that do you not understand?
Stop trying to attribute “smoothness” of an image to G-Sync/FreeSync when that technology have almost nothing to do with that to begin with.
It was you that started to compare “smoothness” of an image using a FPS limiter vs. “using G-Sync/FreeSync” (that is, no FPS limiter at all). It’s an invalid comparison and one that multiple people have told you by now. You have a weird misconception where G-Sync/FreeSync would somehow do something with the overall “smoothness” of an image – even going so far as to completely ignore Kaldaien’s statement of the very opposite.
An FPS limiter is key to ensure a good overall smoothness, as this is the only way to improve frame pacing to an overall target. G-Sync/FreeSync has nothing to do with that at all.
Stop trying to argue that it does.
As far as I am concerned. The way G-sync / Freesync is marketed. IT states its smoother with less latency right? Or am I wrong? B/c I keep reading on all the statements about G-sync marketing and it all states its smoother experience with no tearing.
WE can now agree that it is not the case. I know when I first got my Freesync Monitor I was under the impression my games would run smoother and such at low and high fps. Considering on what I was reading. Obviously I was wrong and didn’t understand the whole gimmick fully. It was until very recently that I started learning that G-sync / Freesync doesn’t in fact make games smoother and such.
Here is exactly what I am talking about:
"NVIDIA G-SYNC is groundbreaking new display technology that delivers the smoothest and fastest gaming experience ever. G-SYNC’s revolutionary performance is achieved by synchronizing display refresh rates to the GPU in your GeForce GTX-powered PC, eliminating screen tearing and minimizing display stutter and input lag. The result: scenes appear instantly, objects look sharper, and gameplay is super smooth, giving you a stunning visual experience and a serious competitive edge."
EDit:
Can you see why I thought such? Right smack on the page for G-sync that is exact wording. So for past couple weeks I am just learning that this isn’t the case. SO excuse me for getting played so hard.
Edit2:
Do have a question though. Is Input Lag same as Latency in terms of hardware? Or is there a clear difference. B/c I am just starting to realize that is something I maybe getting confused on.
Edit3:
On side note framepacing to me directly correlates with Smoothness of game. Tighter the framepacing is the smoother the game is or am I wrong?
G-Sync can only be said to be “smoother” in specific use cases – those scenarios where the FPS is variable enough that where introducing V-Sync would result in an even more stuttery experience than not due to its nature of capping to the refresh rate.
For example, if you were playing a game that ran around ~45 to ~60+ FPS on a 120 Hz monitor then you might feel that G-Sync is “smoother” compared to conventional V-Sync but that’s solely because of how V-Sync works.
Because V-Sync caps the frame rate to a factor of the refresh rate, it would either cap it to 120 FPS, 60 FPS, 30 FPS, or 15 FPS. What that means is that an uncapped frame rate that fluctuated between 45~60 FPS you would end up with a scenario where V-Sync occasionally allows the game to output at 60 FPS (16.67ms per frame), or occasionally caps it to 30 FPS (33.33ms). Basically a massively uneven experience in terms of frame pacing.
That is the single scenario where G-Sync/FreeSync would be experienced as “smoother” – due to it syncing the refresh rate of the monitor to the frame rate of the game the massive fluctuations between the outputted frames (16.67ms → 33.33ms) would be decreased and instead you’d end up with a frame variation much lower between individual frames.
But since G-Sync/FreeSync does nothing in terms of overall frame pacing (only some way of cohesively capping the frame rate does) you’d still experience the end result as “stuttery” if you were to compared it to actually capping the FPS as well.
If we take the 45~60 FPS scenario:
-
V-Sync – Would cap the FPS to either 30 FPS (33.33ms) or 60 FPS (16.67ms). Wildly fluctuating frame pacing because of that.
-
G-Sync – Would sync the refresh rate of the monitor to the frame rate of the game and remove the additional frame variation difference introduced by V-Sync, so basically the end result would be tearing-free variable frame pacing between 45 FPS (22ms) and 60 FPS (16.67ms).
-
Neither G-Sync nor V-Sync – The end result would be tearing and variable frame pacing between 45 FPS (22ms) and 60 FPS (16.67ms).
-
Nether G-Sync nor V-Sync but with a FPS limiter capped at 45 FPS – Tearing but otherwise perfect frame pacing at 45 FPS (22ms).
-
G-Sync + FPS limiter capped at 45 FPS – Tearing-free perfect frame pacing at 45 FPS (22ms).
As you can see, G-Sync alone does not do anything of the overall frame pacing unless one is comparing it directly to a V-Sync scenario where the use of V-Sync frequently changes between 60 FPS and 30 FPS – but so would simply disabling V-Sync do as well.
1 Like
Thank you for the more detailed explaination. So it really only seems smoother in very specific areas of fps then. WHich would explain why Freesync usually is 100 to 144 or 48 to 144 if using Freesync2. At least that’s my monitors VRR rate in CRU info. If it dips below 48 or above 144 it then goes very bad then?
FreeSync is, sadly, a bit annoying in its limitations. What you speak of here is the VRR range – the variable refresh rate range. This is the range of where the monitor can sync up the refresh rate to the frame rate of a game.
-
So a range of 48 to 144 means that if your FPS is between 48 and 144, VRR can kick in and sync up the refresh rate of the monitor to the frame rate of the game.
-
If you were to fall below this minimum FPS you’d start to (typically speaking) experience weird issues such as flashing, hitching, or whatever.
AMD introduced “frame doubling” in their drivers a few years back to counter this issue, but it isn’t a perfect solution due to, again, the nature of FreeSync and how it doesn’t require any form of certification. “Frame doubling” basically means the driver automatically outputs the same image multiple times in a row to achieve a refresh rate within the compatible VRR range.
- So if you were to fall below the VRR range, down to e.g. 47 FPS, each frame would be displayed twice so in reality the refresh rate of the monitor would be 94 Hz (47*2).
- That behavior would continue down to 24 FPS (48/2), and past that each frame would be displayed thrice, so if a game was running at 23 FPS each frame would be displayed thrice so that the effective refresh rate would become 69 FPS (23*3).
- And so on and so far.
This is a feature G-Sync have had since its inception, and all G-Sync monitors (except for “G-Sync Compatible” aka FreeSync) have basically supported a VRR range from 0 FPS to 144 FPS. As Nvidia also requires certification for G-Sync monitors this meant they could ensure the best possible experience by demanding quality from monitor manufacturers.
But this was never the case with FreeSync. They added frame doubling a few years later down the line to prevent flashing etc below the minimum VRR range, but since they don’t have any certification process the quality on FreeSync monitors is highly variable, and especially on how they react at or around the minimum VRR range of that monitor.
AMD began having some form of quality assurance first for FreeSync 2, I believe, which actually have a certification process. However I am not well versed in the requirements regarding passing that certification process.
- On a similar note, the best FreeSync monitors an AMD user can get nowadays is possibly those listed in Nvidia’s “G-Sync Compatible” list. That list is one of FreeSync monitors which fulfills Nvidia’s requirements (on Nvidia hardware) to deliver a somewhat comparable experience as G-Sync at various frame rates etc. One of the things Nvidia’s certification process looks at is specifically flashing, blinking, etc, that can occur at the lower end of the VRR range.
1 Like