GSYNC Breaks In-Game VSYNC

Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by PhantomGamers, May 22, 2017.

  1. PhantomGamers

    PhantomGamers Guest

    Messages:
    79
    Likes Received:
    2
    GPU:
    NVIDIA GTX 1080 EVGA FTW
    Anyone else experience this?

    I just got my first GSYNC monitor, the AOC AG352UCG, and whenever GSYNC is enabled (happens with both fullscreen only and windowed+fullscreen) VSYNC seems to break in every game I've tried (CSGO/Skyrim/Elite Dangerous/+more), and my FPS goes way past my refresh rate of 100hz.

    Forcing VSYNC to ON in the NVIDIA Control Panel works though.

    I'm running a GTX 1080 as my GPU, and I've just clean installed the latest drivers (382.19) using DDU to remove my older drivers which were the current latest non-hotfix drivers, and it's still happening.
     
  2. archie123

    archie123 Guest

    Messages:
    2,475
    Likes Received:
    17
    GPU:
    Aorus 1080ti Extreme
    Yes i have this issue in some older games there doesnt seem to be anything you can do about it , I dont use vsync anymore with gsync. You are better off leaving vsync off and capping your framerate either ingame or with Riva Tuner that way you wont get any vsync induced mouse lag either.
     
  3. Mr_ALLroy

    Mr_ALLroy Ancient Guru

    Messages:
    2,013
    Likes Received:
    28
    GPU:
    EVGA RTX 3080 FTW3
    Try using Fast Sync instead of Vsync. It seems smooth for me when I go above my monitor's max refresh rate. Fast Sync seems to cause a bit less latency than regular Vsync. I always disable Vsync globally and set my "Sync" options in the individual game profiles. You can also try to set the max pre-rendered frames to 1.

    Don't know if this helps. I hope so. :)
     
    Last edited: May 22, 2017
  4. PhantomGamers

    PhantomGamers Guest

    Messages:
    79
    Likes Received:
    2
    GPU:
    NVIDIA GTX 1080 EVGA FTW
    Yes, I'm doing this myself but I was just wondering why it was broken in the first place and if it was maybe something I was doing wrong, it works as intended for my friend who has a Titan X Pascal, so I'm not sure if it's a 1080 issue or something else with my setup.
     

  5. RandomDriverDev

    RandomDriverDev Guest

    Messages:
    111
    Likes Received:
    0
    GPU:
    1080 / 8GB and 1060 / 6GB
    Gsync overrides the games vsync configuration.
    Originally gsync would then imply vsync to prevent exceeding the vertical refresh but nvidia has since made vsync optional.
     
  6. RealNC

    RealNC Ancient Guru

    Messages:
    4,894
    Likes Received:
    3,168
    GPU:
    RTX 4070 Ti Super
    It's best to disable in-game vsync and configure it in the control panel instead. In-game vsync can have side effects that are either not beneficial (like triple buffering) or sometimes harmful (like frame pacing that assumes fixed refresh) with g-sync.

    Even better, just make sure you always use g-sync by capping your frame rate to 2FPS below max refresh using an in-game limiter, or if the game doesn't have one, with RTSS. Avoid the NVidia frame limiter (configured through Inspector) as that has input lag that is almost as high as vsync.

    With that setup, you are going to always have minimum input lag, no stutter and no tearing by avoiding vsync or fastsync.

    Full details here:

    http://forums.blurbusters.com/viewtopic.php?f=5&t=3073
     
    Last edited: May 22, 2017
  7. Han2K

    Han2K Master Guru

    Messages:
    306
    Likes Received:
    6
    GPU:
    MSI GTX1080 GX
    I've to add:
    Set V-Sync to Enable in NV Control Panel.
     
  8. Mda400

    Mda400 Maha Guru

    Messages:
    1,087
    Likes Received:
    199
    GPU:
    4070Ti 3GHz/24GHz
    Was inspector tested with Multicore Rendering disabled as RTSS was later on?
    There's a problem with that article as a reference, it keeps getting updated as new information is figured out that was already known and is not ready to be a viable source. We're not even viable sources, but some may choose to take our advice.

    DirectX 9 was never made for rendering parallelism with the CPU. Valve coded it into source starting with HL2: EP1 with the intention of splitting up animation thread, network thread, etc. Its very hacky and the scheduling is not there like with DirectX 11+.

    [​IMG]

    [​IMG]

    Over time you probably figured that out yourself, but truth is that we can go back and forth arguing which method of frame limiting is superior.

    There's too many combinations to take into account for everyone's individual setup whether they use G-Sync or not.

    I'll also throw another hypothetical theory of mine into the mix and say that if you are using Windows 10, borderless fullscreen exhibits less delay than exclusive mode now does. Makes a bit more sense when UWP games like GoW 4 or Forza Horizon 3 don't offer the option. Windows 10 and DirectX 12 take away a lot of driver overhead. That overhead is reintroduced from enabling things like driver controlled V-Sync, frame limiting, anti-aliasing, aniso filtering, etc.

    Everything runs on top of the UWP now. Desktop is now an 'app' as noted by Windows 8, so any program you run on Windows 8 and later interacts on a different platform with the hardware.

    Unfortunately we need frame limiting with V-Sync and variable refresh technologies like G-Sync. But i think its only safe to say that In-game frame limiting is fact for being the best way to limit frames. Else we can't say which third party method is superior and control panel options should only be forced if the game doesn't offer a working option.
     
    Last edited: May 22, 2017
  9. RealNC

    RealNC Ancient Guru

    Messages:
    4,894
    Likes Received:
    3,168
    GPU:
    RTX 4070 Ti Super
    Here's a very easy test that should convince people once and for all that nvidia's limiter adds too much lag:

    http://forums.guru3d.com/showpost.php?p=5433683&postcount=20

    Yes, feel free to use it. But please, for the love of god, don't claim that it's better or lower latency than RTSS.

    Bottom line: any claim made needs to be backed up by data or a reproducible test. Otherwise it's just placebo. Are there tests that show that windowed mode is faster than exclusive? Because otherwise, you're just asking me to take your word for it :-/
     
    Last edited: May 23, 2017
  10. Orbmu2k

    Orbmu2k Member Guru

    Messages:
    116
    Likes Received:
    21
    GPU:
    RTX 4090
    yes limiter v2 adds a lot of input lag, please try this setting with latest profile inspector 2.1.3.6 and latest drivers R381

    [​IMG]

    what i see on a 60Hz GSYNC setup:

    Limiter V2 + FORCE VSYNC ON = lot of input lag (even below refreshrate)
    Limiter V2 + FAST SYNC = less input lag than VSYNC ON
    Limiter V1 (FORCE V2 OFF) = very low input lag
     
    Last edited: May 23, 2017

  11. RealNC

    RealNC Ancient Guru

    Messages:
    4,894
    Likes Received:
    3,168
    GPU:
    RTX 4070 Ti Super
    Thanks for the update, Orbmu2k! Will try as soon as possible when I'm at my PC.

    Looking through your NVAPI changes, it seems the "limiter delay" is actually configurable? Or is it hardcoded to "on" and "off"?

    Also, do you happen to know what NVidia's limiter is actually intended for? (I would suspect it's was mainly intended for low power mode on laptops with no regards to latency until now?)
     
  12. Orbmu2k

    Orbmu2k Member Guru

    Messages:
    116
    Likes Received:
    21
    GPU:
    RTX 4090
    as far as i can detect it only applies to limiter v1 and removes the non accurate FPS setting
     
  13. Netherwind

    Netherwind Ancient Guru

    Messages:
    8,813
    Likes Received:
    2,396
    GPU:
    GB 4090 Gaming OC
    It did override but doesn't anymore, that's why we have the V-sync option (On or Off). For example, if you enable Gsync and then select "let the 3d app decide", that's what will happen.

    That's how I understand it.
     
  14. RealNC

    RealNC Ancient Guru

    Messages:
    4,894
    Likes Received:
    3,168
    GPU:
    RTX 4070 Ti Super
    That's correct. Initially G-Sync was forcing VSync. In a later driver update, they allowed configuration of the behavior for when you reach your max refresh.

    Forcing vsync with g-sync is still the "best" setting though ("best" in quotes because if you don't care about tearing and stutter prevention, then it's obviously not the "best" setting anymore :))
     
  15. RealNC

    RealNC Ancient Guru

    Messages:
    4,894
    Likes Received:
    3,168
    GPU:
    RTX 4070 Ti Super
    I just tested the new Profile Inspector setting using the 23FPS cap method (it's the lowest it allows now; I don't know if it's possible to set a lower cap like 10FPS, which would make input lag testing more accurate due to higher frame times).

    Input lag is reduced to levels where I can't clearly detect a difference between RTSS and driver limiter. However, that's only true when using"Fast Sync". When forcing vsync off, on, or setting "Use the 3D application setting", the v1 driver limiter appears to have double the input lag of RTSS. And the v2 limiter appears to have triple the input lag of RTSS.

    So it seems Fast Sync + v1 Limiter seems to be treated specially by NVidia.

    To summarize:

    (Tested with both G-Sync and without G-Sync with same results.)

    • v2 limiter: ~3 times higher input lag than RTSS (regardless of Fast Sync on/off.) Basically the same input lag as full vsync.
    • v1 Limiter with vsync on or off: ~2 times higher input lag than RTSS.
    • v1 Limiter with Fast Sync: Cannot detect a difference compared to RTSS.

    Looks good! It seems using the driver limiter is now a viable option. I don't know if the extra input lag when not using Fast Sync is a driver bug or intended though.
     

  16. cryohellinc

    cryohellinc Ancient Guru

    Messages:
    3,535
    Likes Received:
    2,974
    GPU:
    RX 6750XT/ MAC M1
    Im totally confused now, it used to be Gsync on + Vsync on + cap FPS via limiter v2 to ~2fps less than your monitor refresh rate.

    This has been patched?

    Sorry a bit out of context here.
     
  17. RealNC

    RealNC Ancient Guru

    Messages:
    4,894
    Likes Received:
    3,168
    GPU:
    RTX 4070 Ti Super
    Nope. The least input lag setup was G-Sync On + Vsync on (doesn't increase input lag) + RTSS cap to 2FPS below max refresh. NVidia's limiter (via Profile Inspector) was adding too much input lag (almost as bad as g-sync off vsync on.)

    What changed is that now there's an option in Profile Inspector to force a low latency frame limiter, which only works well when using G-Sync + Fast Sync. It seems to be equally good as RTSS + G-Sync + VSync on. But testing is needed with high frame rate cameras to be 100% sure that the new NVidia limiter mode is really as good as RTSS.

    Btw, all of the above only applies to games that don't have an in-game limiter. If you're playing a game with a built-in limiter (like Overwatch or CS:GO), you should always use that. Neither RTSS nor the nvidia limiter can ever be as low-latency as an in-game limiter. The in-game limiter is always the best choice as it doesn't add any extra input lag.
     
    Last edited: May 23, 2017
  18. cryohellinc

    cryohellinc Ancient Guru

    Messages:
    3,535
    Likes Received:
    2,974
    GPU:
    RX 6750XT/ MAC M1
    Let's test it out.
     
  19. CK the Greek

    CK the Greek Maha Guru

    Messages:
    1,315
    Likes Received:
    37
    GPU:
    RTX 2060S

    IF ingame limiter works as it should and doesn't really make issues with performance (many games had this issue until a patch fixed it..)
     
  20. Mda400

    Mda400 Maha Guru

    Messages:
    1,087
    Likes Received:
    199
    GPU:
    4070Ti 3GHz/24GHz
    I did note 'hypothetical theory' which is basically saying in my opinion. I was interjecting on the point of all the different things that could affect these results.

    Same thing goes for the 'evidence' that RTSS is lower in frame-limit latency.
    That is depended on how the user has their machine set up and if they use driver related settings or application ones.



    And you say you need hard evidence.. If we both make quick claims without having some baseline on how we tested and any media proof of it happening, then everything here is opinion and up to the individual to try. The only fact is this: RTSS uses its own frame rate limiter, NVinspector uses the driver's.
     

Share This Page