New frame limiting modes vs Scanline Sync x/2 for 30FPS lock

Discussion in 'Rivatuner Statistics Server (RTSS) Forum' started by Lamarcke, Feb 15, 2021.

  1. Lamarcke

    Lamarcke New Member

    Messages:
    1
    Likes Received:
    0
    GPU:
    AMD RX 580 4GB
    I've been testing the new framerate limiting modes on the RTSS Beta and the Scanline Sync x/2, and i can't reach a conclusion about which mode is best in terms of frametime smoothness.

    Scanline Sync /2 works really well, when the target game accepts it, you get no tearing and no extra inputlag (Normal VSYNC 30FPS input lag is horrible, makes everything unplayable), and the frametimes are smooth. frametime fluctuations are minimal, +-0.02 up or down on frametime graph. However if the framerate drops to something below the 30FPS mark on any given moment (like on a disk acess microstutter), the stutter is rather severe (like a frame of freezing). I don't personally mind the scanline hitching while on high GPU usage, i'm way more sensitive to (micro)stutters.

    While testing the new framerate limit modes, i've noticed some things on my 60hz TV:
    (All tests done using a controller and Vsync OFF)

    - Tearing is obviously present.
    - With async mode, no frametime graph fluctuation, even when (by eye) you can spot the difference, the graph remains untouched.
    - With front edge sync, the frametime graph fluctuates +-0.01ms, much like Scanline Sync fluctuations, however at an even lower margin.
    - With back edge sync, the frametime graph remains untouch much like in async mode, however (by eye) it looks a little more smooth than the async method.


    While you can get the best of both worlds with Scanline Sync, some games really hate it, like The Division 2, which has an infamous bug where you CPU can easily get to 100%, and every enemy encounter gives you some stutters, Scanline Sync makes the game's unplayable while in combat.

    So, for someone who doesn't mind the tearing, which method would be the most advisable for a 30FPS lock on a 60hz TV/Monitor?

    If using VSync off, is it a good idea to use the "true refresh rate" - 0.01(/2 for 30FPS) like the BlurBusters LowLag Vsync guide does, or is a 33.3ms lock sufficient?
     
  2. Unwinder

    Unwinder Moderator Staff Member

    Messages:
    15,309
    Likes Received:
    2,608
    Frametime graph can be solid flat line with each of those modes. That's a question of slightly different frametime term definition and frametime measurement method. Async and back edge sync modes are adjusting framepacing by altering frame start times. Frametime measured as frame start-to-start delta will be flat for those two modes. That's default frametime measurement mode for RTSS (it is selectable via <Frametime calculation point> option option in global properties) and almost any game engine and ingame FPS/frametime counter.
    Front edge sync mode is adjusting framepacing by altering frame present times. If you define frametime as frame present-to-present delta (e.g. select <frame presentation> in <Frametime calculation point>), frametime graph will be flat for this mode. Some thidr party frametime monitoring solutions (e.g. PresentMon and CapFrameX derived on it) can only calculate the frametime this way.
    I'd recomment to read context help for <Frametime calculation point> to get more info.
     
  3. RealNC

    RealNC Ancient Guru

    Messages:
    3,302
    Likes Received:
    1,504
    GPU:
    EVGA GTX 980 Ti FTW
    In that case, using the in-game FPS limiter if there is one, or a 30FPS cap in RTSS if not.

    That's only useful with vsync. The -0.01 thing doesn't have any benefits without vsync.

    With that being said, you can also try scanline sync with vsync or enhanced sync (amd) / fast sync (nvidia). I find input lag is quite consistent that way and not too bad. Obviously it's not going to be as low as vsync OFF, but still, it might be OK for you.
     

Share This Page