Page 2 of 5 FirstFirst 1 2 3 4 ... LastLast
Results 11 to 20 of 50

Thread: VSync without increased latency

  1. #11
    Basic Member
    Join Date
    May 2015
    Posts
    9
    There is a difference between falling to 45/30 fps because the card is consistently too slow to render at 60fps, and has to wait out double the frame refresh time every single frame (30Hz) or every other frame (45Hz), and a few frame drops per second because the prediction of the rendering time is sometimes too optimistic.

    That's why you are seeing close to 60Hz instead of the mentioned 45/30Hz.

  2. #12
    Basic Member
    Join Date
    Jul 2015
    Posts
    29
    Ok, I did some more testing. Yesterday I was playing a long match and noticed my FPS dropped to about 50, while it normally is always fixed at 60 with VSync on. I tested now watching the replay for the match while zoomed out (which usually lowers FPS). With the option off, FPS is at 60; if I turn it on, it dropped to about 40. Turning it off and on would switch it from 60 to 40 FPS consistently.

    Of course this might mean that my PC is not strong enough, but I assumed that if I could keep the FPS at 60 normally then the option wouldn't affect me negatively. Based on these results, I'd think that there's a slight delay introduced by the scheduling of the tasks to coincide with the monitor refresh, and this delay can build over several frames to become noticeable. With that in mind it would seem, to me, that this is expected behavior and that I'd need a better PC to use this option.

  3. #13

  4. #14
    Basic Member
    Join Date
    Jul 2015
    Posts
    29
    Quote Originally Posted by aveyo View Post
    Please note that VSync is meant to be run in exclusive fullscreen mode, not borderless window.
    I've always used it with borderless window and it has always worked. I have also tested the experimental lag limiter on both settings and the FPS drop is exactly the same in both configurations.
    Last edited by lixoman100; 10-30-2015 at 09:25 AM.

  5. #15
    Basic Member
    Join Date
    Jul 2015
    Posts
    29
    Quote Originally Posted by aveyo View Post
    I don't care what have you used all your life, I've just stated an irrefutable fact that VSync is meant to be run in exclusive fullscreen (if you want the best - unhindered by window managers - performance). 'Cause science. Google can provide further why's.
    And I don't care if you think your statement is irrefutable, because it isn't. I have absolutely zero performance difference in Dota 2 between exclusive fullscreen and borderless window, and vsync works in both exactly the same. Your statement was true a few years ago on older engines and/or with shittier hardward. Test it yourself if you like, it has worked like this on most modern games for a good while now.

  6. #16
    Basic Member aveyo's Avatar
    Join Date
    Aug 2012
    Location
    EU West
    Posts
    2,728
    I think the OP is quite clear on what this function does: limits input lag generated by VSync ON.
    What's not quite clear, is that this has a cost in performance. If your hardware does not offer a little more than what's needed for VSync ON, you will definitely get more/larger drops.

    For testing this, a bindtoggle would be handy:
    bindtoggle del "r_experimental_lag_limiter"

    where del is the key you want.

    JayStelly: can we have mat_vsync added back to cfg files, there is absolutely no reason for it's removal, every other Video Advanced GUI settings have a cfg counterpart.

  7. #17
    Basic Member
    Join Date
    Apr 2013
    Posts
    284
    Quote Originally Posted by aveyo View Post
    If your hardware does not offer a little more than what's needed for VSync ON, you will definitely get more/larger drops.
    I get 70-80 fps with vsync off. 60 constant fps with vsync on. 50-54 fps with this option on. So a little more is not enough. The other user stated that he reaches 100 fps with vsync off and it's still not enough. Maybe you need to have 200 fps for this magic option to work properly. If that's the case why would you want to lose 140 fps?!? As far as I'm concerned this option doesn't work. Whatever input lag it removes it's not worth it as the frame loss ads more. So yeah...

  8. #18
    Basic Member
    Join Date
    Feb 2012
    Posts
    95
    The other user stated that he reaches 100 fps with vsync off and it's still not enough.
    Sounds like the scheduler missed it's target too often (10%+), delaying the rendered frame until the next display scanout.
    Maybe there is a way to add a extra few ms of buffer before the next scanout to mitigate this.
    I don't have a powerful GPU to reach the 100 fps to test this though (edit: with lowered resolution maybe)

    edit: tested @720p. fps with vsync off 100 fps average.

    105
    101
    103
    97
    98
    99
    101
    104
    100
    98
    99
    99
    99
    97
    100
    93
    93
    89
    93
    96
    102
    100
    101
    103
    104
    106
    108
    103
    102
    104
    102
    with experimental v-sync.

    59
    56
    55
    59
    53
    55
    58
    59
    58
    54
    59
    59
    59
    61
    59
    59
    55
    60
    58
    60
    59
    53
    59
    59
    59
    58
    58
    59
    58
    59
    61
    frame times attached (very inconsistent, maybe fraps measurement error?)
    Attached Files Attached Files
    Last edited by ScepticMatt; 10-31-2015 at 03:52 AM.

  9. #19
    Basic Member
    Join Date
    Apr 2013
    Posts
    284
    Quote Originally Posted by ScepticMatt View Post
    Sounds like the scheduler missed it's target too often (10%+), delaying the rendered frame until the next display scanout.
    Maybe there is a way to add a extra few ms of buffer before the next scanout to mitigate this.
    I don't have a powerful GPU to reach the 100 fps to test this though (edit: with lowered resolution maybe)
    I just tested it again. I put everything to minimum or off and reached 200 fps that way. It then stays stable at 60 fps with vsync and this option turned on. So it seems you need to have 100+ fps to make this work. It still feels like normal vsync to me, the lag was really noticeable.

  10. #20
    Basic Member
    Join Date
    Feb 2012
    Posts
    95
    frame time analysis:
    vsync disabled:

    double buffered vsync (fullscreen exclusive):

    Windows aero tripple buffered vsync + rivatuner statistics server frame limiter (borderless windowed mode, in-game v-sync disabled):

    experimental vsync:


    something seems to be not working correctly here.
    Edit: no difference between DX9 and DX11.

    System info attached.
    Attached Files Attached Files
    Last edited by ScepticMatt; 10-31-2015 at 05:37 AM.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •