The truth about PRE-RENDERING 0?

Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by Tastic, Jul 16, 2012.

  1. RealNC

    RealNC Ancient Guru

    Messages:
    3,111
    Likes Received:
    1,339
    GPU:
    EVGA GTX 980 Ti FTW
    That's not an FPS spike. I thought you meant you're playing at 120FPS and it went higher than 120FPS (that's what a spike is), which of course sounded weird since you can't get more than 120FPS if you did everything correctly.

    Since you're not getting a solid 120FPS, the "low latency vsync" method won't work. This method should only be used if you can maintain the target FPS cap for the majority of the time. You need to constantly hit your cap. If you're not, you're not getting a latency reduction and you get stutter on top of that.

    In your case, I would recommend using CRU to add a 90Hz mode, and then use this method with a 90FPS cap. Again, same thing: find out the true refresh you're getting, and cap to 0.007 lower. However, since you're introducing a new mode, you can just make sure to make that mode be 90.007Hz then then cap to exactly 90FPS. Make the 90.007Hz mode and then test it on vsynctester to make sure that's what you get. It doesn't have to be exactly 90.007. Anything between 90.006 and 90.012 will work just fine.

    In-game limiters aren't accurate enough. Some may be, but usually they're not. RTSS is accurate enough. Since this method requires ~0.007FPS accuracy, you really need RTSS. For g-sync, in-game limiters are better since it's not important to be super-accurate, but for vsync, RTSS is better.

    You can try though. In-game limiters usually only allow whole numbers, so if you change your refresh rate to be anywhere between XX.006Hz to XX.012Hz, then you can set the in-game limiter to XX FPS. But most limiters will produce either slight stutter or add input lag since they can't stay within 0.007FPS accuracy.

    Yes. With 1/2 vsync, your effective refresh rate becomes half your actual refresh. So you need to cap "as if" you were running 60Hz.

    However, it's better to just add another refresh rate. 90Hz is much better if you can maintain 90FPS.

    Prior to getting a g-sync monitor, I had added a couple extra refresh rates just for doing this method: 80.007Hz, 90.007Hz and 100.007Hz. So basically I was trying to "emulate" a very coarse "poor man's g-sync" :p


    Are you using 1/2 vsync in inspector? You need that. Note that setting that option might require a reboot. The driver has a bug where the option doesn't apply unless you reboot.

    You need to check if you're actually using 1/2 vsync. Disable the frame cap (set 0FPS in RTSS). The game should stick to 60FPS. If it climbs higher, then 1/2 vsync is NOT active.

    But again: 90.007Hz + 90FPS RTSS cap will give you a much better experience.
     
    Last edited: Jul 31, 2017
  2. Enclose

    Enclose Member

    Messages:
    26
    Likes Received:
    3
    GPU:
    GTX 980TI
    Then I was wrong, i apologize if i confused u with this.

    It's ON for sure because without this Option i would have massive tearing in every game i would play on the PC. So i just put the 1/2 Refresh Rate option with the Vertical Sync Tear Control to Standard in every Game.

    Before i updated to the newest version of Inspector i used the Frame Limiter V2 (60Fps) constantly, because it reduced the input lag even more.
    This option seems to be gone or not working as it should anymore (correct me if im wrong), now im trying something new. The Gameplay with the Line Limit=59984 feels really good and very smooth, im just speaking about the Stutter when the Fps reaches 58-59. Im trying to find a solution how to avoid the stutter to be so agressive in the Gameplay. The timeframe the stutter happens seems to me very equal (as you can see from the graph). Maybe i should try a higher pre rendered frames? Im playing with 1 atm. (0=same results)
    Will try the 90.007Hz Method aswell. But i can try this method only in Games i know i can hold the 90FPS, so the 60FPS Method is working better overall in the most Cases for me.
     
    Last edited: Jul 31, 2017
  3. gdallsk

    gdallsk Active Member

    Messages:
    71
    Likes Received:
    2
    GPU:
    MSI RTX 2080 Ventus
    What if you tried Fast Sync instead? That should get rid of all the random stuttering you're getting.
     
  4. RealNC

    RealNC Ancient Guru

    Messages:
    3,111
    Likes Received:
    1,339
    GPU:
    EVGA GTX 980 Ti FTW
    So if you switch your monitor to 60Hz, use full vsync, and don't use a frame capper (so you get 60FPS with vsync) you don't get these stutters?
     

  5. Enclose

    Enclose Member

    Messages:
    26
    Likes Received:
    3
    GPU:
    GTX 980TI
    Don't i need for Fast Sync more Frames do work correctly? I read somewhere that even more frames i can get, then better Fast Sync will work.

    Stutters are still there like in the Graph i posted but not so aggressive like with the other Method 120Hz + 1/2. Im just wondering why the Frametimes between the Stutters are so identical...
     
    Last edited: Jul 31, 2017
  6. gdallsk

    gdallsk Active Member

    Messages:
    71
    Likes Received:
    2
    GPU:
    MSI RTX 2080 Ventus
    The more frames your card outputs, the less input lag there is, as it simply discards the frames that are not going to be displayed, but it still renders them all of them anyway. I have used it on a 60 Hz panel with framerates lower than my refresh rate and it has worked wonderfully. It may seem if your framerate is lower than the refresh rate of your monitor it simply works as if you have Vsync on.
     
    Last edited: Jul 31, 2017
  7. RealNC

    RealNC Ancient Guru

    Messages:
    3,111
    Likes Received:
    1,339
    GPU:
    EVGA GTX 980 Ti FTW
    Fast sync is not useful if you cap your frame rate to almost the refresh rate. And you need to cap in order to reduce latency.

    I don't know. Could be anything. Background processes, game bugs, driver issues, a gazillion other things.
     
  8. mirc00

    mirc00 New Member

    Messages:
    7
    Likes Received:
    0
    GPU:
    12GB
    I just registered here just to say thank you for your low latency vsync guide. Holy ****, its the first time i ever have a smooth running Dota 2 without feeling any input delay. No stutter, no tearing.

    Limited my FPS to 59.990 via RTSS and used max pre rendered frames 1 and triple buffering for Dota 2 (not sure if triple buffering does anything as i read its just for OpenGL and not DX).
     
  9. Xtreme512

    Xtreme512 Master Guru

    Messages:
    665
    Likes Received:
    7
    GPU:
    GTX1060 6GB
    a small note on BF1. when Im using RTSS limiter, cpu usage is higher than using only in-game fps limiter and disabling RTSS limiter.

    So, RTSS limiter at 0 and in-game limiter at 60, plus pre-render 2 gives me best results. in-game vsync is on btw.
     
  10. Mineria

    Mineria Ancient Guru

    Messages:
    3,842
    Likes Received:
    19
    GPU:
    Asus Strix GTX 1080
    In game frame limiters seems to have less lag than any external limiter.
    RTSS is the next best option if a game offers no support for frame rate limiting.
    This goes especially for G-Sync, where I found that having V-Sync set to on (via NVCPL) and in game frame limiter set slightly below max yields the best result.
     
    Last edited: Aug 8, 2017

  11. mirc00

    mirc00 New Member

    Messages:
    7
    Likes Received:
    0
    GPU:
    12GB
    Is there a difference between capping at [Refresh Rate]-0.007 and capping at straight 58 or 59 FPS on a 60hz monitor? I can guess capping at -0.007 gives you the most FPS while still being slightly under the refresh rate vs just being 2 FPS away from the refresh rate basically wasting ~1.9 FPS?
     
  12. RealNC

    RealNC Ancient Guru

    Messages:
    3,111
    Likes Received:
    1,339
    GPU:
    EVGA GTX 980 Ti FTW
    The issue is stutter. Cap to 59 or 58 and you'll see there's frame skipping.

    If you don't mind the stutter, then it's fine. You can cap to whatever you want. The -0.007FPS cap is for when you don't want stutter.
     
  13. RealNC

    RealNC Ancient Guru

    Messages:
    3,111
    Likes Received:
    1,339
    GPU:
    EVGA GTX 980 Ti FTW
     
  14. mirc00

    mirc00 New Member

    Messages:
    7
    Likes Received:
    0
    GPU:
    12GB
    Alright thanks. Why is it -0.007 tho? Why not 0.0014? Can you explain that from a technical point of view?
     
  15. yasamoka

    yasamoka Ancient Guru

    Messages:
    4,730
    Likes Received:
    178
    GPU:
    EVGA GTX 1080Ti SC
    Accurate framerate limiters have to use a wait loop in order to ensure accurate timing. This is reflected in CPU usage which Windows then reports as 100%. Note that CPU usage is active CPU time, not CPU usage per unit of time. That is, if you're using 1% of your CPU all the time, Windows will report 100% CPU usage. But if you're using 100% of your CPU 1% of the time, Windows will report 1% CPU usage. This also explains why certain software, like Prime95, is much more stressful than other rendering applications, even though all maintain a 100% CPU usage.

    Unwinder explains here:
    http://forums.guru3d.com/showpost.php?p=4443926&postcount=3
     

  16. RealNC

    RealNC Ancient Guru

    Messages:
    3,111
    Likes Received:
    1,339
    GPU:
    EVGA GTX 980 Ti FTW
    The closer you get to the refresh rate, the more likely it is that the cap won't be reached. If the cap isn't reached, you get vsync back-pressure, which increases input lag. Vsync back-pressure is what happens if the game submits a rendered frame to the API (D3D, GL, Vulkan) and then goes ahead to render more frames. If the frame cap is actually being reached, this doesn't happen; the game will be blocked from trying to render more frames. That's what's reducing input lag; preventing the game from submitting frames to be displayed at a pace that is faster than what the monitor can keep up with.

    0.007 is a small enough value to not result in stutter, but still high enough to make sure the cap is reached.

    It's not a magic value, btw. 0.007, 0.008, 0.010, 0.012, all would work fine too. But you get too high, frame skipping will start to become more prominent.
     
  17. mirc00

    mirc00 New Member

    Messages:
    7
    Likes Received:
    0
    GPU:
    12GB
    thanks that helped a lot.
     
  18. Mars73

    Mars73 Member

    Messages:
    39
    Likes Received:
    4
    GPU:
    Gigabyte GTX1080Ti
    I've been reading this RTSS cap thing for a while and until now I've never used it.
    I just set (on most games) vsync on and it's smooth and not too concerned about input lag.
    But was still intrigued by this and went ahead and configured it.
    With the vsynctester I found out that my monitor was 59.934 so I capped some games at 59.927.
    Before this cap thing I got a frametime of 16.6ms, with this cap I also have 16.6ms so I must honestly say I can't notice if it's doing less lag then before.
    It's doing a consistent 16.6ms though.
    Could be my age (44), but is there something where I can see/feel this input lag? (sure when playing a game of course).
    Been testing it with Fallout 4, Ghost Recon Wildlands etc.
     
  19. RealNC

    RealNC Ancient Guru

    Messages:
    3,111
    Likes Received:
    1,339
    GPU:
    EVGA GTX 980 Ti FTW
    Are you playing in windowed borderless mode? This trick usually only helps in exclusive fullscreen mode.

    (For Fallout 4 you need to edit an ini setting to disable in-game vsync; it interferes with the cap otherwise.)

    The latency difference isn't huge, but it's discernible. I have no issues ABX-ing it on 60Hz. The mouse is less "floaty". On 120Hz it's much more difficult to tell though (you get ~17ms less latency on 60Hz, but only ~8ms less on 120Hz.)

    Also, it's only with a mouse where you can feel latency. If you're using a gamepad, you won't be able to tell.
     
  20. mirc00

    mirc00 New Member

    Messages:
    7
    Likes Received:
    0
    GPU:
    12GB
    How do you measure the frametime while gaming?
     

Share This Page