Stuttering - cannot resolve

Discussion in 'Technical & Support' started by benkay, Dec 19, 2018.

  1. d0nd33

    d0nd33 Registered

    Joined:
    Dec 4, 2016
    Messages:
    107
    Likes Received:
    78
    That's strange but I don't own an NVIDIA so I can't verify. Sooner or later, I'll make a test at a friend's place.
     
    Alex72 likes this.
  2. DrivingFast

    DrivingFast Registered

    Joined:
    Mar 1, 2018
    Messages:
    1,638
    Likes Received:
    1,083
    Alex72 and d0nd33 like this.
  3. Alex72

    Alex72 Registered

    Joined:
    Dec 27, 2016
    Messages:
    1,317
    Likes Received:
    992
  4. d0nd33

    d0nd33 Registered

    Joined:
    Dec 4, 2016
    Messages:
    107
    Likes Received:
    78
    Thank you, in fact I did that but I noticed that without synchronization there is a ~15px difference measured with the scrolling bar when I look at the tearing line, between last frame and current frame (in fact we can mesure the refresh time knowing the speed of the bar: 15px/960px/s=15.625s - I used my "eye" estimation of 15px but obviously it is not precise, it is in fact 960px/s*0.01667s=16px, I wasn't that far). This means that if the frame get synchronized, I'd look at the "old" frame, the one that when uncapped is 16px backward or, more precisely, 5~10ms old respect the "current" frame (this depends on the tearing line position, if it's about at 1/3 from the top 1/3*16.67ms= 5.56ms).
    One thing I don't understand: the tearing line is in a fixed position even when closing and restarting the game. I'd guess it would be in a different starting position every time the game boots, instead it looks like it's got a fixed place, like if there is some sort of synchronization in the graphic engine (I'm not talking about the vertical one).
    That guide is good to avoid excessive frame queue when synchronization is on, but I rather control the tearing line while displaying a "recent" frame.

    When shopping for a new monitor, I'll consider 144Hz and low lag. I didn't with the one I currently have because it was so cheap and I'm stingy. I know that being stingy may result in bad experiences...:oops:
     
    Last edited: Jan 15, 2019
  5. d0nd33

    d0nd33 Registered

    Joined:
    Dec 4, 2016
    Messages:
    107
    Likes Received:
    78
    Moreover, if you did measure 0.01s, it isn't too bad at all, some gayming monitors are even higher than that (have a look at TFT Central reviews...)! I guess my TV is about 0.04-0.05s, an estimate that comes from my experience with monitors that are way faster. It's really a noticeable lag when directly compared to others. This is why I try to limit "software" lag to its lowest.
     
    Last edited: Jan 15, 2019
  6. Alex72

    Alex72 Registered

    Joined:
    Dec 27, 2016
    Messages:
    1,317
    Likes Received:
    992
    Maybe that is my issue also. Crap monitors. Surely not the most expensive ones, lol.
     
  7. Highlandwalker

    Highlandwalker Registered

    Joined:
    Nov 11, 2016
    Messages:
    806
    Likes Received:
    2,254
    Have any of you tried the record to memory that I suggested in my post on the 11th January, see reply no-10. It works, I have had no stutter what so ever since I set to record to memory. Try this and then play around with VSYNC.
    VSYNC makes the GPU work harder which can cause stuttering. The reason people have success with playing around with graphic settings is because the amount of information it has to write to the hard drive is less and it varies with track and car mod and this changes the length of the stutter. It seems that when set to record to hard drive it interrupts the actual running of RF2 which causes the stutter and it may be system connected as well, Windows may be running some sort of back ground processes. When set to record to memory it does not interrupt the running of RF2 so no stutter.
    When I was getting stuttering I tried a lot of the suggestions in other posts and none really worked totally, some times it improved and other times made no difference.
    These are my thoughts and findings. I am totally happy the way RF2 runs now.
     
  8. d0nd33

    d0nd33 Registered

    Joined:
    Dec 4, 2016
    Messages:
    107
    Likes Received:
    78
    Thank you. Yes, I did but for me, it does not change anything.
    The stutter I get is not of 1sec and it is noticeable even in the menu if I use the overlay bar.
    [​IMG]
    On the left, "ideal" tearing of 240fps, 60hz refresh rate. On the right, stuttering because of inconsistent frame time. The average frame time of both cases is 4,17ms. The one in the right looks worse and is what I'm experiencing. I think OP is talking about this, too.
    I had this problem years ago, it went away but sometime in the past year it came back. I have an AMD HD7950, the very same happens to a friend of mine with a NVIDIA GTX 970. I don't have any other game that got it this bad. I'm going to get a 980 TI, I'll report back.
    To solve this, I use the frame limiter from RTSS.
     
    Last edited: Jan 20, 2019
  9. Alex72

    Alex72 Registered

    Joined:
    Dec 27, 2016
    Messages:
    1,317
    Likes Received:
    992
    The framelimiter from RTSS included using VSYNC anyway so i really didnt notice any difference. I am using those numbers and i limit it in RTSS, but i also use VSYNC since that was the tips so i still have small input lag. I cant tell difference from doing that and to have "maximmum pre-rendered frames=1" in Nvidia CPL. I cant tell any difference with my eyes anyway.

    Record to memory i shall test. The thing is i have NEVER been able to run games without VSYNC smoothly in the last bunch of years. If games arent vsynced they will tear and stutter. Always. But i shall try.
     
  10. d0nd33

    d0nd33 Registered

    Joined:
    Dec 4, 2016
    Messages:
    107
    Likes Received:
    78
    The guide shows the benefit of frame limiting with synchronization but the limiter can be used independently from it.
     
    Alex72 likes this.
  11. d0nd33

    d0nd33 Registered

    Joined:
    Dec 4, 2016
    Messages:
    107
    Likes Received:
    78
    I've bought an used 980 TI, stuttering has disappeared, even without the frame limiter.
    Before I had an HD7950 and played with lowest settings except for textures at full, anti aliasing level 3 and anisotropic filter at 16x. I kept vram usage under control and it never went over 3GB, lowest FPS was ~70, uncapped average ~100.
    I use Windows 10 LTSC, maybe it has something to do with this.
     
    Last edited: Jan 22, 2019
  12. Alex72

    Alex72 Registered

    Joined:
    Dec 27, 2016
    Messages:
    1,317
    Likes Received:
    992
    @d0nd33 Mate, i tested to turn VSYNC off and it worked pretty nicely except a small tear mostly in the sky. In AMS it was stuttery + tearing but in rF2 it was smooth! Maybe i need to tweak the numbers in RTSS a Little bit and be able to get rid of the tear even?
     
  13. d0nd33

    d0nd33 Registered

    Joined:
    Dec 4, 2016
    Messages:
    107
    Likes Received:
    78
    Without synchronization tearing is bound to happen. You can reduce the number of tearing "lines" by matching "Framerate limit" to your screen refresh rate exactly (ms accuracy), so you'll get only a single tearing line in a fixed position but you can set its vertical starting position with the "Scanline sync" value. Every time a frame will get an higher time to render (some random stutter), the tear line will move. I alt tabbed fast to get it back to its starting position, because it may get at road height and it becomes very distracting.
    If you use synchronization these steps are meaningless, DrivingFast posted the proper guide.
     
    Last edited: Jan 22, 2019
  14. Alex72

    Alex72 Registered

    Joined:
    Dec 27, 2016
    Messages:
    1,317
    Likes Received:
    992
    Yeah and i followed his guide. Tested my monitors and got the numbers. Just thought maybe changing the numbers slightly up or down could remove the tear, but maybe thats just not possible. I also Think i left "max frames ahead" to "1" so maybe i'll test setting that to default "none" and see what the difference is. Maybe its only related to vsync. Not sure.
     
  15. d0nd33

    d0nd33 Registered

    Joined:
    Dec 4, 2016
    Messages:
    107
    Likes Received:
    78
    You can actually "hide" the tear line if you find the correct value of "Scanline sync" and you get the same result as synchronization without queue.
    "Max frames ahead" is the length of the queue that there is when synchronization is active. The default is a queue of 3 frames, if you force it at 1 you should get lower lag.
    If you use synchronization with a frame limiter as the guide says, the queue won't be filled and it will look like there isn't one at all, which means low lag.
    I'm not perfectly sure whether Fast sync uses the queue, it should not, even without a limiter, but temperatures will be higher because the GPU will work constantly without any pause.
     
    Last edited: Jan 22, 2019
    Alex72 likes this.
  16. Lars Otto

    Lars Otto Registered

    Joined:
    Mar 8, 2013
    Messages:
    43
    Likes Received:
    104
    Tested anything in this thread, nothing helps :-(.

    PC: AMD 8350, 16GB Ram Gskill, Gigabyte 1070ti 8GB, Asus Sabertooths Rev.3, Samsung 27 TFT, LG 19 TFT,2x 1000GB HDD, 1 x 128GB SSD,Win10
     
  17. Alex72

    Alex72 Registered

    Joined:
    Dec 27, 2016
    Messages:
    1,317
    Likes Received:
    992
    Even with VSYNC on and limit rendered frames ahead? Without VSYNC i cant get it stutter free either. Well, after using @DrivingFast RivaTuner trick by limiting the FPS video RivaTuner profile and do a monitor test with the websites he linked (https://forum.studio-397.com/index....intaining-perfect-fluidity.61352/#post-963007) with that its much better even though i have a tear. If you havent tested that yet could be worth a try.
     

Share This Page