Gsync + Vsync - reflex limiter bypass

Discussion in 'Rivatuner Statistics Server (RTSS) Forum' started by bt_sam, Apr 1, 2024.

  1. bt_sam

    bt_sam Member

    Messages:
    12
    Likes Received:
    1
    GPU:
    1650
    Hi sorry to bother,

    so i've found now that when using vsync + Gsync + Reflex (in game reflex not rtss) an automatic cap to around is applied so for my 280hz monitor i'm seing 260

    the guys over at blurbusters have for a long time advised to use gsync + vsync and use rtss to cap -3 (i.e. 277 for me) below the refresh rate of the monitor

    i cant seem to work out if i can do this anymore if using gsync + vsync + reflex

    obcs i can set 277 in rtss but it gets set to 260 by reflex

    anyone seen this?
     
  2. FatBoyNL

    FatBoyNL Ancient Guru

    Messages:
    1,890
    Likes Received:
    267
    GPU:
    RTX 4080 Suprim X
    Yes, it's a new standard feature introduced by reflex ;)
     
  3. octoreach

    octoreach Active Member

    Messages:
    77
    Likes Received:
    23
    GPU:
    RTX 4070
    You can use the limits set by Nvidia Reflex. Instead of a fixed minus 3, it uses a formula based on monitor refresh rate for determining the limiter value.
    Resulting values are 116 for 120Hz, 138 for 144Hz etc.
    Usually, if I run with Reflex enabled I disable the RTSS limiter or other in game limiters; further advanced discussion can be found here.
     
    bt_sam likes this.
  4. Mufflore

    Mufflore Ancient Guru

    Messages:
    14,806
    Likes Received:
    2,769
    GPU:
    Aorus 3090 Xtreme
    When the rules are hard set, NVidia should provide this information as a mouseover at least.
    It feels lazy to not tell us what happens or how to use features when its easy to do.
     

  5. RealNC

    RealNC Ancient Guru

    Messages:
    5,334
    Likes Received:
    3,581
    GPU:
    4070 Ti Super
    The numbers by Blur Busters were derived by trial and error latency measurements. Nvidia probably knows more than us. But don't sweat it. 260FPS vs 277FPS isn't actually anything noticeable. It's only a frame time difference of 0.24 milliseconds. It's not gonna make much difference in motion clarity or fluidity.
     
    bt_sam likes this.
  6. aufkrawall2

    aufkrawall2 Ancient Guru

    Messages:
    4,702
    Likes Received:
    1,974
    GPU:
    7800 XT Hellhound
    Isn't really surprising when you see how often your monitor hits max refresh rate, despite the 5% lower limit.
    Though enforcing something is always stupid (but still the lesser evil vs. no low latency limiter at all when you care about input lag). Input lag with in-game limiters with high variance usually still is fine even very close to maximum refresh rate.
     
  7. bt_sam

    bt_sam Member

    Messages:
    12
    Likes Received:
    1
    GPU:
    1650

    Yeah I gotcha. Just a shame as to use rtss to limit I would now have to cap lower than 260
     
  8. octoreach

    octoreach Active Member

    Messages:
    77
    Likes Received:
    23
    GPU:
    RTX 4070
    Then don't use Reflex.
    It brings a higher power consumption because of keeping the GPU core clock constantly high and for some games it just doesn't feel justified for that potential extra drop of smoothness.
    This input latency difference is seemingly very little between Reflex Low Latency and RTSS sub-refresh rate cap with G-Sync.
    Witcher 3, for example, feels very smooth and snappy just with the RTSS cap, and I don't bother enabling for Reflex Low Latency for it, I just can't feel any extra snapiness at 116 FPS / 120 Hz.
     
    Last edited: Apr 1, 2024
  9. aufkrawall2

    aufkrawall2 Ancient Guru

    Messages:
    4,702
    Likes Received:
    1,974
    GPU:
    7800 XT Hellhound
    Input lag difference in Witcher 3 vs. in-game limiter/Reflex and RTSS async limiter is very distinct...
     
  10. bt_sam

    bt_sam Member

    Messages:
    12
    Likes Received:
    1
    GPU:
    1650
    yeah i just tested reflex off and traditional rtss -3 cap (g-sync + v-sync monitor is 280hz so circa 3.8 frametime ) i used the new reflex overlay but ussing rtss limiter set to 277 adds about 6ms vs the reflex auto cap 260 (in fortnite) which reads about 3.8ms for total render

    i believe rtss costs you 1 frame to work it's magic is that right

    i do love rtss so much for rock solid frametimes

    upload_2024-4-1_20-8-38.png upload_2024-4-1_20-9-23.png
     
    aufkrawall2 likes this.

  11. aufkrawall2

    aufkrawall2 Ancient Guru

    Messages:
    4,702
    Likes Received:
    1,974
    GPU:
    7800 XT Hellhound
    Yes, vs. Reflex/in-game limiter. Without Reflex, RTSS async probably also can lower lag when games have more than 1 frame CPU prerender (e.g. Witcher).
     
  12. octoreach

    octoreach Active Member

    Messages:
    77
    Likes Received:
    23
    GPU:
    RTX 4070
    I surely can't feel it. They are equally smooth to my perception. Did you measure it by any chance?

    RTSS limiter doesn't cost you one extra frame of input latency, that is false.

    Some time ago jorimt made a test/measurement with Overwatch and there there was only a 2 ms difference between Reflex and RTSS cap with G-sync at 240 Hz. Now I haven't measured for Witcher 3, because I don't have the equipment to do a proper scientific measurement, but I surely didn't feel it and assumed the difference can't be that much higher.

    Also, not sure how reliable RTSS' aproximation of latency is when using Reflex, or when not using it, if the former is otherwise the most accurate. I recall reading a post about it not being so accurate in a certain scenario but can't remember in which one, maybe I can find the source.
     
  13. aufkrawall2

    aufkrawall2 Ancient Guru

    Messages:
    4,702
    Likes Received:
    1,974
    GPU:
    7800 XT Hellhound
    No, but it's very obvious at 75fps. Though smooth and direct aren't the same thing, "smoothness" of in-game limiter is actually kinda bad in that case.
     
  14. octoreach

    octoreach Active Member

    Messages:
    77
    Likes Received:
    23
    GPU:
    RTX 4070
    Right, I actually meant snappy instead of smooth in all previous mentions, so strictly referring to input lag.

    I think this was the post detailing RTSS' latency approximation and its limitations: https://forums.guru3d.com/threads/msi-ab-rtss-development-news-thread.412822/page-209#post-6167178
     
    bt_sam likes this.
  15. octoreach

    octoreach Active Member

    Messages:
    77
    Likes Received:
    23
    GPU:
    RTX 4070
    More precisely, this part I feel is connected to this discussion and the observations you have shown, of higher latency when using the RTSS framelimiter:
     
    bt_sam likes this.

  16. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,249
    Likes Received:
    7,018
    PresentMon’s latency approximation approach mentioned there has zero relation to Reflex latency markers, which are precisely set by application with native Reflex support.
     
  17. octoreach

    octoreach Active Member

    Messages:
    77
    Likes Received:
    23
    GPU:
    RTX 4070
    With the current available tools in RTSS, can the input latency of a game be reliably measured & compared between having Reflex ON vs Reflex OFF and a RTSS framelimit set? This is what I would like to understand better. And thought the latter of the two might show incorrect approximated latency values. But I have to admit this topic seems quite complicated to me and over my head, so your input on this would be valuable.
     
  18. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,249
    Likes Received:
    7,018
    Yes, if game natively sets simulation marker in both cases. It applies to Reflex latency markers based calculation ONLY, PresentMon still uses approximation in any case.
     
    BlindBison and octoreach like this.

Share This Page