CPU Render Performance is faster using in-game FPS Cap in comparison to RTSS

Discussion in 'Rivatuner Statistics Server (RTSS) Forum' started by mytommy, Sep 11, 2020.

  1. mytommy

    mytommy New Member

    Messages:
    5
    Likes Received:
    0
    GPU:
    1050Ti
    I've been doing benchmark comparisons (in Rainbow Six Siege) using RTSS and in-game fps cap, and there is a big difference in the CPU render performance as shown below. Is this normal? how can I fix this? I also showed my RTSS settings below.

    In-Game fps cap benchmark results:

    [​IMG]

    RTSS cap benchmark results:

    [​IMG]

    My RTSS Settings:

    [​IMG]
     

    Attached Files:

  2. Unwinder

    Unwinder Moderator Staff Member

    Messages:
    15,405
    Likes Received:
    2,674
    You’re limiting framerate by 60, i.e. request CPU render time to be fixed at 16.6ms. That’s exactly what you’re seeng.
     
    BlindBison likes this.
  3. mytommy

    mytommy New Member

    Messages:
    5
    Likes Received:
    0
    GPU:
    1050Ti
    That's true, but im also capping the framerate to the same digits, 60 using the in-game cap settings, and i get better cpu render performance in that case
     
  4. BlindBison

    BlindBison Master Guru

    Messages:
    757
    Likes Received:
    151
    GPU:
    RTX 2080 Super
    @mytommy I could be missing something, but it looks to me like RTSS is simply delivering more accurate frametimes (60 FPS = 16.7 ms).

    In-engine FPS limiters "can" have a bit better input latency going off of BattleNonSense' tests (Overwatch/BFV for example) -- perhaps Rainbow Six: Siege is also in that camp (not all games are such as Far Cry 5 if I remember Hardware Unboxed's test correctly).

    Typically RTSS does deliver very accurate frametimes -- in a lot of cases more accurate than the internal/in-engine ones from what I’ve read previously.

    Occasionally I have run into some games that don't seem to "play nice" with external fps limiting (MCCCollecion -> Reach for example -- microstutters unless I cap the fps to 60 in-game despite a flat RTSS frametime graph), but from what I gather that's more to do with those applications/specific games than RTSS.
     
    Last edited: Sep 12, 2020

  5. mytommy

    mytommy New Member

    Messages:
    5
    Likes Received:
    0
    GPU:
    1050Ti
    The graph definitely shows that RTSS delivers consistent framerates. However, I still don't see how 16.7ms is better than 4.5ms. This shows that when using RTSS, its taking the CPU more time to render compared to in-game cap.
     
    Last edited: Sep 12, 2020
  6. Unwinder

    Unwinder Moderator Staff Member

    Messages:
    15,405
    Likes Received:
    2,674
    That’s not CPU “performance”. Avoid
    limiting framerate if you don’t understand the basics AT ALL.
     
    Dan Longman likes this.
  7. mytommy

    mytommy New Member

    Messages:
    5
    Likes Received:
    0
    GPU:
    1050Ti
    where did i say this was the CPU performance? the graph says its cpu render performance. And my question to this thread is still not answered. I wanna know if 4.5ms is better than 16.7ms? i dont get any stutter when using rtss or in-game cap, so im not worried about fps smoothness here.
     
  8. Andy_K

    Andy_K Master Guru

    Messages:
    618
    Likes Received:
    111
    GPU:
    MSI GTX 960 OC
    The ingame analytics can't distinguish between cpu takes 16.7ms to render and RTSS is holding back cpu render until 16.7ms is reached, while doing nothing but waiting.

    So the cpu render is not taking 16.7ms when capping with RTSS, but RTSS is waiting to release the cpu render until 16.7ms.
     
    GSDragoon, Dan Longman and BlindBison like this.
  9. mytommy

    mytommy New Member

    Messages:
    5
    Likes Received:
    0
    GPU:
    1050Ti
    So which is better to use for this case, the in-game cap or RTSS? I'm confused. According to your explanation that means the game takes about 4.5ms to render when using the in game-cap
     
  10. BlindBison

    BlindBison Master Guru

    Messages:
    757
    Likes Received:
    151
    GPU:
    RTX 2080 Super
    @mytommy if frame time accuracy/frame pacing is your priority than RTSS is likely preferred. If input lag is your priority in-game is likely preferred.

    Again though not all in-engine limiters improve latency over RTSS (as per Hardware Unboxed tests with Far Cry 5 iirc) — I don’t actually know in this case since I haven’t seen any testing for the Rainbow Six Siege limiter. Also, as I said above occasionally you may run into a game that doesn't play nice with external fps limiting so just be aware of that.

    @Andy_K Yeah, I think how you explained it there is probably correct. Thanks,
     
    Last edited: Sep 12, 2020
    Andy_K likes this.

  11. Andy_K

    Andy_K Master Guru

    Messages:
    618
    Likes Received:
    111
    GPU:
    MSI GTX 960 OC
    And it takes 4.5ms when capping with RTSS. There is no difference in cpu load.

    The difference is when capping ingame the game waits until 16.7ms (more or less) or capping done by then RTSS waits until 16.7ms (more precise).
    Like @BlindBison said, if frame pacing is your first priority (sensitive to microstutter) then go with RTSS
    Which is better in regards to input lag you have to try and test it, as it is dependent to the game. I would prefer RTSS unless I get noticeable input lag.
     
    Last edited: Sep 13, 2020
    BlindBison likes this.

Share This Page