Do you use Nvidia's Driver Limiter or RTSS to limit your framerate? Any pros and cons to either?

Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by BlindBison, Sep 2, 2022.

  1. BlindBison

    BlindBison Ancient Guru

    Messages:
    1,628
    Likes Received:
    607
    GPU:
    RTX 3070
    I've seen some videos on YT like BattleNonSense that seem to suggest the limiters function roughly in the same way as each other and have a similar frametime accuracy and input delay (though going off MSI afterburner or RTSS's readings it seems to be a hair more accurate in terms of frametime).

    But I've also noticed some games don't seem to like having their framerate limited externally and even though the frametime readings look "perfect" on a graph there's clearly camera stutter (or something like that) going on which makes me wonder if perhaps some games need a proper in-engine limiter to work right or some such, I'm not sure (I use G-Sync + V-Sync ON).

    Anyhow I've tested a few games with each limiter and they do seem to work similarly in my tests, is RTSS preferred then because you can set it on the fly/in real time during gameplay? One thing I've always wondered about is whether or not limiting framerate externally has some degree of overhead cost on the CPU since it's a CPU side limiter, but it seems if there is such a cost it's likely very minimal.

    Something I would like to get around to testing at some point would be Nvidia's old limiter from Inspector as maybe that would be "smoother"/play nicer with certain titles, I can't say without testing it further. Anyhow, which limiter do you prefer and why? One potential plus for RTSS is that Unwinder may keep it more up to date than Nvidia keeps theirs as Nvidia may just have added their limiter ages ago then not touched it since, I'm uncertain. Thanks!
     
  2. tsunami231

    tsunami231 Ancient Guru

    Messages:
    13,280
    Likes Received:
    1,278
    GPU:
    EVGA 1070Ti Black
    I dont use it less game is "windowed" and said game dont have vsync options, in which case i will limit to 60 fps, i general just let Vsync limit to 60fps, i will use either or fps limiter if there is pacing issues, which usual fixed by either or providing the game engine isnt the cause

    one istance i use it in hellblade, there alots frametime spikes in that game, if i limit the game to 60fps either nvidia or rtss the frametime spikes are gone, leaving the only stutter i see in that game from shader? loading? when entering a new area
     
    BlindBison likes this.
  3. cucaulay malkin

    cucaulay malkin Ancient Guru

    Messages:
    6,672
    Likes Received:
    3,818
    GPU:
    RTX 3060 Ti
    rtss is just faster if you use afterburner.
     
    BlindBison likes this.
  4. BlindBison

    BlindBison Ancient Guru

    Messages:
    1,628
    Likes Received:
    607
    GPU:
    RTX 3070
    Sorry, but what do you mean by "faster" -- more convenient? Faster to enable and switch values? Less input delay? Thanks,
     

  5. tsunami231

    tsunami231 Ancient Guru

    Messages:
    13,280
    Likes Received:
    1,278
    GPU:
    EVGA 1070Ti Black
    Faster in that you just turn on and off in AB/RTSS change rate without messing with nvcp which general mean the game have be relaunched, in that reguard it would be faster if game it self has support for it, but all fps limiters are not equal some bad some are ok some good. it really depends

    I dont like using external programs for such things so less i have good reason i just use what in nvcp and in game
     
    BlindBison likes this.
  6. yobooh

    yobooh Master Guru

    Messages:
    260
    Likes Received:
    15
    GPU:
    Gigabyte 970 G1
    In my opinion In Game Frame limiter is the best choice when correctly implemented, but sadly often it's not stable enough.
    RTSS and NVCPL have essentially the same input lag and overhead, I prefer to use RTSS simply cause it`s on the fly and has more option (Like the Scanline Sync).
    NVCP fps limit gives less problem with recording softwares so it should be better in this case

    Best video about the topic:
     
    Last edited: Sep 2, 2022
    BlindBison likes this.
  7. windrunnerxj

    windrunnerxj Master Guru

    Messages:
    447
    Likes Received:
    99
    GPU:
    MSI Gaming 1060 6G
    I was just wondering about certain features of RTSS but didn't want to create a new topic in the RTSS sub-forum.
    Would it still be beneficial to have RTSS frame limiter set to display refresh rate on a 60hz display without gsync/freesync WHILE having in-game vsync enabled? I feel like there's a bit less of an input lag this way but it might as well be placebo.
    Regarding scanline sync, it's fine to leave the framerate limiter mode on "async", right? And no need to enable fast sync or anything else for Scanline Sync x2?
    For hybrid sync it's best to set the limiter to "front edge sync", correct?
    Oh and does changing limiter mode from sync to async in RTSS settings requires a game restart or it's applied immediately like the rest of RTSS options?
     
  8. tsunami231

    tsunami231 Ancient Guru

    Messages:
    13,280
    Likes Received:
    1,278
    GPU:
    EVGA 1070Ti Black
    as i stated above setting AB RTSS/NVCP limiter to 60 fps or same hz of monitor can and does smooth out Frametime pacing when used in conjuctions with vsync
     
    BlindBison and cucaulay malkin like this.
  9. oneoulker

    oneoulker Active Member

    Messages:
    81
    Likes Received:
    67
    GPU:
    NVIDIA RTX 3070
    This is a personal experience and I don't have evidence to back it up but when I lock to 30 FPS in Spiderman with RTTS, I had intense camera judders and stutters. When I used NVCP framelimiter, they weren't there.

    Maybe in that specific game with VRR, RTSS did not play nicely with it. No idea tbh.
     
    TheDeeGee and BlindBison like this.
  10. aufkrawall2

    aufkrawall2 Ancient Guru

    Messages:
    2,267
    Likes Received:
    751
    GPU:
    3060 TUF
    Made the same experience once in Battlefield 4: Mouse input was choppy with RTSS limiter, driver limiter was fine.
    But there are also some games with broken input like Days Gone or Horizon Zero Dawn that have mouse input jumping with the driver limiter as well, I guess there's nothing to be done about it in these particular cases.
     
    BlindBison likes this.

  11. BlindBison

    BlindBison Ancient Guru

    Messages:
    1,628
    Likes Received:
    607
    GPU:
    RTX 3070
    See the "BlurBusters Low Lag Vsync Guide" via Googling for it -- the answer is that yes, you want to do the following:

    1) V-Sync ON (or Half-Refresh Rate V-Sync via Nvidia Inspector if you're targeting 30 fps for example)
    2) Use Hztester or UFO or whatever the best site is to test your "true" exact refresh rate (e.g. many 60 hz panels are actually 59.94 or 60.02, etc).
    3) Set an RTSS framerate cap to your precise monitor refresh rate out to at least 2 decimals (some recommend doing precise refresh - 0.01 -- if you cap under it can reduce input lag, but cap too low and you'll get stutter as the v-sync backbuffer won't be filled in time occasionally)
    4) You may also want to enable ULLM (control panel low lag mode) to further reduce input delay while using v-sync (especially if you're targeting 30 fps)

    Note, if you want to target 30 fps then the best way is to do the same setup above but get Nvidia inspector and set half refresh rate proper vsync (the control panel defaults only often adaptive which doesn't play nice with capping to "true" reresh/the low lag approach that fills the backbuffer "just in time"). For 30 you'd just take the true refresh (lets assume 60.02 for sake of example) then cap to half that amount in conjunction with half refresh rate vsync.

    Note, if you're on a laptop then many nvidia inspector or control panels features won't work like half refresh rate vsync but full vsync with an accurate enough fps limit to "true" half refresh still can work OK in my experience with only occasional deviations.

    Personally I have not had a good experience with scanline sync as I've found it to be far too unstable in general terms and it requires a substantial amount of GPU overhead at all times -- there is another thread you could probably search for via Google or Guru3d for the "scanline sync tutorial guru3d" and that should pull it up.
     
    Luckz likes this.
  12. BlindBison

    BlindBison Ancient Guru

    Messages:
    1,628
    Likes Received:
    607
    GPU:
    RTX 3070
    I've been testing exactly this scenario and wasn't sure if it was placebo or not -- RTSS reads out as "totally perfect" frametimes yet there's occasional camera "jitter" nonetheless. This still happened with the driver limiter, but 'seemed' to be less severe/less often, but I can't really tell for sure if it's placebo without more objective testing (EDIT: eh, after retesting I think this is just placebo as the same thing basically happens with either one, apologies I did not mean to mislead anyone)

    My tin foil hat crackpot theory which is probably totally wrong is that perhaps nvidia put somewhat "looser" timings on frametime accuracy for some "reason" -- it's still very accurate, but not quite as accurate going off the frametime readings and the frametimes do not indicate camera smoothness oftentimes in my experience/depending on the game. Sometimes an in-engine limiter will read out much crappier frametimes yet will not stutter so hard when the fps dips below the limiter value or may not have the same camera "jitter" on occasion that capping externally can have (I noticed that in Deathloop for example where the only way I could get the darn thing to run smooth was to set the internal fps limiter to a cap of 60 -- otherwise even when reaching an fps limit 100% of the time, it would stutter. Even a 60 fps cap with RTSS externally seemed to have occasional camera stutter but I'm pretty sure that's the fault of the game really, not RTSS -- Digital Foundry talked about this camera issue iirc so I expect other games may also operate similarly on occasion).
     
    Last edited: Sep 3, 2022
  13. tsunami231

    tsunami231 Ancient Guru

    Messages:
    13,280
    Likes Received:
    1,278
    GPU:
    EVGA 1070Ti Black
    camera jittery could come from how game handles camera movements, some game have fix camera rates that cant be changed and a limiters can either help or make worse. at lest in my experience a game could be 60fps but camera is not updated 60 times second. and limiter can either make thing better or worse.

    I cant remeber games that did this, off top of my head but i know there games out there that were like this as i remeber have games running 60fps and where nice and smooth, but camera movements where anything but

    i could be wrong but just my experience with some game over the years
     
    Last edited: Sep 3, 2022
    BlindBison likes this.
  14. Mapson

    Mapson Member Guru

    Messages:
    117
    Likes Received:
    178
    GPU:
    GTX 1060 6GB
    An older analysis may be of interest if only for some insight into the difference between Nvidia framerate limitiers, in game framerate limiters and RTSS.

    Summary from https://www.reddit.com/r/allbenchmarks/comments/fbuk9x/comprehensive_benchmarking_of_nvidias_fps/

    DX11:
    Stability-wise, the RTSS and NV v3 (NV CP) limiters were almost on par and they were significantly better than all the rest of methods too.
    In terms of approximate and expected input lag, the NV v3 (NV CP) was the best method overall.

    DX12::
    Stability-wise, the NV v2 limiter was the best method overall and the in-engine limiter the worst one. RTSS limiter was close to NV v2 one, and significantly better than both the NV v1 & v3 (NV CP) (worse L-Shapes distributions and Lows avg numbers with both NV v1 & v3, stutters).
    In terms of approximate and expected latency, the NV v3 (NV CP) was also the best method overall.

    Vulkan:
    Stability-wise, the RTSS limiter was clearly superior than all the Nvidia's limiters (worse L-Shapes distributions and Lows avg numbers with all the NVIDIA's limiters, stutters), being the RTSS limiter the best method overall and the in-engine limiter the worst one.
    In terms of approximate and expected input lag, there were almost no differences between all the FPS limiters.
    Full analysis along with comments / feedback in the Reddit link, bear in mind that this is an older analysis, has issues in the testing methodology and is for reference only.

    EDIT: there's also an old analysis of SpecialK, NVIDIA and RTSS limiters in a DirectX11 scenario at https://www.reddit.com/r/allbenchma...ecialks_frame_rate_limiter_review_dx11_based/
     
    Last edited: Sep 3, 2022
    Luckz and BlindBison like this.
  15. BlindBison

    BlindBison Ancient Guru

    Messages:
    1,628
    Likes Received:
    607
    GPU:
    RTX 3070
    Yeah, some games work fine at very specific framerates (why I bring up deathloop which has that camera issue), but other games I think it's not actually the frametimes that are the issue but rather something to do with how they're updating the camera position or some such. I'm just speculating of course.
     

  16. BlindBison

    BlindBison Ancient Guru

    Messages:
    1,628
    Likes Received:
    607
    GPU:
    RTX 3070
    Thank you very much, that's helpful -- so, basically you just gotta test "case by case" is what I'm getting out of this lol. That's very useful, much appreciated

    As far as I know RodroG's work here is the most comprehensive that anybody's done at this point. Now, reading over it and going through some of the comments it seems there may be some limitations to the way it was tested, but it's what we've got at this point at least.
     
    Mapson likes this.
  17. Unwinder

    Unwinder Moderator Staff Member

    Messages:
    16,205
    Likes Received:
    3,984
    Why I always see BlindBison in something misinformative I’m forced to comment in my limited weekend time? :)

    Have to disagree. RodroG’s framerate limiters reviews contains quite a lot of wrong misinformative statements and it resulted in several myths blindly quoted by not too knowledgeable users in the net. Comparing the “efficiency” of different limiters with CapFrameX by comparing “present-to-present” timings only is just plain stupid. That’s exact reason why RTSS introduced settings allowing you to switch between “frame-start-to-frame-start” and “present-to-present” frametime measurements approaches and seeing how they reflect different ways of framerate limiting.
     
    BlindBison likes this.
  18. BlindBison

    BlindBison Ancient Guru

    Messages:
    1,628
    Likes Received:
    607
    GPU:
    RTX 3070
    Sorry, I'm not intending to provide incorrect information. I do try to include it in my comments when I'm speculating about something/when i'm likely to be incorrect, a big part of why I make these threads is to hear what other people think on a given issue I don't fully understand. In any case it's very much appreciated when you do comment as you tend to clear things up substantially. That's helpful to know that Rodro's tests had flaws, in that case I guess nobody's really done objective testing that holds up as far as I know, but it "seems like" Nvidia's approach is quite similar to your own just based on what I've read thus far (apologies if I'm mistaken there, I don't mean to create work for you to clean up).

    I'm sorry to take up your time, your comments are always appreciated :) Thanks for the work you do on RTSS
     
    Last edited: Sep 3, 2022
    oneoulker likes this.
  19. aufkrawall2

    aufkrawall2 Ancient Guru

    Messages:
    2,267
    Likes Received:
    751
    GPU:
    3060 TUF
    It was worse in the past, yes (limited at a value that was slightly off and some weird variance). But with recent drivers, I don't think that's true anymore: RTSS frame time graph is super flat with Nvidia driver limiter throughout the entire Strange Brigade Vulkan benchmark.
     
    BlindBison and Mapson like this.
  20. kman

    kman Master Guru

    Messages:
    464
    Likes Received:
    80
    GPU:
    3080 tuf OC edition
    I use the nvidia fps limiter.It's just too convenient.
    The only game I had to change the limiter was Apex because at a certain driver (452.?? can't remember) The nvidia fps limiter wasn't controlling the frametime anymore for Apex and was getting bad frame pacing so I had to switch to RTSS.
    I even alerted Manuel with a video of the issue but the guy couldn't see an issue apparently lol.Couple of ppl noticed the issue too but nothing was ever done anything about it.
     
    BlindBison likes this.

Share This Page