Reimplement the frame rate limiter!

[Gammler]

The Remaster is missing a Framerate Limiter.
To have the best stable and most predictable frame times a frame limiter is very important.
The Framerate Limiter is also needed to stay in the VRR Range. Running at V-Sync Threshold would increase input delay.

Explaination:
As with every game, on a 120 Hz panel for example the best frame times and input delay can be achieved when not GPU bound and 3 or 4 frames below V-Sync Threshold for example 116 FPS. On the original version of the game i was able to set the game atleast to 110 FPS. Reflex Low Latency does not help, because this is only relevant when running at uncapped framerates.

[Websteroni]

True, but you should be using RTSS anyway. Most in-game limiters can't beat it's frame pacing. Although, I suppose an in-game option wouldn't hurt for more casual PC gamers who don't want to use RTSS.

[il Boh]

in my experience the best way is to use the limiter trough nvidia driver, i use 116 for 120Hz 4K GSync and 236 for 240Hz h5K GSync monitors, last time i used the cap via rtss i had some micro-stuttering when used with DLSS-FG.

[Aemony]

Originally posted by Gammler:

Reflex Low Latency does not help

That's odd, and indicates something is amiss. If Reflex Low Latency works as it should then it should automatically cap the frame rate below V-Sync and within the VRR range.

[Gammler]

I'm glad that there are people that understand this topic.

Originally posted by TheWebDoctor:

True, but you should be using RTSS anyway. Most in-game limiters can't beat it's frame pacing. Although, I suppose an in-game option wouldn't hurt for more casual PC gamers who don't want to use RTSS.

Originally posted by il Boh:

in my experience the best way is to use the limiter trough nvidia driver, i use 116 for 120Hz 4K GSync and 236 for 240Hz h5K GSync monitors, last time i used the cap via rtss i had some micro-stuttering when used with DLSS-FG.

In-Game Frame Rate Limiters are mostly superior to external limiters. This was confirmed by tests from YouTuber "battle(non)sense". Allthough this might be also game dependend if the frame rate limiters implementation is bad.

I only use the nvidia FPS limiter if a game doesn't support limiting FPS.

Originally posted by Aemony:

Originally posted by Gammler:

Reflex Low Latency does not help

That's odd, and indicates something is amiss. If Reflex Low Latency works as it should then it should automatically cap the frame rate below V-Sync and within the VRR range.

Thats weird. The reason i said Reflex does not help was because the FPS Limiting does not work with multiple monitors attached, i guess the V-Sync is broken. I opened another thread for this.
Indeed NVIDIA Reflex Low Latency limits the FPS to 116 in my case. To my knowledge and from all the games i tested Reflex Low Latency FPS Limiter only engages its FPS Limiter when running GPU bound. I mean that is fine, but isn't that wrong and limited to NVIDIA?

[Aemony]

Originally posted by Gammler:

Originally posted by Aemony:

That's odd, and indicates something is amiss. If Reflex Low Latency works as it should then it should automatically cap the frame rate below V-Sync and within the VRR range.

Thats weird. The reason i said Reflex does not help was because the FPS Limiting does not work with multiple monitors attached, i guess the V-Sync is broken. I opened another thread for this.
Indeed NVIDIA Reflex Low Latency limits the FPS to 116 in my case. To my knowledge and from all the games i tested Reflex Low Latency FPS Limiter only engages its FPS Limiter when running GPU bound. I mean that is fine, but isn't that wrong and limited to NVIDIA?

You're touching upon a few different things here, but basically V-Sync On + Nvidia Reflex Low Latency On should automatically engage a FPS limited below the V-Sync range of your monitor. Based on my test, in a multi-monitor setup, that functionality works as it should in this game for me, and it caps the game properly to 116 FPS on my 120 Hz display.

I'm not exactly sure what you mean by Reflex Low Latency only engages its FPS limiter when running GPU bound though? The point of the basic Low Latency FPS limiter is to set an upper bound when using V-Sync (which you need to use on VRR monitors) to prevent the game from going into the V-Sync range of the display.

It was a few months since I left the Special K project and community (where we have a whole bunch of smart folks that knows how Nvidia Reflex works in and out), but if I remember my time from that project correctly, Nvidia Reflex Low Latency has various modes that can engage in difference scenarios and whatnot. But its most basic of modes is to enable an FPS limiter to keep the frame rate within the VRR range when playing on a VRR display with V-Sync enabled (which VRR requires).

[Gammler]

Originally posted by Aemony:

Originally posted by Gammler:

Thats weird. The reason i said Reflex does not help was because the FPS Limiting does not work with multiple monitors attached, i guess the V-Sync is broken. I opened another thread for this.
Indeed NVIDIA Reflex Low Latency limits the FPS to 116 in my case. To my knowledge and from all the games i tested Reflex Low Latency FPS Limiter only engages its FPS Limiter when running GPU bound. I mean that is fine, but isn't that wrong and limited to NVIDIA?

You're touching upon a few different things here, but basically V-Sync On + Nvidia Reflex Low Latency On should automatically engage a FPS limited below the V-Sync range of your monitor.

You seem to be right. My knowledge was the FPS Limiter of NVIDIA Reflex Low Latency only is relevant if you have your FPS uncapped, so the GPU doesn't run at 100% but at for example 98% utilization. (to prevent being GPU bound which would result in higher unstable frame times). That is new to me.

I just tested this in Counter Strike 2 and it indeed does limit the FPS to 116 FPS. So it seems to limit FPS in two cases.

V-Sync enabled (120Hz) = keep it 3-4 frames below V-Sync Threshold
V-Sync disabled (uncapped FPS) => keep utilization just a tiny bit below 100% - keeping the render queue empty at all times

But still an In-Game FPS Limiter should be always available. Not everyone can use Reflex Low Latency.

[Chump]

Originally posted by Websteroni:

True, but you should be using RTSS anyway. Most in-game limiters can't beat it's frame pacing. Although, I suppose an in-game option wouldn't hurt for more casual PC gamers who don't want to use RTSS.

In game fps limiters are always better than RTSS. Mostly because they can lower input latency even further

[Aemony]

Originally posted by Chump:

Originally posted by Websteroni:

True, but you should be using RTSS anyway. Most in-game limiters can't beat it's frame pacing. Although, I suppose an in-game option wouldn't hurt for more casual PC gamers who don't want to use RTSS.

In game fps limiters are always better than RTSS. Mostly because they can lower input latency even further

This is really not the case as it's entirely dependent on how the game developers implemented the in-game FPS limiter and how it works, which usually comes down to experience and understanding. And sadly, despite how importance frame pacing (and so frame limiting) is, that area is wildly misunderstood and underestimated by game developers to the point where a lot (possibly even most of them) of FPS limiters implemented by game developers fail utterly at their intended purpose.

Take the Crash Bandicoot N. Sane Trilogy and the NieR: Replicant (on release) games as two examples. Both of these stupidly implements an """ FPS Limiter """ through an implementation using dynamically adjusting v-sync. Basically v-sync would engage and disengage constantly in an attempt to hit your target FPS.

At best, you'd only get weird occasional stutters while suffering the input latency penalty of v-sync (oh boy, how fun!). At worst, you'd get constant stutters as a result of having a non-standard refresh rate (e.g. 144 Hz, or 165 Hz) while the game devs could never imagine a non-60 Hz multiplier being used as the refresh rate of a display.

So no, in-game FPS limiters are not always better. Sometimes its better, though if I've seen anything in all of my years of working as part of the the Special K modding team as well as the PCGamingWiki project, those are the exceptions and not the rule. Normally, the in-game FPS limiter tend to such compared to third-party solutions developed and implemented by engineers that fully understand the render and presentation pipeline of games to a degree that the individual game developer rarely does.

[LilDipper]

You should never, EVER, use a frame limiter inside of a video game, but rather set the limit inside of NCP or w/e is the equivalent for AMD, CCC?

[LilDipper]

Originally posted by Gammler:

In-Game Frame Rate Limiters are mostly superior to external limiters. This was confirmed by tests from YouTuber "battle(non)sense". Allthough this might be also game dependend if the frame rate limiters implementation is bad.

Please no one listen to this guy, he hasn't a clue. Driver-based limiters will always, ALWAYS be better than the ones inside of games. That is why you should always disable vsync in game and enable it in NCP or CCC, as driver-based limiting is superior in every single way imaginable.

[Gammler]

Originally posted by LilDipper:

Originally posted by Gammler:

In-Game Frame Rate Limiters are mostly superior to external limiters. This was confirmed by tests from YouTuber "battle(non)sense". Allthough this might be also game dependend if the frame rate limiters implementation is bad.

Please no one listen to this guy, he hasn't a clue. Driver-based limiters will always, ALWAYS be better than the ones inside of games. That is why you should always disable vsync in game and enable it in NCP or CCC, as driver-based limiting is superior in every single way imaginable.

Show us your test results. Trust me bro is not a valid test by the way.

[LilDipper]

Originally posted by Gammler:

Originally posted by LilDipper:

Please no one listen to this guy, he hasn't a clue. Driver-based limiters will always, ALWAYS be better than the ones inside of games. That is why you should always disable vsync in game and enable it in NCP or CCC, as driver-based limiting is superior in every single way imaginable.

Show us your test results. Trust me bro is not a valid test by the way.

If it works for me and I don't really give half a crap what you think or do with that info, then it for sure is a valid test.

[Crimsongz]

Originally posted by Gammler:

Originally posted by LilDipper:

Please no one listen to this guy, he hasn't a clue. Driver-based limiters will always, ALWAYS be better than the ones inside of games. That is why you should always disable vsync in game and enable it in NCP or CCC, as driver-based limiting is superior in every single way imaginable.

Show us your test results. Trust me bro is not a valid test by the way.

He is right. Both RTSS and NVCP are better than ANY ingame fps limiter. Those from Unreal Engine games are the worst one.

[Majestic]

Originally posted by LilDipper:

You should never, EVER, use a frame limiter inside of a video game, but rather set the limit inside of NCP or w/e is the equivalent for AMD, CCC?

For AMD you should use Radeon Chill in the Adrenaline software. Which is a heck of a lot better than any third party software or what Nvidia has to offer. Especially in a single-player game like this, with cutscenes and dialogue, meaning the framerate can chill out in those sections.

[Gammler]

https://www.youtube.com/watch?v=W66pTe8YM2s

Apparently there is a difference between limiters that provide

- the lowest input lag
OR
- the most stable frametimes.

I thought most stable frametimes also results in the lowest input lag, but that is not the case as you can see in this video with actual "Click to photon" latency test results.

---
For lowest input lag:
In-Game > NVCPL > RTSS
---
For most stable frame times i don't have test results.
---

There can be some games with a bad implementation though.

RTSS was a good option back when NVIDIA did NOT have a FPS Limiter in the driver. The hidden driver FPS Limiter was worse than RTSS. But when they implemented an official limiter this changed and RTSS is now the worst option, atleast for input lag reduction.

If you just look at the frametime graphs you might come to a conclusion that doesn't give you the lowest input lag.