Blur Busters Forums

Hello,

Can some1 summarize the best settings for the following 3 scenarios please ?!

1) CPU bound like CSGO, LOL, Fortnite etc...

2) GPU bound like Warzone

3) Gsync

From my year old understanding, is that low latency mode should be kept at "On" as long as CPU is not at like 99% load, but it should be put to "Ultra" if the game is GPU bound like a Warzone for example.

So for scenario 1) for games like valorant, csgo, league and all the non GPU heavy games default settings for this should be low latency mode left to "On" while scenario 2) for games like Warzone where my GPU is actually at 99% then low latency should be "Ultra"

What about Gsync? I don't think the settings change I can enable gsync + low latency + -3 FPS cap + vsync no issues yeah?

Then what about this new setting called "nvidia reflex"

I quit gaming for a few months so im rusty, from my readings is that it doesn't hurt to turn this on? The only different is that if you have it "On" compared to "On + boost" is that the second option basically puts your GPU in max performance for the gaming session

So is this true? Enabling nvidia reflex for any game regardless of CPU bound or GPU bound doesn't hurt?

Can I stack it with low latency and gsync and vsync and -3 fps cap?

So in summary:

High fps esports games that don't use too much GPU settings should be: gsync on + vsync on + -3 fps cap + low latency "On" + Nvidia reflex "On+Boost"

Graphics games and games that use GPU such as Warzone or single player graphics heavy games settings should be: gsync on + vsync on + -3 fps cap + low latency "Ultra" + Nvidia reflex "On+Boost"

This is okay?!

.

BTRY B 529th FA BN wrote: ↑

20 May 2021, 09:21

Reflex will over-ride the Low Latency mode.

Correct, Reflex is considered a replacement to LLM and will override it, regardless of what LLM is set to when Reflex is engaged.

GFresha wrote: ↑

19 May 2021, 20:17

From my year old understanding, is that low latency mode should be kept at "On" as long as CPU is not at like 99% load, but it should be put to "Ultra" if the game is GPU bound like a Warzone for example.

So for scenario 1) for games like valorant, csgo, league and all the non GPU heavy games default settings for this should be low latency mode left to "On" while scenario 2) for games like Warzone where my GPU is actually at 99% then low latency should be "Ultra"

Actually, the only known behavioral difference between LLM "On" and "Ultra" with G-SYNC specifically, is that the latter sets an auto FPS limit slightly below the refresh rate when used with G-SYNC + V-SYNC in support games (just like Reflex does), and LLM "On" doesn't. Both LLM "On" and "Ultra" set internal MPRF to "1," however.

GFresha wrote: ↑

19 May 2021, 20:17

Enabling nvidia reflex for any game regardless of CPU bound or GPU bound doesn't hurt?

It shouldn't hurt, but it also won't really be doing anything to help in non-GPU-bound scenarios either (beside auto-limiting the FPS slightly below the refresh rate when using it with G-SYNC + V-SYNC to keep G-SYNC in range).

I.E. Reflex is safe to keep enabled in supported games as fallback to reduce the render queue during GPU-bound scenarios, regardless of whether your system is always GPU-bound in the given game. But be aware it will set it's own auto FPS limit when used with G-SYNC + V-SYNC, so if you want to use an in-game or an external limiter instead of it, you'll have to set their values below Reflex's auto limit for them to take precedence.

It shouldn't hurt, but it also won't really be doing anything to help in non-GPU-bound scenarios either (beside auto-limiting the FPS slightly below the refresh rate when using it with G-SYNC + V-SYNC to keep G-SYNC in range).

So would you say in non-GPU bound scenarios like valorant, fortnite, overwatch its best to enable LLM to "On" and turn off reflex? If reflex overrides LLM and since it doesn't do anything in non-GPU bound scenarios its best to turn off reflex for these games so I can use the benefit of LLM?

Vice versa for games like call of duty, its best to have reflex turned on?

In both scenarios I have gsync + vsync on so I either use an ingame limiter if available or use external limiter and set to -3 FPS

Side question, is NVCP frame rate limiter better than riva turner in terms of input lag?

Side question 2, is LLM on "Ultra" the best set up for gsync + vsync scnarios? Is input delay lower with LLM "On" and in game limiter -3 FPS + gsync + vsync OR LLM "Ultra" and no in game limiter (cos "Ultra" automatically reduces FPS rate down to slightly under the refresh rate of your monitor) or is input delay here the same? Basically does in game FPS limiter still the best for gsync + vsync scnarios or does the FPS limiter by NVCP LLM on "Ultra" beats it?

GFresha wrote: ↑

22 May 2021, 16:44

So would you say in non-GPU bound scenarios like valorant, fortnite, overwatch its best to enable LLM to "On" and turn off reflex? If reflex overrides LLM and since it doesn't do anything in non-GPU bound scenarios its best to turn off reflex for these games so I can use the benefit of LLM?

Both LLM and Reflex address the same thing; they reduce the render queue in GPU-bound situations, but Reflex does it at the engine level, and does it better/more, and as such, is considered a replacement to LLM.

LLM "On" or "Ultra" never has an advantage over Reflex in either GPU-bound or non-GPU-bound scenarios. Where Reflex is available, LLM is obsolete.

Where Reflex isn't available, with G-SYNC, if you're going to have LLM enabled, unless you desire the auto FPS limit LLM "Ultra" sometimes provides, it's "best" to use G-SYNC + V-SYNC + LLM "On" + minimum -3 FPS in-game or external limit (typically in-game for the lowest possible frame delay).

Similarly, both LLM and Reflex do nothing for the render queue when it isn't saturated by max GPU usage, so it doesn't matter whether you have LLM or Reflex on in non-GPU-bound situations, unless, again, you're using them as fallback for when your system does become GPU-bound, at which point you'd always want to use Reflex over LLM.

GFresha wrote: ↑

22 May 2021, 16:44

In both scenarios I have gsync + vsync on so I either use an ingame limiter if available or use external limiter and set to -3 FPS

Assuming you have a 144Hz+ refresh rate, and G-SYNC + V-SYNC + Reflex, it will set a lower than -3 limit of its own (G-SYNC + V-SYNC + LLM "Ultra" will also do this in supported non-DX12/Vulkan games), preventing your manually set in-game or external limit from taking effect.

GFresha wrote: ↑

22 May 2021, 16:44

Side question, is NVCP frame rate limiter better than riva turner in terms of input lag?

As far as is known, they're both about the same from an input lag and frametime performance standpoint, so it's basically down to preference and convenience.

GFresha wrote: ↑

22 May 2021, 16:44

Side question 2, is LLM on "Ultra" the best set up for gsync + vsync scnarios? Is input delay lower with LLM "On" and in game limiter -3 FPS + gsync + vsync OR LLM "Ultra" and no in game limiter (cos "Ultra" automatically reduces FPS rate down to slightly under the refresh rate of your monitor) or is input delay here the same? Basically does in game FPS limiter still the best for gsync + vsync scnarios or does the FPS limiter by NVCP LLM on "Ultra" beats it?

LLM "Ultra" uses the NVCP "Max Frame Rate" limiter, which is similar to the RTSS limit, and thus typically incurs 1 frame of additional delay over a good in-game limiter. However, while external limiter such as NVCP "Max Frame Rate" and RTSS can incur slightly more frame delay than an in-game limiter, they typically have steadier frametime performance, so it can be a trade-off.

As for Reflex's auto FPS limiter with G-SYNC + V-SYNC (same behavior as LLM "Ultra"), there has been speculation that it has as low input lag as an in-game limiter, since Reflex is implemented by the devs at the engine-level, but I haven't tested that, and I don't personally know of any other sources that have tested/confirmed it either.

TLDR; Render queue related settings such as LLM and Reflex (which aren't directly related to G-SYNC operation, by the way) only work to reduce input lag/frame delay in GPU-bound scenarios, and Reflex will always do it better than LLM.

Thank you so much for the reply and taking the time man you are a beast. I am just fascinated by all this info, apologies if my questions seem repetitive, I am kind of just writing my thoughts outload tbh like rubber duck debugging

So essentially reflex and LLM both do the same thing in that in GPU bound scenarios they reduce the render queue but in non GPU bound scenarios, which is almost all the shooter titles like Ow,valorant, fortnite, apex legends, they are useless yeah?

Games that don't support reflex but are not D12 or Vulkan then I can use LLM. Now for these types of scenarios, you mentioned that LLM "Ultra" is different than "On" in that it will cap your FPS using NVCP frame limiter and is more optimal in vsync + gsync but aside from that their is no difference between the two? I don't know why but I keep remembering from my research a year ago in non GPU-bound scenarios for esports titles like valorant,csgo,apex legends to keep LLM to "On" and this would be same as limiting queued frames to 1 but in GPU-bound scenarios like Warzone its best to set LLM to "Ultra" and this would be same as setting limiting queued frames to 0 or is their no basis for this? I remember when LLM first came out it was stuttering for games like AC:Origins but I think it was cos that game pushed the CPU to 99% and so when you slap on it LLM it starts to stutter, especially if you leave it on "Ultra."

So my question here is just like reflex, will LLM ever hurt me to turn it on? Regardless of gpu bound or cpu bound games? Or will it hurt me in non GPU bound scenarios to have LLM? For example in games where my CPU is already under stress, I would assume LLM would cause stutters or does LLM "On" not cause stutters even if CPU is at 99% but LLM "Ultra" causes stutters?


So really in summary, in scenarios where my game does not have reflex option but is also not a D12 or Vulkan game, this means I can use LLM. Isn't it just better to put LLM to "On" and set in game cap to -3 and enable gsync + vsync? I can only see my self using LLM "Ultra" if my game does not have an in game limiter.


Side question, does reflex set fps cap -3 or only -2? So 144 hz does it cap to 141 or 142? If its only 142 then its best to use in game or external limiter for vsync + gsync per your gsync article that recommends -3 fps cap yeah?

Side question 2, when I set reflex to reflex + boost does it hurt if I have that on? Or is it over kill cos I already have msi afterburner on max and also have NVCP to "prefer maximum performance"

GFresha wrote: ↑

23 May 2021, 14:00

So essentially reflex and LLM both do the same thing in that in GPU bound scenarios they reduce the render queue but in non GPU bound scenarios, which is almost all the shooter titles like Ow,valorant, fortnite, apex legends, they are useless yeah?

They are essentially useless from an input lag reduction standpoint whenever the GPU usage isn't maxed.

GFresha wrote: ↑

23 May 2021, 14:00

Games that don't support reflex but are not D12 or Vulkan then I can use LLM. Now for these types of scenarios, you mentioned that LLM "Ultra" is different than "On" in that it will cap your FPS using NVCP frame limiter and is more optimal in vsync + gsync but aside from that their is no difference between the two?

Not with G-SYNC.

With G-SYNC off, LLM "Ultra" behaves differently ("just-in-time" frame delivery in fixed refresh rate scenarios; G-SYNC already does this, so it doesn't need it), whereas LLM "On" behaves the same whether you have G-SYNC on or off.

GFresha wrote: ↑

23 May 2021, 14:00

So my question here is just like reflex, will LLM ever hurt me to turn it on? Regardless of gpu bound or cpu bound games? Or will it hurt me in non GPU bound scenarios to have LLM? For example in games where my CPU is already under stress, I would assume LLM would cause stutters or does LLM "On" not cause stutters even if CPU is at 99% but LLM "Ultra" causes stutters?

Potentially in non-G-SYNC scenarios. Otherwise LLM On/Ultra are the same with G-SYNC but for the auto FPS limit of "Ultra."

I only recommend "On" with G-SYNC as it will 1) not cause any harm in non-GPU-bound scenarios, 2) it can reduce the render queue by about 1 frame in supported games during GPU-bound scenarios, and 3) it doesn't set an auto FPS limit, leaving you free to set one yourself with your chosen limiter.

GFresha wrote: ↑

23 May 2021, 14:00

So really in summary, in scenarios where my game does not have reflex option but is also not a D12 or Vulkan game, this means I can use LLM. Isn't it just better to put LLM to "On" and set in game cap to -3 and enable gsync + vsync? I can only see my self using LLM "Ultra" if my game does not have an in game limiter.

Again, with G-SYNC on, the only difference between LLM "On" and "Ultra" is the auto FPS limit. Also, there's no guarantee LLM will work in the given game, and there's no easy way to tell if it will.

Render Queue settings are typically the most situational and one of the least impactive where perceivable input lag reduction is concerned.

GFresha wrote: ↑

23 May 2021, 14:00

Side question, does reflex set fps cap -3 or only -2? So 144 hz does it cap to 141 or 142? If its only 142 then its best to use in game or external limiter for vsync + gsync per your gsync article that recommends -3 fps cap yeah?

At 144Hz, the auto FPS limit of both Reflex and LLM "Ultra" is ~138 FPS. So as I said in a previous reply, if you want to use Reflex, but set your own FPS limit, you'll have to set it under the Reflex limit for it to take effect (under 138 FPS at 144Hz).

GFresha wrote: ↑

23 May 2021, 14:00

Side question 2, when I set reflex to reflex + boost does it hurt if I have that on? Or is it over kill cos I already have msi afterburner on max and also have NVCP to "prefer maximum performance"

Boost = Reflex + Prefer Maximum Performance.

Man you are beast thanks!!

My games are smooth but you intrigued me with setting FPS limiter using riva turner or NVCP for the consistent frametimes = games feel more smooth and consistent at the cost of 1 MS input delay.

Gotta test this see if I like thanks!

GFresha wrote: ↑

23 May 2021, 16:26

at the cost of 1 MS input delay.

1 "frame," not 1ms; at 144Hz, a single frame is worth 6.9ms. In most of my past tests, external limiters such as RTSS have up to 1 frame more delay than many in-game limiters (Apex, Overwatch, CSGO, etc).

Okay just noticed something weird, is my FPS supposed to be capped when using reflex or LLM to "Ultra"?

Just tried a bunch of different combos but my FPS didn't cap? Using 240hz and these tests below were for valorant which is a D11 game.

gsync + vsync + reflex "Off" + LLM "Ultra" + in game limiter "unlimited" = FPS 240

gsync + vsync + reflex "On" + LLM "Off" + in game limiter "unlimited" = FPS 240

gsync + vsync + in game limiter "unlimited" = FPS 240

So I noticed that with vsync on my FPS was limited to 240 so I turned vsync off and ran these

gsync + vsync "Off" + LLM "Off" + reflex "Off" + in game limited "unlimited" = 600+ FPS

gsync + vsync "Off" + LLM "Ultra" + reflex "Off" + in game limited "unlimited" = 600+ FPS

gsync + vsync "Off" + LLM "Off" + reflex "On" + in game limited "unlimited" = 600+ FPS

Sorry for the repetitiveness but in summary, every time my in game limiter was unlimited, my FPS would run maximum regardless of LLM "Ultra" or reflex "On" but I thought that LLM "Ultra" and Reflex supposed to cap my FPS? Like regardless if gsync and vsync were on, both LLM and Reflex wouldn't cap.

The only time my FPS ever stayed at 240 FPS was when vsync was on, gsync, llm and reflex wouldn't cap.

Is this normal? Am I supposed to see an FPS when I get either LLM or Reflex turned on?