@@hialgo1808 No. Nvidia ultra low latency exists because of Radeon Anti-Lag. Nvidia Reflex is beyond either of those, and the precedent is that not hitting max GPU utilization has a significant impact on response times. Not sure if Chris was the first to discover this, but he was likely the one who did the most research to impact Nvidia's implementation of Reflex.
@@Voidward I believe that console devs have known it for a long time. I read some docs about game development of a very popular fighting game from 2010 where they talk about keeping 60 fps but always having 2-4% "overhead", never quite got why overhead is so important before I watched Chris videos. I think that both knowledge from console sport/fighting games and Chris publicity helped create the perfect storm and paved the way for Reflex. Really glad that it did.
@@ndesuuu that doesn't seem like a lag reduction thing, 60fps was max back then. Seems they just wanted to target a frame rate and wanted headroom for potential frame drops so the 60fps remains stable, unless they specifically mentioned latency.
This guy definitely deserves more subs ! He is the hero we all needed when it comes to gaming ! The video quality, data processing and nice readable graphs are on the level of Tech Jesus himself !
I was always convinced that limiting the framerate to get below 100% GPU utilisation is more advantageous than a few extra frames, but had no way to prove it. Thank you for this comprehensive analysis!
Comment: _"Radeon BOOST is NOT meant to reduce system latency! That is what Anti-Lag is for!"_ *FYI:* It was AMD who positioned Radeon Boost as *"Ultimate Latency Reduction Technology"* - more powerful than Radeon Anti-Lag according to their own slides. I thought the snip from the press conference I included 4:42 made that clear, guess it didn't - however this should: twitter.com/BattleNonSense/status/1373021156345847813
I would like to have a look at system latency and what could effect latency in games. Things like spatial sound virtual 7.1 or higher bit audio, full screen optimization, GPU scaling and image sharpening on older Nvidia cards. There are also some people saying that discord voice chat or any background desktop application can increase latency especially on low thread count CPUs (4 threads). Also something like G-sync VS No sync VS scanline sync that comes with RTSS in term of latency on caped stable FPS count. I personally found all these fake info around the web and even if there was any latency difference it would be like no difference, but I want it to be proved. There is also an MSI tool that change system scheduling for certain stuff like setting higher priority for the graphic driver and I am not sure if it really does anything. This could be an interesting topic for u in the upcoming video. I would love to see that. Thank you very much for all precious info we got from u.
Great video as always Can we by any chance get a short summary video of what are the most Important settings you need to make to let a game run optimally? Basically summarizing all of your videos on this topic and allowing us to reference a video that contains everything one might need without it containing that much background information :)
This is a *very* subjective topic- not just to personal goal/preference on the matter, but also the massive difference in how/why you would go about it for any particular game given the massive differences in some engines, etc.
Disable vsync, lower settings in-game for max framerate, but don't let you GPU hit that framerate. Limit it to a stable framerate below that which you are unlikely to hit in complex scenes.
I been watching since very early on before I even built my pc. First monitor I bought was Gsync and religiously been following since. Thank You for the Tests!
Fascinating stuff, and it's completely changed my approach to configuring my frame-rate limits (which I've typically done to reduce system power consumption for games that do not require high frame-rates).
Hey Chris, as always, your content is awesome. Quick question then : We know that we have input latency increase in a G-Sync enable scenario with framerate at the top-notch of the monitor (for example, when your framerate touches 165Hz on a 165Hz monitor with G-Sync enable, you increase momentarily your input lag then). What's the setting to use then? NULL OFF / G-SYNC ON / V-SYNC ON / In game or RTSS FPS limiter ON at 158Hz to prevent G-SYNC ceiling and REFLEX ON in game? I hope my question is not too hard to understand :D ! Cya
@@BattleNonSense Would you be interested in analyzing Quake Champions netcode? I know it is not a mega popular game, but it has a passionate community :)
You still plan on making a video comparing keyboards and input lag? :) I would love to see more Corsair comparisons like you did a bit of in the last video! :)
@Falcon_ haha you again, being stupid. Lmao your pseudo intelligence is hilarious. What you’re trying to say doesn’t even make sense. The travel distance of a key is not “subjective” anyway?????
@Falcon_ Yeah but how in the hell is that "subjective", it's a fact and it doesn't even affect the test.. It's not like it's hard to understand the actuation point of different keyboard switches. Please learn the definition of subjective if you want to carry on using it, it seems you think word definitions are "subjective" too lmfao
2 Questions: Would setting max power to 95% help NOT getting gpu bound? Would adaptive power management help in any case where you do not want to be gpu bound?
Interesting as always. Great to see amd and nvidia taking this seriously (thanks to you no doubt!) For me using a frame rate limiter and having rock solid frame times is the most practical and transferable solution for all games. What I would like is a setting for target frame rate with solid frame times in game that adjusts resolution and settings dynamically to achieve this. Until then I am happy to find the settings that give me the most stable performance.
Thanks Chris for the explanation of the functionality at the beginning of the video. I thought it was a marketing gimmick but it's some actual usable technology. 👍🏾
@@marcoferrari4285 I actually like to look of Overwatch when I use the image sharpening at 20% with Freesync. Especially when I have all graphic settings to low except for the render scale, which I set to 100% because anything lower just looks too ugly for me. I'm using 5700XT and AW2521HFL. But overall, I'm starting to regret buying AMD GPU now because of all the features like DLSS, RTX and now Reflex that I'm missing out on. I don't regret buying the Ryzen 3600 though and I still think the CPU was well worth the money.
I am curious about the frametime consistency when comparing with (RTSS) using the nvidia reflex , can you tell me an avg and how much standard deviation does it have comparing to RTSS? (Let's say you cap the framerate at 90% load on GPU as the maximum limit using RTSS) Is the frametime all over the place with Reflex? Why I ask this: You said the in-game framelimiter has bad frametimes (Inconsistent) so I am trying to get the best of the best in terms of options (G-sync/Free Sync) + No GPU Bound (let's say best of both worlds, no tearing (SYNC OPTIONS) + Smooth (Frametime Consistency) + Lowest latency possible while having the first 2 (Not the fastest) Also if you can test the new feature of RTSS (Scanline SYNC) You have a great mind and I thank you for dedicating your time to do this for us and I hope you will continue to do so.
WTF is Radeon Boost doing? You buy a high refresh rate monitor to get a responsive and as clear as possible gaming experience during fast movement and Radeon Boost doesn't really benefit the responsiveness, but completely ruins clarity. Who thought this was a good idea? Even ignoring any stability issues, I can't see a single reason why anyone would use this.
Great content as always :) Just one question: throughout the video you've always talked about a "good" in-game frame limiter. Would you say the limiter in CSGO through console (fps_max x) is good? Or can you achieve even lower latency with an external limiter like rtss?
I think there's a bit of a misunderstanding here. There are two separate measurements for "good" fps limiters: 1) Good framepacing (how smooth image looks) 2) Good responsiveness (doesn't introduce additional input lag) Both RTSS and Nvidia's limiter v3 offer smooth framepacing (which is GOOD), however since they are external limiters, they introduce additional input lag (which is BAD). In-game limiters usually have worse framepacing (which is BAD), however they usually don't add any additional input lag (which is GOOD). You see where this is getting? Anyway to answer your question, using fps_max to minimize latency is good enough, and using RTSS would actually increase the latency.
In terms of system latency Ubisoft games have horrible ingame FPS limiters (R6S is just one example - requires cfg edit to enable) Overwatch, Source, Unreal, Frostbite Engine have good ingame FPS limiters based on my system latency tests. External FPS limiters also perform worse than in-engine limiters (in terms of system latency) as just like NULL or Anti-Lag they try to pace the game from the outside.
Love the content as always Chris. I've used Afterburner as a framelimiter for some time, any thoughts on using it or an external equivalent outside of the game? Thanks
When you are talking about smoothness, there is thing to remember. While higher framerate may reduce latency, when you are already at your monitors refresh rate going higher framerate can make it less smooth. Having framerate very uneven with your refresh rate causes it to not be smooth. For example, having 150fps on 144hz monitor is smoother than 180fps. And for more noticeable example, having 150fps on 120hz feels like stuttering to me. That is, with no sync, im not sure how it works above your refresh rate on freesync or g-sync monitors, i only know in those smoothest is to limit fps little below your refresh rate. So, if i had game that supports reflex and nvidia gpu, i would enable reflex but limit fps anyway (I assume there is no downside to enabling it). But i dont play any game that supports it and i have amd gpu currently. I dont believe that improvement of radeon anti-lag is that low at least in every case. In Quake champions i keep my fps limited to 150, and its clear to me that i get lower input lag with radeon anti-lag. On my laptop i limit it to 118 fps (for freesync 120hz) and it doesnt keep fps up there always, on it i feel huge difference. Could it be that somehow it affects input lag in Quake Champions differently than in games you test?
With G-Sync if I enable Reflex in Overwatch, the game auto-caps FPS to 138 (on my 144 hz monitor). Is this the Nvidia driver level limiter kicking in like it does for ULLM automatically or is it the Overwatch FPS limiter kicking in? It’s “probably” the driver level limiter I would think given its behaving just like it does for the ULLM mode+Gsync. In your prior videos you demonstrated that the in engine limiter has less latency and we know RTSS and the Driver limiter have basically the same latency. Would this mean for users who are reaching the Max refresh of their Gsync panel and hitting that automatic FPS cap that they’ll actually get worse latency with reflex ON since they’ll be running into the driver limiter? They could instead do Reduce Buffering ON + a manual cap at 140 for example, but we only know how capped + FPS limit compared to Reflex uncapped currently. It would be useful to see how Reflex + Nvidia driver level limiter / the auto FPS cap behavior on Gsync panels compares when you’re actually reaching that target. Currently there’s no way to disable this auto capping behavior so if you use another limiter you have to cap beneath the auto limiter by say 3 or 4 frames which wastes overhead. Thanks,
hey dude, would be awesome to see an updated look at Escape from Tarkov. they've supposedly done a lot of server modernization in the past year and the community can't decide whether it's better or worse. could use some cold hard data from the man himself
You sir are brilliant!!!! I love it. Liked & Subscribed! I need you to help me decide if I should get the Zowie BenQ XL2456k dyac+ 240hz or Alienware AW2521H Ulmb 360hz! No explanation needed! 😕
Very good video. Conclusion: Avoid using such low latency features (unless your gpu load is at fixed 99%), stick to in-game fps limiters (capped to slightly lower framerate that reduces gpu load).
There's a free Quake-like game on Epic, called Diabotical, that has optional multi-threaded input where an input thread runs at ~500hz, as well as giving you control over buffering (1 to 5 frames) settings, and a pretty good in-game FPS limiter. If you have time to test whether multi-threaded input is actually a boon there, and what combination of settings is "best", I would love to see what results you get.
The way that AMD is playing catchup here is interesting in implementation but embarrassing in practice. You get your visuals set to mud in a way AMD thinks you won't notice to not even get the same benefits as Nvidia's Reflex does at full resolution. No matter how much people dump on Nvidia and their business practices, software advances like this, DLSS, RTX, integer scaling, Gsync, hardware acceleration in productivity apps, etc, are what will keep me loyal to this brand. They are a significant value add to me.
@Falcon_ Their version is RTX is also very underwhelming. They're trying to put out pretty much whatever they can to claim they have feature parity, but thus stuff clearly doesn't have the R&D behind it.
One thing that is very interesting to consider is how much latency remains after subtracting frame draw time and the result of the SOL test (which covers scanout time, hardware latency, polling rate, ect) to see how much latency is caused by imperfection of the games input handling (careful not to confuse framecap delay with gpu draw time)
There is actually another very interesting aspect that games which provide the Reflex Flash Indicator allow you to measure. And that is how much extra delay is caused by the game's animation system as well as (in some games) intentional trigger delay.
That AMD tech is interesting, but it creates an inconsistency in response, which seems like the exact opposite of what you would want, even if its more responsive at times.
@@WalrusFPGA Static scene does not need high fps, only high fps we need is in the motion and when we control this motion. I need higher FPS when i moving around and higher sharpness when i just looking
What's the current consensus for limiting FPS in games with no engine limiter? still RTSS? And has the opinion on GSYNC changed at all? Is latency better or worse with it enabled?
So how's the fps limiter in Call of Duty Cold War? It seems to fluctuate all over the place for me rather than holding a constant fps, does it also have bad input lag?
Love your videos, im a new at PC gaming, can you make a video showing the best NVIDIA control panel options for gaming ? Its so many options and kinda confusing.
Honestly, you don't need to touch anything in the NVCPL. Except for when you own a G-Sync monitor there are 2 options that you want to look at 1. G-Sync is enabled 2. always choose highest available refresh rate.
@@BattleNonSense Thx for the reply ! I see a lot of these "guides" for better performance after some tweaks on the NVCPL. I put the 'Power Management mode' in Prefer Maximum Performance and 'Texture Filtering - Quality' at High Performance. Does this settings actually change anything ? How about 'Threaded Optimization' , 'Shader Cache' ? Does Low Latency Mode works and NVIDIA Reflex is on ?
Apex Legends has some issues with it's Lag compensation, try turning it off(cl_lagcompensation 0) and go into the fiering range. you can easily hit consecutive wingman shots while draging the aim over. with Lag compensation on(cl_lagcompensation 1) tracking is the only way of hitting consecutiv shots. in game with the lag compensation off(cl_lagcompensator 0) you will get a ton of no-regs. But the lag compensation favours tracking aim over dragging/flicking. if you flick and rest your aim over the target you will hit, but if the mouse is dragged over target it will count as a miss
I’ve been using NULL with Reflex and my latency feels amazing. Also use a timer resolution but the one i own also is a standby list cleaner which i still don’t know is all the way safe. But on controller you feel the latency even more when it jumps than you do on kbm, atleast i do. I also cap my frames on all GPU heavy games because I don’t have a Gsync monitor (i have one on the way now)
I know this sounds odd but could you look into this for me, War Thunder has both support for DLSS 2.0 and Nvidia Reflex but no site (including Nvidia's) show "no support" for Reflex and any article as of Feb 2021 and newer do not show DLSS support for WT either. just find this highly odd considering WT is a fairly large game as its a direct competitor to WoT and WoWs and also are the same company behind "enlisted" which supports Reflex.
@@BattleNonSense Depends, Chill with 120fps during movement should have lower input lag than an in-engine limit with 60fps Further, not all in-engine limiter are created equal. To illustrate: Your GPU is capable of ca. 70fps in standard resolution. Chill limits to 60fps without movement and reduces the GPU utilization thus reduces input lag. When you move the cursor with Boost your GPU is capable of 150fps but again with ca. 100% utilization. Chill can also limit your fps to 120fps during movement and can so guarantee you never reach 100% GPU utilization.
But where comparison latency between 1600 dpi mouse and 800 dpi? :P is there any difference at 1k hz? or difference is only visible like u show to us on 8khz? because looks like 3200 dpi or 1600 is much faster than 400 dpi if it comes to latency. Like it!
Hi! I've never watched one of your video's before, but are you from germany? I started the video and the german accent insta punched me in the face xD Edit: This is not a hate commentary, just a simple question, because I live in germany
Hmmm... so if I want to use an in-game frame rate limiter in order to both prevent gpu bound, as well as enable input delay consistency and frame consistency, is there any point to be using nvidia reflex since limiting FPS at a consistently sustainable value simulates a cpu bound case? At least Boost surely gives benefit, right? Or does it..? Hmm. I'm guessing not. I suppose that's why you mentioned multiple times that reflex seems to be like a very nice variable frame rate limiter, because it produces similar results to limiting frames at a consistent/reliable number, without having the frames actually limited. Am I on the right track here?
Great video, as always. I am just wondering if there is a difference when you use in-game limiter or the one in nVidia settings. I have i7 6700k and GTX 1060, while playing Apex Legends in 1080p I get around 144fps on low settings, where cpu load is 50%, and gpu load is 99%. I use the limiter from nvidia settings, set at 140 fps, but the game still feels a bit laggy, maybe it's because of 20 tick servers, I don't know. What do you guys think?
the nvidia FPS limiter in the NVCPL as well as AMD's CHILL perform worse than the in-engine frame rate limiter because just like NULL and Anti-Lag they try to pace the game from the outside. Apex Legends is based on the source engine which has a good in-game FPS limiter, but Respawn does not offer an FPS cap option in the settings, and the fps_max parameter in the launch options does not work for me.
Thank you so much , love your channel ❤️ as a conclusion, for me an AMD user i should enable anti-lag and 'try and error' to find a frame rate that my pc can maintain consistently in each game .. am i right?
Sorry if this was covered already, but do I need to use the Overwatch frame rate limiter when I'm using Reflex+Boost or is that redundant? One or the other? Thanks!
A really nice thing about radeon boost: It also works with vega igpus. So basically any amd laptop. This gives me in overwatch about 100fps on my amd apu laptop and 70fps in gta online both on fullhd. This is really awsome, because you can game on laptops, which are not really meant for gaming.
@@buttflies7597 im playing on my lenovo slim 7 with an Ryzen 4800u on full hd with medium settings. But also lower end ryzen cpus can achieve that, because the gpus are realtively similar.
You can just reduce the render scale or the game resolution instead. That way you will get a stable frame rate and frame times while with Boost FPS and Frame Times (as well as system latency) are all over the place.
Thanks a lot, awesome video!! What basically is not clear for me yet is the following: While having Nvidia Reflex Enabled, should the Low Latency Mode (LLM) in the Control Panel (NCP) be off as shown in the benchmarks? Would LLM on or ultra negatively affect the input lag? Or is it just not relevant anymore as soon as Reflex is on and only becomes affective when the game doesn't support Reflex?
Hi Chris ist das Problem "Latenz reduzieren mit AntiLag" mit AMD Grafikkarten immer noch nachteilig? Ist Nvidia immer noch die bessere Wahl mit Reflex low latency in Games? Danke Dir
You didn't mention that boost also does variable rate shading. It's more than just simply adjusting render scale now. Also Apex Legends works fine for me on a 165hz 1440p monitor with a 5700xt so strange it didn't work for you. Did you DDU? Take this as constructive criticism cause I really do appreciate your work on this and your findings. Keep it up.
I feel like Nvidia Reflex exists because of your testing, and you'll never get official credit for it. So I'll just thank you personally.
@Falcon_ nice theory you have come up with there. FoRtNiTe KiDs DuMB. 🤡 look in a mirror
I feel like Nvidia Reflex exists because of Anti-Lag, personally...
@@hialgo1808 No. Nvidia ultra low latency exists because of Radeon Anti-Lag. Nvidia Reflex is beyond either of those, and the precedent is that not hitting max GPU utilization has a significant impact on response times.
Not sure if Chris was the first to discover this, but he was likely the one who did the most research to impact Nvidia's implementation of Reflex.
@@Voidward I believe that console devs have known it for a long time. I read some docs about game development of a very popular fighting game from 2010 where they talk about keeping 60 fps but always having 2-4% "overhead", never quite got why overhead is so important before I watched Chris videos. I think that both knowledge from console sport/fighting games and Chris publicity helped create the perfect storm and paved the way for Reflex. Really glad that it did.
@@ndesuuu that doesn't seem like a lag reduction thing, 60fps was max back then. Seems they just wanted to target a frame rate and wanted headroom for potential frame drops so the 60fps remains stable, unless they specifically mentioned latency.
This guy definitely deserves more subs ! He is the hero we all needed when it comes to gaming ! The video quality, data processing and nice readable graphs are on the level of Tech Jesus himself !
"himself"... heh
Exactly, it's what I say since when I "found him" on TH-cam
He’s beyond tech Jesus because he’s rational, and not sensationalistic. Doesn’t try to editorialize his content to fan the flames of fanboyism
this is youtube.
clowns and clickbait rule here.
I was always convinced that limiting the framerate to get below 100% GPU utilisation is more advantageous than a few extra frames, but had no way to prove it. Thank you for this comprehensive analysis!
Comment: _"Radeon BOOST is NOT meant to reduce system latency! That is what Anti-Lag is for!"_
*FYI:* It was AMD who positioned Radeon Boost as *"Ultimate Latency Reduction Technology"* - more powerful than Radeon Anti-Lag according to their own slides.
I thought the snip from the press conference I included 4:42 made that clear, guess it didn't - however this should: twitter.com/BattleNonSense/status/1373021156345847813
Where did you manage to buy an LDAT? I've been looking to buy one
Nvidia GeForce Twitter just shared your video. Nice.
Underrated channel - you should make a video on testing CRT monitors vs modern flat-panel monitors. That would be very interesting
wth? who uses those?
@@FcoEnriquePerez CRT monitors have 0ms latency
@@Fals3Agent Yeah, so? lol
@@Fals3Agent If i could find good CRTs for gaming I would use them to this day.
@@FcoEnriquePerez Real men use CRT
We can't thank you enough for your hard work.
I would like to have a look at system latency and what could effect latency in games. Things like spatial sound virtual 7.1 or higher bit audio, full screen optimization, GPU scaling and image sharpening on older Nvidia cards.
There are also some people saying that discord voice chat or any background desktop application can increase latency especially on low thread count CPUs (4 threads).
Also something like G-sync VS No sync VS scanline sync that comes with RTSS in term of latency on caped stable FPS count. I personally found all these fake info around the web and even if there was any latency difference it would be like no difference, but I want it to be proved. There is also an MSI tool that change system scheduling for certain stuff like setting higher priority for the graphic driver and I am not sure if it really does anything.
This could be an interesting topic for u in the upcoming video. I would love to see that. Thank you very much for all precious info we got from u.
Great video as always
Can we by any chance get a short summary video of what are the most Important settings you need to make to let a game run optimally? Basically summarizing all of your videos on this topic and allowing us to reference a video that contains everything one might need without it containing that much background information :)
This is a *very* subjective topic- not just to personal goal/preference on the matter, but also the massive difference in how/why you would go about it for any particular game given the massive differences in some engines, etc.
Disable vsync, lower settings in-game for max framerate, but don't let you GPU hit that framerate. Limit it to a stable framerate below that which you are unlikely to hit in complex scenes.
@@Voidward what about llm?
I been watching since very early on before I even built my pc. First monitor I bought was Gsync and religiously been following since. Thank You for the Tests!
Incredibly in-depth as always. Great job and thanks for this useful video!
As always, thanks for the real testing with all this hard work!
Keep it up and thanks for the video!
Your analysis is so organized and clear...Thanks!!!
Thanks for covering this topic. It was interesting to listen to!
Fascinating stuff, and it's completely changed my approach to configuring my frame-rate limits (which I've typically done to reduce system power consumption for games that do not require high frame-rates).
Hey Chris, as always, your content is awesome. Quick question then : We know that we have input latency increase in a G-Sync enable scenario with framerate at the top-notch of the monitor (for example, when your framerate touches 165Hz on a 165Hz monitor with G-Sync enable, you increase momentarily your input lag then). What's the setting to use then? NULL OFF / G-SYNC ON / V-SYNC ON / In game or RTSS FPS limiter ON at 158Hz to prevent G-SYNC ceiling and REFLEX ON in game? I hope my question is not too hard to understand :D ! Cya
You are the hero we need!
I'm with you up to the point where you start wishing more games add proprietary features instead of just improving their rendering pipeline.
Remember when battle)non)sense used to do netcode videos for games ?
Eventually I will do that again, right now system latency interests me more.
@@BattleNonSense Would you be interested in analyzing Quake Champions netcode? I know it is not a mega popular game, but it has a passionate community :)
Well done!
Superb video as always, you are on another planet, so much work, quite spectacular.
You still plan on making a video comparing keyboards and input lag? :) I would love to see more Corsair comparisons like you did a bit of in the last video! :)
@Falcon_ No, it was fine! Go watch his last video! :-)
Yup, something is in the works. Will be just a few more weeks until I get all the hardware. :)
@@BattleNonSense Ohhh great to hear! Can´t wait! :)
@Falcon_ haha you again, being stupid. Lmao your pseudo intelligence is hilarious. What you’re trying to say doesn’t even make sense. The travel distance of a key is not “subjective” anyway?????
@Falcon_ Yeah but how in the hell is that "subjective", it's a fact and it doesn't even affect the test.. It's not like it's hard to understand the actuation point of different keyboard switches. Please learn the definition of subjective if you want to carry on using it, it seems you think word definitions are "subjective" too lmfao
2 Questions: Would setting max power to 95% help NOT getting gpu bound?
Would adaptive power management help in any case where you do not want to be gpu bound?
You are doing god's work. Thank you so much.
man you're the best
no marketing BS
just facts
Interesting as always. Great to see amd and nvidia taking this seriously (thanks to you no doubt!) For me using a frame rate limiter and having rock solid frame times is the most practical and transferable solution for all games. What I would like is a setting for target frame rate with solid frame times in game that adjusts resolution and settings dynamically to achieve this. Until then I am happy to find the settings that give me the most stable performance.
Thanks Chris for the explanation of the functionality at the beginning of the video. I thought it was a marketing gimmick but it's some actual usable technology. 👍🏾
Thanks for uploading.
I will re-watch this in 2023 when the hardware is available.
This tech works with GPU's that are more than 3 years old.
2023 lol send me a message in 2032
My GPU is 6 year old :(
Can't even buy any 4+gb GPU in the current market
At least I can still use the old Boost in DX9 games.
how is your "available hardware in 2023" thingy goin lil bro???
@@user-pdogiP78V you've been living under a rock or smth, LIL BRO?
Broooo you are the must youtuber care about the knowledge at this area thank's u 4 real u make a hard work . Hope see more and more from you
Amazing breakdown, so comprehensive
thx you deserve more subscriber
Thank you for all of your hard work making this useful in depth video for us!
So for AMD Users keep using AntiLag, cap frames in game where your card is more stable and for the moment avoid Radeon Boost.
Thanks for the video!
and also make shure to disable the image sharpening because it's complete garbage
what if a game is capped at 60 fps should I still keep AntiLag on?
@@marcoferrari4285 I actually like to look of Overwatch when I use the image sharpening at 20% with Freesync. Especially when I have all graphic settings to low except for the render scale, which I set to 100% because anything lower just looks too ugly for me. I'm using 5700XT and AW2521HFL. But overall, I'm starting to regret buying AMD GPU now because of all the features like DLSS, RTX and now Reflex that I'm missing out on. I don't regret buying the Ryzen 3600 though and I still think the CPU was well worth the money.
@@heinstar6206 Same gpu and cpu. In overwatch works flawless but in other games doesn’t work very well like Warzone.
i hope you can try compare other competitive games too like DoTA 2 for example
the hero we need but dont deserve, love your work, thank you so much!
I am curious about the frametime consistency when comparing with (RTSS) using the nvidia reflex , can you tell me an avg and how much standard deviation does it have comparing to RTSS? (Let's say you cap the framerate at 90% load on GPU as the maximum limit using RTSS) Is the frametime all over the place with Reflex?
Why I ask this: You said the in-game framelimiter has bad frametimes (Inconsistent) so I am trying to get the best of the best in terms of options (G-sync/Free Sync) + No GPU Bound (let's say best of both worlds, no tearing (SYNC OPTIONS) + Smooth (Frametime Consistency) + Lowest latency possible while having the first 2 (Not the fastest)
Also if you can test the new feature of RTSS (Scanline SYNC)
You have a great mind and I thank you for dedicating your time to do this for us and I hope you will continue to do so.
This video shows a guy that knows A LOT about this and it's a pleasure learning from it. 🤟
so nvidia reflex on is better than the boost option?
Watch the video
I guess you mean RLL ON vs. RLL ON+BOOST?
Boost just disables the power saving setting of the GPU, that will only result in minor gains.
WTF is Radeon Boost doing? You buy a high refresh rate monitor to get a responsive and as clear as possible gaming experience during fast movement and Radeon Boost doesn't really benefit the responsiveness, but completely ruins clarity. Who thought this was a good idea? Even ignoring any stability issues, I can't see a single reason why anyone would use this.
Radeon Boost is looking awful, you can see the difference even with YT compression and everything. The menu test, holy shit.
Yeah, not sure what they were thinking...
@@bad-sackingen5892 It might have been inspired by dynamic resolution in consoles
I'm so frustrated with my Radeon card the worst drivers, Always, the worst technology and always so so far behind of what the industry is providing...
Hey, could you test Escape From Tarkov netcode again? They said they have changed netcode a lot, but seems the same.
Did you disable fullscreen optimizations for any of these? Awesome video btw.
Great content as always :)
Just one question: throughout the video you've always talked about a "good" in-game frame limiter. Would you say the limiter in CSGO through console (fps_max x) is good? Or can you achieve even lower latency with an external limiter like rtss?
I think there's a bit of a misunderstanding here. There are two separate measurements for "good" fps limiters:
1) Good framepacing (how smooth image looks)
2) Good responsiveness (doesn't introduce additional input lag)
Both RTSS and Nvidia's limiter v3 offer smooth framepacing (which is GOOD), however since they are external limiters, they introduce additional input lag (which is BAD).
In-game limiters usually have worse framepacing (which is BAD), however they usually don't add any additional input lag (which is GOOD). You see where this is getting?
Anyway to answer your question, using fps_max to minimize latency is good enough, and using RTSS would actually increase the latency.
In terms of system latency Ubisoft games have horrible ingame FPS limiters (R6S is just one example - requires cfg edit to enable)
Overwatch, Source, Unreal, Frostbite Engine have good ingame FPS limiters based on my system latency tests.
External FPS limiters also perform worse than in-engine limiters (in terms of system latency) as just like NULL or Anti-Lag they try to pace the game from the outside.
@@BattleNonSense does this mean r6 limiter is not good or I can edit the cfg file and use it ?
@@BattleNonSense so if i have to cap fps in R6S, should i use rtss instead of ingame limiter? TIA
Fantastic video as always, liked
I like how scientific you are!
Awesome work. Your patreon link is broken btw.
Treasure trove of info do keep making these man
Love the content as always Chris. I've used Afterburner as a framelimiter for some time, any thoughts on using it or an external equivalent outside of the game? Thanks
you could make a video showcasing the input lag effect of the parsec latency in a local network with and without wifi, it would be pretty interesting
When you are talking about smoothness, there is thing to remember. While higher framerate may reduce latency, when you are already at your monitors refresh rate going higher framerate can make it less smooth. Having framerate very uneven with your refresh rate causes it to not be smooth. For example, having 150fps on 144hz monitor is smoother than 180fps. And for more noticeable example, having 150fps on 120hz feels like stuttering to me. That is, with no sync, im not sure how it works above your refresh rate on freesync or g-sync monitors, i only know in those smoothest is to limit fps little below your refresh rate.
So, if i had game that supports reflex and nvidia gpu, i would enable reflex but limit fps anyway (I assume there is no downside to enabling it). But i dont play any game that supports it and i have amd gpu currently.
I dont believe that improvement of radeon anti-lag is that low at least in every case. In Quake champions i keep my fps limited to 150, and its clear to me that i get lower input lag with radeon anti-lag. On my laptop i limit it to 118 fps (for freesync 120hz) and it doesnt keep fps up there always, on it i feel huge difference. Could it be that somehow it affects input lag in Quake Champions differently than in games you test?
blud really went and compared 2 completely different things that have absolutely nothing to do with each other 💀💀💀💀💀
With G-Sync if I enable Reflex in Overwatch, the game auto-caps FPS to 138 (on my 144 hz monitor). Is this the Nvidia driver level limiter kicking in like it does for ULLM automatically or is it the Overwatch FPS limiter kicking in? It’s “probably” the driver level limiter I would think given its behaving just like it does for the ULLM mode+Gsync.
In your prior videos you demonstrated that the in engine limiter has less latency and we know RTSS and the Driver limiter have basically the same latency.
Would this mean for users who are reaching the Max refresh of their Gsync panel and hitting that automatic FPS cap that they’ll actually get worse latency with reflex ON since they’ll be running into the driver limiter? They could instead do Reduce Buffering ON + a manual cap at 140 for example, but we only know how capped + FPS limit compared to Reflex uncapped currently.
It would be useful to see how Reflex + Nvidia driver level limiter / the auto FPS cap behavior on Gsync panels compares when you’re actually reaching that target.
Currently there’s no way to disable this auto capping behavior so if you use another limiter you have to cap beneath the auto limiter by say 3 or 4 frames which wastes overhead. Thanks,
hey dude, would be awesome to see an updated look at Escape from Tarkov. they've supposedly done a lot of server modernization in the past year and the community can't decide whether it's better or worse. could use some cold hard data from the man himself
very late reply but seconded, I love this game and felt like it has improved quite a bit.
You sir are brilliant!!!! I love it. Liked & Subscribed! I need you to help me decide if I should get the Zowie BenQ XL2456k dyac+ 240hz or Alienware AW2521H Ulmb 360hz! No explanation needed! 😕
Very good video.
Conclusion: Avoid using such low latency features (unless your gpu load is at fixed 99%), stick to in-game fps limiters (capped to slightly lower framerate that reduces gpu load).
This includes nvidia reflex? Im cpu bound what is better fps cap or nvidia reflex?
Good video, you covered everything I think.
There's a free Quake-like game on Epic, called Diabotical, that has optional multi-threaded input where an input thread runs at ~500hz, as well as giving you control over buffering (1 to 5 frames) settings, and a pretty good in-game FPS limiter. If you have time to test whether multi-threaded input is actually a boon there, and what combination of settings is "best", I would love to see what results you get.
And all my friends still call me crazy when i tell them to use in game fps cap smh at least i try. 😂
Share this vid.
The way that AMD is playing catchup here is interesting in implementation but embarrassing in practice. You get your visuals set to mud in a way AMD thinks you won't notice to not even get the same benefits as Nvidia's Reflex does at full resolution. No matter how much people dump on Nvidia and their business practices, software advances like this, DLSS, RTX, integer scaling, Gsync, hardware acceleration in productivity apps, etc, are what will keep me loyal to this brand. They are a significant value add to me.
@Falcon_ Their version is RTX is also very underwhelming. They're trying to put out pretty much whatever they can to claim they have feature parity, but thus stuff clearly doesn't have the R&D behind it.
AMD did catch up to nvidia in terms of raw processing power.
Now they must step up their game in the software department to compete with DLSS and RLL.
@@BattleNonSense I specifically meant software. Check your email, I want a non-Patreon donation option.
Thanks for what you do!
One thing that is very interesting to consider is how much latency remains after subtracting frame draw time and the result of the SOL test (which covers scanout time, hardware latency, polling rate, ect) to see how much latency is caused by imperfection of the games input handling (careful not to confuse framecap delay with gpu draw time)
There is actually another very interesting aspect that games which provide the Reflex Flash Indicator allow you to measure.
And that is how much extra delay is caused by the game's animation system as well as (in some games) intentional trigger delay.
Where did you manage to buy an LDAT? I've been looking to buy one
That AMD tech is interesting, but it creates an inconsistency in response, which seems like the exact opposite of what you would want, even if its more responsive at times.
Look like its need to be polished and will be perfect
@@iraklimgeladze5223 i dont think there is anything "perfect" about variable resolution and inconsistencies that it introduces.
@@WalrusFPGA Static scene does not need high fps, only high fps we need is in the motion and when we control this motion. I need higher FPS when i moving around and higher sharpness when i just looking
@@iraklimgeladze5223 sounds like this is something you dont want then, as it decreases your resolution when you move around..
@@WalrusFPGA lol, when you move around its harder to see. It's how humans eyes see.
Good Video as always
What's the current consensus for limiting FPS in games with no engine limiter? still RTSS? And has the opinion on GSYNC changed at all? Is latency better or worse with it enabled?
So how's the fps limiter in Call of Duty Cold War? It seems to fluctuate all over the place for me rather than holding a constant fps, does it also have bad input lag?
Love your videos, im a new at PC gaming, can you make a video showing the best NVIDIA control panel options for gaming ? Its so many options and kinda confusing.
Honestly, you don't need to touch anything in the NVCPL.
Except for when you own a G-Sync monitor there are 2 options that you want to look at
1. G-Sync is enabled
2. always choose highest available refresh rate.
@@BattleNonSense Thx for the reply ! I see a lot of these "guides" for better performance after some tweaks on the NVCPL. I put the 'Power Management mode' in Prefer Maximum Performance and 'Texture Filtering - Quality' at High Performance. Does this settings actually change anything ? How about 'Threaded Optimization' , 'Shader Cache' ? Does Low Latency Mode works and NVIDIA Reflex is on ?
Apex Legends has some issues with it's Lag compensation, try turning it off(cl_lagcompensation 0) and go into the fiering range. you can easily hit consecutive wingman shots while draging the aim over. with Lag compensation on(cl_lagcompensation 1) tracking is the only way of hitting consecutiv shots. in game with the lag compensation off(cl_lagcompensator 0) you will get a ton of no-regs.
But the lag compensation favours tracking aim over dragging/flicking. if you flick and rest your aim over the target you will hit, but if the mouse is dragged over target it will count as a miss
Which file within the game should I use: cl_lagcompensation 1? Where do I find? Could you inform? Thanks!
I’ve been using NULL with Reflex and my latency feels amazing. Also use a timer resolution but the one i own also is a standby list cleaner which i still don’t know is all the way safe. But on controller you feel the latency even more when it jumps than you do on kbm, atleast i do. I also cap my frames on all GPU heavy games because I don’t have a Gsync monitor (i have one on the way now)
@Falcon_ should hpet be disabled before using any kind of timer resolution tool?
I know this sounds odd but could you look into this for me, War Thunder has both support for DLSS 2.0 and Nvidia Reflex but no site (including Nvidia's) show "no support" for Reflex and any article as of Feb 2021 and newer do not show DLSS support for WT either. just find this highly odd considering WT is a fairly large game as its a direct competitor to WoT and WoWs and also are the same company behind "enlisted" which supports Reflex.
Oh thanks for the info, this might help with something I am working on.
You aren't going to post a video about how I flew a jet into the palm tree right before you were about to unlock the black dice camo?
Literally just searched for a video about this and you made this today. Lets dive in
Boost should be used with chill. Higher frames during movement without being GPU limited.
That makes no sense because you handicap Boost with Chill.
And chill performs much worse than an in-game FPS limiter just like nvidia's FPS limiter.
@@BattleNonSense Depends, Chill with 120fps during movement should have lower input lag than an in-engine limit with 60fps
Further, not all in-engine limiter are created equal.
To illustrate:
Your GPU is capable of ca. 70fps in standard resolution.
Chill limits to 60fps without movement and reduces the GPU utilization thus reduces input lag.
When you move the cursor with Boost your GPU is capable of 150fps but again with ca. 100% utilization.
Chill can also limit your fps to 120fps during movement and can so guarantee you never reach 100% GPU utilization.
@Battle(non)sense any chance you can do a motion delay test for a good few of the top mice? GPW, Finalmouse, Endgame gear, ZOWIE
Can you check what happens if we cap the fps with RTSS to 59 and have reflex enabled, what's the behaviour of latency then?
Starkes Video wie immer! :)
Huh. Nvidia Reflex has always felt off to me but I also used a 300 fps limit in warzone. Maybe setting to unlimited will feel better.
Reflex Low Latency will not engage at all when you cap your FPS and you reach that FPS cap value.
But where comparison latency between 1600 dpi mouse and 800 dpi? :P is there any difference at 1k hz? or difference is only visible like u show to us on 8khz? because looks like 3200 dpi or 1600 is much faster than 400 dpi if it comes to latency. Like it!
Come on AMD, time to step-up! Anti-lag 2, based on Nvidia Reflex technology!
Great video!
But what is best? 1. Ingame fps limiter on 241fps.
2.Ore activate reflex and disable ingame fps limiter. Games then run in 400 fps.
great job Battle(non)sense
Nice video thank you
Hi! I've never watched one of your video's before, but are you from germany? I started the video and the german accent insta punched me in the face xD
Edit: This is not a hate commentary, just a simple question, because I live in germany
Hmmm... so if I want to use an in-game frame rate limiter in order to both prevent gpu bound, as well as enable input delay consistency and frame consistency, is there any point to be using nvidia reflex since limiting FPS at a consistently sustainable value simulates a cpu bound case? At least Boost surely gives benefit, right? Or does it..? Hmm. I'm guessing not. I suppose that's why you mentioned multiple times that reflex seems to be like a very nice variable frame rate limiter, because it produces similar results to limiting frames at a consistent/reliable number, without having the frames actually limited.
Am I on the right track here?
thx for the video!!!
Hi! thx for this video. Do you know if super rendering increase input lag?
Great video, as always. I am just wondering if there is a difference when you use in-game limiter or the one in nVidia settings. I have i7 6700k and GTX 1060, while playing Apex Legends in 1080p I get around 144fps on low settings, where cpu load is 50%, and gpu load is 99%. I use the limiter from nvidia settings, set at 140 fps, but the game still feels a bit laggy, maybe it's because of 20 tick servers, I don't know. What do you guys think?
the nvidia FPS limiter in the NVCPL as well as AMD's CHILL perform worse than the in-engine frame rate limiter because just like NULL and Anti-Lag they try to pace the game from the outside.
Apex Legends is based on the source engine which has a good in-game FPS limiter, but Respawn does not offer an FPS cap option in the settings, and the fps_max parameter in the launch options does not work for me.
@@BattleNonSense Thank you, I just set the parameter in Origin and it locks at 135 fps, as intended. (used +fps_max 135). Keep up the great work!
Unbelievable! Radeon features need more work!
AMD did catch up to nvidia in terms of raw processing power.
Now they must step up their game in the software department to compete with DLSS and RLL.
Thank you so much , love your channel ❤️
as a conclusion, for me an AMD user i should enable anti-lag and 'try and error' to find a frame rate that my pc can maintain consistently in each game .. am i right?
Well said from 18:45
so what is the best config for the least latency while having the best fps in csgo since new features like this arent supported in csgo
Monitor your GPU utilisation with rivatuner and then limit FPS to 10% below your lowest framerate on the worst map
Sorry if this was covered already, but do I need to use the Overwatch frame rate limiter when I'm using Reflex+Boost or is that redundant? One or the other? Thanks!
Thanks for the awesome explanation! I just have 1 question, does it matter if NULL is set to ON while reflex is also on in a game?
Reflex Low Latency overrides NULL.
So when RLL is enabled ingame NULL is disabled by the driver for that game.
good work!
A really nice thing about radeon boost: It also works with vega igpus. So basically any amd laptop. This gives me in overwatch about 100fps on my amd apu laptop and 70fps in gta online both on fullhd. This is really awsome, because you can game on laptops, which are not really meant for gaming.
Woah your getting high framerate from an iGPU?! Thats amazing! What CPU and resolution are you playing?
@@buttflies7597 im playing on my lenovo slim 7 with an Ryzen 4800u on full hd with medium settings. But also lower end ryzen cpus can achieve that, because the gpus are realtively similar.
You can just reduce the render scale or the game resolution instead.
That way you will get a stable frame rate and frame times while with Boost FPS and Frame Times (as well as system latency) are all over the place.
@@BattleNonSense Nah, it then looks shitty. With radeon boost it still looks good and you dont have any stutters when moving...
Would be awesome to see an updated look at Escape from Tarkov.
Turn off game mode in Windows helps with tarky
What kind of performance differences would you see in a higher end overwatch config? 9700K/3080?
depends on ingame GFX settings and resolution.
nvidia tech support says reflex doesn't work on every monitor.?
Thanks a lot, awesome video!! What basically is not clear for me yet is the following: While having Nvidia Reflex Enabled, should the Low Latency Mode (LLM) in the Control Panel (NCP) be off as shown in the benchmarks? Would LLM on or ultra negatively affect the input lag? Or is it just not relevant anymore as soon as Reflex is on and only becomes affective when the game doesn't support Reflex?
Reflex overrides control panel options to avoid conflicts.
Hi Chris ist das Problem "Latenz reduzieren mit AntiLag" mit AMD Grafikkarten immer noch nachteilig? Ist Nvidia immer noch die bessere Wahl mit Reflex low latency in Games? Danke Dir
Hast du eine Antwort bekommen?
@@k0ns491 nein.
Das Thema hat sich erübrigt, da ich auf Nvidia setze auch für Streaming.
You didn't mention that boost also does variable rate shading. It's more than just simply adjusting render scale now. Also Apex Legends works fine for me on a 165hz 1440p monitor with a 5700xt so strange it didn't work for you. Did you DDU? Take this as constructive criticism cause I really do appreciate your work on this and your findings. Keep it up.