CS2 says using VSync + GSync + Reflex has lowest input latency

[phaze]

yo guys,

Today CS2 released a patch claiming using VS + GS + Reflex provides the smoothest looking game + lowest input latency, what's your take about this?

I tried but I feel the mouse is kinda heavy with these settings.

Reference:


[ VIDEO SETTINGS ]
Settings Recommendations
[*]Added a Settings Recommendation popup if your display's refresh rate is set below its maximum at startup. Using your display's highest refresh rate is generally recommended.
[*]Added a Settings Recommendation popup if NVIDIA G-Sync support is detected but not enabled at startup. Using G-Sync is generally recommended if your system supports it.
Added a Settings Recommendation popup if NVIDIA G-Sync is enabled but not V-Sync and/or NVIDIA Reflex. It is generally recommended to enable all three settings together when they are available. Note that applying these settings will limit your frame rate to your display's refresh rate or slightly lower, which is usually the smoothest-looking and lowest input latency settings combination. See https://help.steampowered.com/en/faqs/v ... -B0DA-9032 for technical details.

[F1zus]

Reflex does not limit the frame rate. You need to limit the fps (4-6 less than the monitor's refresh rate) to avoid enabling the vsync buffer.
The second point is that Reflex turns on when the GPU is loaded at 95% or higher.
Thus, for the smoothest image possible, you need gsync on (or freesync) + vsync + fps limit.
I've been playing with this setting for about 2 years and don't feel any additional mouse lag (technically it increases by 1-2ms).

[joseph_from_pilsen]

Reflex does limit the fps in dx11 render (doesnt in vulkan only but vulkan generally sucks as its not native, only emulated = fps count is always lower), in my case it limits whenever is gsync+vsync+reflex about 7-8 percent fps below the refresh rate (223fps at 240hz, 260fps at 280Hz etc.). However, i didnt test it for few months as im using vsync fast since i upgraded my PC few months ago (and i cant see much difference in term of latency, its maybe from a simple reason that cs2 minimums are never over 200fps).
Because there is a way outside using vulkan how to disable the limiter - use vsync fast (very suitable for 7800X3D and similar fast CPUs), with vsync fast however i cant tell much perceivable difference, maybe due to 0.1 and 1 percentile minimum fps which are always the same and always deep below refresh rate in cs2. Also servers have huge impact, 9 of 10 valve (and not only valve) servers cause the game to run choppy with stutters (netgraph must be a straight green line without holes, without fluctuating and without any yellow/red bars - impossible to get at most servers).

Edit: i compared actual vsync+gsync+reflex cap vs gsync+reflex+vsync fast and vsync fast is a very very tiny snappier on feeling but vsync with the hard fps cap feels a little more resilient to any imperfections causing microstutters, microlags and worse persistence. However its a trade off, at good quality server + good fps over 500 is vsync fast a little better choice. The problem is that the game engine is trash and 500+fps stable is a myth, quality server another myth. Offline it runs always perfect. Funny fact - the new mills (trash)map stutters at 500-600 fps much more than the other new (trash)map at mid at 300fps (even if the netgraph is green straight line). At both synces, yes vsync fast stutters a little more but even normal vsync isnt capable to remove the stutter totally fully too.

[cursed-gamer]

Ye ye ye Valve developers trying to patch their fucked up game with VSync + GSync solution. What about hiring developers who actually know what they do?

[Luggah]

but what about dx11 having increased input lag after tabbing in/out?

[Slender]

750 fps average
drop to 300-250fps all the time (server issue).
gsync and vsync in comp. game in 2024, clowns in valve.

[ahead]

750 fps average
drop to 300-250fps all the time (server issue).
gsync and vsync in comp. game in 2024, clowns in valve.

The server cannot affect the client's FPS.

[Slender]

750 fps average
drop to 300-250fps all the time (server issue).
gsync and vsync in comp. game in 2024, clowns in valve.

The server cannot affect the client's FPS.

why i dont have any drops when i go with local server (with / without bots), and when no one connect to server (online dm with bots) ?
im use 360hz monitor and i have avg ~ 600-700 fps.
im set 2.78ms (that is 360fps) in settings, when it lower then that value, numbers is red.
im always have red when play ranked.

[joseph_from_pilsen]

750 fps average
drop to 300-250fps all the time (server issue).
gsync and vsync in comp. game in 2024, clowns in valve.

The server cannot affect the client's FPS.

Thats not true. Even in csgo bad server tanked FPS a lot and caused drops. I posted here about 2 yeears ago a measurement between 3 servers at the same map and the differences were clear. It was about 380 vs 350 vs 275fps (offline, wase and then a server with known var/sv issues). You can do similar test today in cs2 on your own if you want, expect similar results. Do a fps/frametime test with wase vs pracc vs valve trash dm when they are all about the same player load and the differences will be obvious too. Also there were tests proving that even players inventory and workshop subscriptions have (negative) impact on fps (thats why brand new acc is always smoother than 3k+ hrs old acc with medals and another clutter).

edit: link to the old test here viewtopic.php?f=10&t=10168&p=82523 - an advice, dont use same servers today as the "reference bad one" and "reference good one". The "bad one" upgraded their hardware past years (after I kept reminding them ) so its no more a reliable "bad server". On the other hand, the "good one" - wase - have well performing but also poor perf servers.

Btw the fps and frametimes are tanking online so terribly that even a blind person must notice it... Mostly at valve crap servers. Not at all of them and not always at all places in the map but very easy to notice. Good thing is also to bind netgraph chart to TAB key so you can see after any lag occured whether it was your impression or a reality.

[ahead]

My statement is true. I looked at your link, and your test involves additional players with different game situations, making it an unclear experiment. You can easily prove my statement by creating a dedicated server and connecting to it from your client. Then, you can change the SERVER FPS via the console and see that your CLIENT FPS does not change.