CS2 NVIDIA Reflex -

[i don't sign profiles.]

Hello,
I got the CS2 Test a few days ago and have been trying to mess around with the settings available to figure out any ways to optimise the transition between CSGO and CS2, and I made this interesting discovery and figured I'd share it in case anyone else was experiencing what I was.

CS2 now has an NVIDIA Reflex setting, which can be either "Off", "On", or "On + Boost". NVIDIA Reflex is something that many games use to reduce latency, so obviously most people's first instinct is to keep it enabled. Many people - myself included - use NVIDIA Control Panel to turn on "Low Latency Mode". The control panel setting works similarly to NVIDIA Reflex, but is not necessarily the same thing and may have differing results, however people say that NVIDIA Reflex is much better at reducing latency than the control panel setting.

On CSGO, I have fairly quick and accurate flicks. I use 1.65*400DPI sensitivity, and a 1.2 Zoom sensitivity multiplier. When I started playing CS2, I noticed that something was off with my timing/flicks - specifically against moving targets. My initial solution to this problem was increasing my Zoom Sensitivity Ratio to 1.25, which allowed me to hit flicks against moving targets, but I would miss stationary targets. This implied that I was having latency, as I needed a higher zoom sensitivity to flick to the moving target.

Eventually, I started to look into NVIDIA Reflex and realised that if you enable NVIDIA Reflex, it will override your Control Panel setting. After disabling the NVIDIA Reflex setting and restarting, I noticed that I was no longer hitting flicks on 1.25 Zoom Sensitivity Ratio - so after switching back to 1.2 I was finally hitting them the same as in CSGO.

I can't say for certain that the control panel setting is FASTER than NVIDIA Reflex, but all I can say is that in my case - it was. This may vary between different hardware and settings, but I think it's still worth a try. Even if you didn't use the control panel setting, I'd recommend trying it out for yourself and seeing if you find the game to be more responsive.

In Summary -
If you are having issues with latency and your muscle memory from CSGO isn't translating to CS2 - especially if you use the "Low Latency" setting in the NVIDIA Control Panel - try disabling NVIDIA Reflex entirely. If you don't use the Low Latency setting in the control panel, I would even recommend trying it out (for CSGO as well) and seeing if it helps. I know some people tend to think control panel settings and these random "FPS Boost" optimisations are placebo, but the Low Latency setting is definitely not one of those and has been incredibly noticeable for me for years.

In the end, introducing a new setting that messes with latency will of course mess with your muscle memory. If you've been playing CSGO for long enough and are very comfortable with your setup/settings, getting rid of these factors all-together may give you a head start. I hope this helps some people out who may have been dealing with similar issues when playing CS2, and if anyone has any additional input - or tests it out for themselves - feel free to comment.

---------------------------------------------

UPDATE - October 9, 2023

I do want to update this thread just to say that although there are some odd things going on with NVIDIA Reflex, there are some other things going on. My initial reason for doing this test was because my shots appeared to hit way off from where I was shooting them - even with weapon_accuracy_nospread 1.

After rewatching some clips I recorded of shots missing that clearly should've hit, I ended up doing a separate test recording my in-game footage at 120fps to see what was happening, and the difference was insane. I posted the test to Reddit over 3 months ago, and handed in the video to the CS2 test bugs email, and then ultimately stopped playing much CS2 after realising how bad the problem was and knowing there was nothing I could do to stop it. The link to the Reddit post can be found here.

About a month ago, it seemed like people started to pick up on this issue after a few YouTubers acknowledged these issues. I recently uploaded a new test that I recorded in 240fps way back in July (with direct comparisons to CS:GO), but never got around to editing until a few weeks ago. The link to that video can be found here.

Ultimately, I find my game to be way more smooth and less choppy with NVIDIA Reflex disabled. That much has been clear, and is shown in my tests from this thread. However, as far as my initial suspicions are concerned regarding the latency being the cause of my missed shots - there is clearly something else going on regarding the "sub-tick" system that's the main cause for why I had to drastically raise my zoom sensitivity to hit shots.

This might be common knowledge to a lot of the community, but I just wanted to make sure people don't see this thread and assume that NVIDIA Reflex alone is causing your shots to go nowhere when you flick. I saw a recent video of Ropz showing his CS2 settings where he mentions disabling NVIDIA Reflex due to the setting possibly having worse performance over leaving it disabled - link to the video can be found here. This might just be word from this thread being spread around, but I wouldn't write off NVIDIA Reflex as a possible setting that is causing some issues with your game's performance.

[Sudetenland ist mein Heimatland]

nice test but im affraid with wrong results.
The reason why you missed in cs2 was because CSGO has huge latency and what you did was "tuning" cs2 to behave same bad like csgo. If your muscle memory from csgo start working again, congratulations, you increased cs2 latency nearly to csgo level.
I have opposite experience, when rifling in cs2, the crosshair is MUCH faster than in csgo with nvidia reflex on, but thats what i want, i dont want sluggish heavy csgo aim not fast enough to do crazy flicks . But faster in term of reaction and responsibility, which i really dont wanna to tune down.
Btw ALWAYS when reflex is on, low latency mode should be off or overriden, its inferior to nvidia reflex and input lag is always worse with LLM only At least by working supported games (cs2 is supported, of course the second condition - working - is still a bit unknown variable).

Now some expected numbers (cs2 is only expectation based on valorant, csgo is known number), with nvidia reflex on the lowest achieavable latency on the input should be 5ms (valorant level), with LLM about 10ms. Csgo lowest possible latency achievable is 12ms. Thats why LLM feels similar ("better") to you, its similar bad. Muscle memory may work, but you will be soon beaten by people who learned to be faster with faster input.

[ގ󠁳ގ󠁳ގ󠁳⁧⁧Redline]

Thanks for the information, we need more helpfull threads like yours.

[i don't sign profiles.]

Автор сообщения: Wunderwaffe

If your muscle memory from csgo start working again, congratulations, you increased cs2 latency nearly to csgo level.

The reason I don't think this is the case is because in order for my muscle memory to properly translate, I had to INCREASE my zoom sensitivity using NVIDIA Reflex. This is because when someone would peak, there would be enough latency that by the time I see they peaked and flick as I would in CS, I would miss because I would shoot behind where they were. When I increased my sensitivity, it compensated for latency meaning by the time I would see the enemy cross, I would have to flick further than I am used to because of the extra delay.

If I were to have added latency, I would have been overshooting the enemy and would have needed to decrease my sensitivity to convert the muscle memory of the flicks and timing. I also noticed that the timing of when someone walked past my crosshair and when I needed to shoot was off as well, and I was shooting too late. I know for a fact that I wasn't overshooting because I had sv_showimpacts on and was clearly seeing where my bullets landed in relation to where the enemy was.

Again, whether this is always the case - I don't know. I've always found my PC behave different on CSGO from everyone else's. Maybe it's a weird circumstantial thing, but for me this almost definitively helped my latency.

[Sudetenland ist mein Heimatland]

steam forum is unfortunatelly not the best place. Sadly reddit csgo is dead for now (and maybe will never come back), and facebook is trash. I would recommend to go to blurbusters forum, there it will be taken seriously and people will validate it with numbers.

[i don't sign profiles.]

Автор сообщения: Wunderwaffe

steam forum is unfortunatelly not the best place. Sadly reddit csgo is dead for now (and maybe will never come back), and facebook is trash. I would recommend to go to blurbusters forum, there it will be taken seriously and people will validate it with numbers.

It's been painful with the GlobalOffensive subreddit being privated at the moment. I'll likely post this in there once it is unprivated, but I just know that I've found many solutions to problems via google searching my issue and finding a steam community thread. Hopefully if someone searches "CS2 NVIDIA Reflex" on Google, this might show up and help them out.

I'll check out Blurbusters as well, not familiar with it.

[68 лет пошлая]

good

[i don't sign profiles.]

Alright, I just did a test as scientifically as I can. I have a camera that shoots 240fps video, and I recorded 4 videos each with 10 AWP shots using the respective settings with my mouse in frame. I restarted the game for each instance, picked the same mode/team and walked to the same exact spot aiming at a wall. I tried to do each click as consistent as possible, and use the same reference points to count the frames after each click. I also tried Reflex with Low Latency mode turned off in the control panel but got essentially the same results. I also removed the worst result out of the 10 for each test.

These average latency times aren't a measurement of my actual latency, but they can be compared based on the two reference points I used to start/stop the frame count.

Here are the results;

Reflex Off -
5, 3, 3, 4, 5, 4, 3, 3, 3 - Frames it takes to display after each click

Averages;
3.67 frames - Average frames rounded to the nearest hundredth
15.25ms - (1000ms/240fps)*average frames

~~~

Reflex On -
4, 6, 3, 5, 3, 3, 4, 5, 6 -

Averages;
4.33 frames
18.02ms

~~~

Reflex On + Boost -
5, 5, 5, 4, 5, 5, 5, 6, 5 -

Averages;
5.00 frames
20.83ms
----------

Obviously there's a lot to take into account here. One frame is about 4.2ms so it's impossible to be precise. I have a 165hz monitor, so that is likely also a factor when it comes to shutter speeds/framerates. We are only dealing with 9 shots. Although they are in the same spot - the differences could increase or decrease depending on what is happening in-game in a normal gameplay situation.

I'm not saying that Reflex Off is the best way for everyone. I'm just saying - for some reason, it reduces my latency and if you are having similar issues, you might as well test it out.

[Sudetenland ist mein Heimatland]

165Hz monitor is bad :( Also post your hardware, you need to be sure its not bottlenecked. Also you need to test more settings (fps_max 400, fps unlimited, frame cap to monitor refresh rate etc) . Also you didnt post your settings, it should be with maximized fps and optimized nvidia control panel (1 prerendered frame only, no sync...)
Your all results are too close from each other and the latency is kinda bad considering valorant can go up to 5-6ms limit on high end hardware. However even if they are in a narrow margin of measurement error range, they can mean something. And not surprisingly - it doesnt seem to be much positive for the game if it runs so badly on your pc with reflex/boost on. Either its not working, or not properly configured (we need to see your settings) or different bottleneck (CPU etc.)

But a nice try, i kinda desperatelly need to get somewhere a high FPS camera, i want to do similar test, in cs2 and also at faceit as it seems to be extremelly input laggy for me.

Btw i have found some numbers in this article, its from pre-release test https://www.techpowerup.com/306912/nvidia-reflex-coming-to-counter-strike-2#g306912 . Dunno why the idiots test nvidia reflex at HIGH settings and HIGH resolution, even better numbers would be with low res + competitive settings but they need to sell overpriced crap gpus... AT low would be visible that you dont need 4070+ to play cs2 with single digit latency.
Your results are comparable to some garbo GTX1060 GPU levels. With opposite impact of nvidia reflex lol.
Sadly, all 3 my smartphones can record only 60fps, so i cant tell you how does my 3060Ti and ryzen 7600 (and tuned 6000mhz rams) at 280hz Asus VG259QM (budget gem monitor) perform in cs2 but i persist a big responsivity improvement in cs2 (reflex + boost on) versus csgo which has incredible lazy aim for me.

[i don't sign profiles.]

Firstly, I just wanna note that I wasn't initially planning on doing a test like this - I figured my initial post with the zoom sensitivity was good enough proof at least for my case. I decided to do the test with the camera because I was curious, but not exactly curious enough to do more combinations or longer tests.

Автор сообщения: Wunderwaffe

7800X3D and at least RTX4070Ti is a bare minimum

Whuh.

Автор сообщения: Wunderwaffe

The latency is kinda bad considering valorant can go up to 5-6ms limit on high end hardware.

As I said, the average ms times aren't my exact latency. Since I was unable to know exactly when my click was registered, I used the reference point of when my finger makes initial contact with the mouse (which isn't necessarily the moment I click), and then count the frames up until when the monitor scopes out of the AWP. Since I counted the frame of the initial contact before I click and the first frame the image is visible, that's 8ms of unknown. There's also the latency of my mouse & monitor that would add to the total.

Автор сообщения: Wunderwaffe

Your all results are too close from each other

That's because we are dealing with increments of 4.2ms and also in a game environment with no movement/action. We are simply shooting at a wall. If we assume 8 ms is shaved off, the results look like this; 7.25ms, 10.02ms, and 12.83ms. Now we are looking at more significant differences.

I have a GeForce RTX 3080 & an Intel i9 9900k. Clearly there is a CPU bottleneck, and I'm sure that is the root of most of my problems with CS. That is why I keep saying that this is likely not the case for everyone, but simply trying to offer a possible easy solution (that doesn't involve buying an $800 GPU and a $450 CPU) for those who may have the same issues as me. A CPU is next on my list, but I still think it's besides my point.

The game is running at 4:3 1440x1080 Stretched Resolution and all of the settings were on low using fps_max 0. The control panel is set to 1 prerendered frame and obviously no sync. I took a whole two days testing around with different settings in-game and on the control panel (not using the camera, but using the zoom sensitivity ratio) and felt the same way.

[Sudetenland ist mein Heimatland]

Yep, 9900K is really bad in cs2, i have seen fps benchmarks not going much over 200fps in cs@. With 3080 in 1440x1080 it may run with 500fps in mirage at higher settings. But with a CPU. Even my budget 7600 and 3060Ti runs the game at 400-500fps in mirage cs2 in 1440x1080.

[ziXXo]

Hope they release for more people to play :( . I want to test CS2 also :D

[i don't sign profiles.]

Автор сообщения: Wunderwaffe

Yep, 9900K is really bad in cs2, i have seen fps benchmarks not going much over 200fps in cs@. With 3080 in 1440x1080 it may run with 500fps in mirage at higher settings. But with a CPU. Even my budget 7600 and 3060Ti runs the game at 400-500fps in mirage cs2 in 1440x1080.

Yeah, I definitely need a CPU upgrade. When I got the 9900k, not only did they come out with a new series shortly after, but they came out with a brand new socket - so I couldn't upgrade while keeping the same motherboard. I just want to make sure I time it right so that doesn't happen again haha. The reason I have a 165hz monitor is because it's 1440p but I would like to get a faster 1080p monitor at some point.

[LTJACK]

Автор сообщения: Wunderwaffe

Yep, 9900K is really bad in cs2, i have seen fps benchmarks not going much over 200fps in cs@. With 3080 in 1440x1080 it may run with 500fps in mirage at higher settings. But with a CPU. Even my budget 7600 and 3060Ti runs the game at 400-500fps in mirage cs2 in 1440x1080.

What's the point of having more than 200 fps anyway? His screen is a 165hz so that's his limit displaying.

Funny to see people saying "I want 500 fps!" but are running 144/165hz screen...

[strb]

Автор сообщения: Jackz

Автор сообщения: Wunderwaffe

Yep, 9900K is really bad in cs2, i have seen fps benchmarks not going much over 200fps in cs@. With 3080 in 1440x1080 it may run with 500fps in mirage at higher settings. But with a CPU. Even my budget 7600 and 3060Ti runs the game at 400-500fps in mirage cs2 in 1440x1080.

What's the point of having more than 200 fps anyway? His screen is a 165hz so that's his limit displaying.

Funny to see people saying "I want 500 fps!" but are running 144/165hz screen...

500 fps is way above what you need for almost any monitor, but you might want to read this https://blurbusters.com/faq/benefits-of-frame-rate-above-refresh-rate/

[76561197967845157]

Автор сообщения: cat

Hello,
I got the CS2 Test a few days ago and have been trying to mess around with the settings available to figure out any ways to optimise the transition between CSGO and CS2, and I made this interesting discovery and figured I'd share it in case anyone else was experiencing what I was.

CS2 now has an NVIDIA Reflex setting, which can be either "Off", "On", or "On + Boost". NVIDIA Reflex is something that many games use to reduce latency, so obviously most people's first instinct is to keep it enabled. Many people - myself included - use NVIDIA Control Panel to turn on "Low Latency Mode". The control panel setting works similarly to NVIDIA Reflex, but is not necessarily the same thing and may have differing results, however people say that NVIDIA Reflex is much better at reducing latency than the control panel setting.

On CSGO, I have fairly quick and accurate flicks. I use 1.65*400DPI sensitivity, and a 1.2 Zoom sensitivity multiplier. When I started playing CS2, I noticed that something was off with my timing/flicks - specifically against moving targets. My initial solution to this problem was increasing my Zoom Sensitivity Ratio to 1.25, which allowed me to hit flicks against moving targets, but I would miss stationary targets. This implied that I was having latency, as I needed a higher zoom sensitivity to flick to the moving target.

Eventually, I started to look into NVIDIA Reflex and realised that if you enable NVIDIA Reflex, it will override your Control Panel setting. After disabling the NVIDIA Reflex setting and restarting, I noticed that I was no longer hitting flicks on 1.25 Zoom Sensitivity Ratio - so after switching back to 1.2 I was finally hitting them the same as in CSGO.

I can't say for certain that the control panel setting is FASTER than NVIDIA Reflex, but all I can say is that in my case - it was. This may vary between different hardware and settings, but I think it's still worth a try. Even if you didn't use the control panel setting, I'd recommend trying it out for yourself and seeing if you find the game to be more responsive.

In Summary -
If you are having issues with latency and your muscle memory from CSGO isn't translating to CS2 - especially if you use the "Low Latency" setting in the NVIDIA Control Panel - try disabling NVIDIA Reflex entirely. If you don't use the Low Latency setting in the control panel, I would even recommend trying it out (for CSGO as well) and seeing if it helps. I know some people tend to think control panel settings and these random "FPS Boost" optimisations are placebo, but the Low Latency setting is definitely not one of those and has been incredibly noticeable for me for years.

In the end, introducing a new setting that messes with latency will of course mess with your muscle memory. If you've been playing CSGO for long enough and are very comfortable with your setup/settings, getting rid of these factors all-together may give you a head start. I hope this helps some people out who may have been dealing with similar issues when playing CS2, and if anyone has any additional input - or tests it out for themselves - feel free to comment.

buy an ad