CS2 0.1% 1% Lows

[fullmetalcsg]

Hello,

I have tried everything so I decided to cry for help in these forums since I've found a lot of valuable information here.

My setup:
7800X3D. No OC
4070 Super.
32GB 6000mhz 30CL RAM EXPO.
Lenovo Legion 1080p 360hz

Edit:
Mouse 8k hz OP1. I also tested with an old logitech 502.
Screen Lenovo Legion y25g-30 360hz gsync
redragon keyboard
Quadcast HyperX Microphone
DT770 headset

My problem:
Low 1% and 0.1% using capframeX on Ancient map

What I have tried:
On the GPU :
Overclock
Undervolt
Stock

On the RAM:
EXPO
DEFAULT

On Nvidia and CS2:
(nvidia screenshots)

-noreflex + low
reflex +gsync +vsync
gsync + low latency
fps cap on nvidia or cs2

On RTSS:
Framerate limit: 440
Framerate limit: 360
Scanline sync: 0
Scanline sync: Various values

On Monitor:
ULMB + 240hz

Extra information:
I have upgraded from a 6800XT. The average frames on that GPU were lower but the lows were better.
The 4070 super is running comfortably, not hitting 100% usage. If i force it to hit 100% with uncapped FPS it will have similar if not worse results. The CPU is nowhere near 100% usage either.
The GPU is scoring great on 3dmark scores and is stable on furmark.
I have run DDU twice.
I have not reinstalled windows yet.

Any ideas and suggestions will be greatly appreciated.

[Bloods]

Can you post your CS2 settings on 1080p and CapframeX screenshot from frame and which testing scenario, I can try compare to my own PC (which is much worse than yours but runs OK on higher res 1920x1440p.

And this problem only occurs on Ancient, I assume you have ran the ingame Dust2 benchmark is your frames good there ?

[joseph_from_pilsen]

post jitter graph results (ingame with telemetry on), networking does most of the lag usually. And yes, the game has bad netcode and even worse engine (networking affects render).

i have the same setup and its atrocious (dust2 offline with bots vs online at half empty server, -noreflex, fps driver lock to 391fps (390hz monitor)).

[kyube]

I have tried everything so I decided to cry for help in these forums since I've found a lot of valuable information here.

My setup:
7800X3D. No OC | 4070 Super. | 32GB 6000mhz 30CL RAM EXPO. | Lenovo Legion 1080p 360hz

My problem:
Low 1% and 0.1% using CapframeX on Ancient map

What I have tried: On the GPU — Overclock | Undervolt | Stock

On the RAM: EXPO | DEFAULT

On Nvidia and CS2:
(nvidia screenshots)
-noreflex + low | reflex +gsync +vsync | gsync + low latency | fps cap on nvidia or cs2

On RTSS:
Framerate limit: 440
Framerate limit: 360

On Monitor:
ULMB + 240hz

Extra information:
I have upgraded from a 6800XT. The average frames on that GPU were lower but the lows were better.
The 4070 super is running comfortably, not hitting 100% usage. If i force it to hit 100% with uncapped FPS it will have similar if not worse results. The CPU is nowhere near 100% usage either.
The GPU is scoring great on 3dmark scores and is stable on furmark.
I have run DDU twice.
I have not reinstalled windows yet.

Any ideas and suggestions will be greatly appreciated.

Could you also mention your motherboard, PSU, SSD, mouse, keyboard & other USB peripherals?
What Nvidia GPU driver are you using? Take a look at a few similar threads where a similarly titled topics have sprung up.
What WIndows version are you on? Run winver command
Could you take a ZenTimings screenshot of your current RAM timings profile?

I'm curious how you've assessed that your "1% & 0,1% lows are suboptimal on Ancient"
CapFrameX should tell you whether you were GPU or CPU limited during the recording.
I would advise you to run the CS2 benchmark map (either Dust2 or Ancient version) and use CapFrameX there.
Set up the "Statistic Parameter settings" like so:


In terms of CS2 in-game settings, I would stick to fps_max 360 and enable Nvidia Reflex.
The only other alternative which you could use is enable VRR (G-sync) & enable Reflex in-game with fps_max 325 (or 320) FPS cap
The larger cap away from the max refresh rate is to avoid the probability to jump out of G-sync range.

Hope this helps.

[gideonmontes]

Hello. Try setting the power plan to "Balanced", reset the BIOS, install XPM/EXPO. Just in case, try manually installing the chipset driver from the folder "C:\Program Files (x86)\AMD\Chipset_Software" run the shortcut "AMD_Chipset_Drivers", after running the shortcut select Modify. Reboot and check the PC.


I had stutters due to disabling power saving functions in the BIOS, as strange as it may be, to enable them it is enough to reset the BIOS settings.

[fullmetalcsg]

Can you post your CS2 settings on 1080p and CapframeX screenshot from frame and which testing scenario, I can try compare to my own PC (which is much worse than yours but runs OK on higher res 1920x1440p.

And this problem only occurs on Ancient, I assume you have ran the ingame Dust2 benchmark is your frames good there ?

My test is simply walk mid ancient and shoot at the big door for 20 settings while strafing and looking around. The results and settings are on the screenshot

[fullmetalcsg]

I have tried everything so I decided to cry for help in these forums since I've found a lot of valuable information here.

My setup:
7800X3D. No OC | 4070 Super. | 32GB 6000mhz 30CL RAM EXPO. | Lenovo Legion 1080p 360hz

My problem:
Low 1% and 0.1% using CapframeX on Ancient map

What I have tried: On the GPU — Overclock | Undervolt | Stock

On the RAM: EXPO | DEFAULT

On Nvidia and CS2:
(nvidia screenshots)
-noreflex + low | reflex +gsync +vsync | gsync + low latency | fps cap on nvidia or cs2

On RTSS:
Framerate limit: 440
Framerate limit: 360

On Monitor:
ULMB + 240hz

Extra information:
I have upgraded from a 6800XT. The average frames on that GPU were lower but the lows were better.
The 4070 super is running comfortably, not hitting 100% usage. If i force it to hit 100% with uncapped FPS it will have similar if not worse results. The CPU is nowhere near 100% usage either.
The GPU is scoring great on 3dmark scores and is stable on furmark.
I have run DDU twice.
I have not reinstalled windows yet.

Any ideas and suggestions will be greatly appreciated.

Could you also mention your motherboard, PSU, SSD, mouse, keyboard & other USB peripherals?
What Nvidia GPU driver are you using? Take a look at a few similar threads where a similarly titled topics have sprung up.
What WIndows version are you on? Run winver command
Could you take a ZenTimings screenshot of your current RAM timings profile?

I'm curious how you've assessed that your "1% & 0,1% lows are suboptimal on Ancient"
CapFrameX should tell you whether you were GPU or CPU limited during the recording.
I would advise you to run the CS2 benchmark map (either Dust2 or Ancient version) and use CapFrameX there.
Set up the "Statistic Parameter settings" like so:


In terms of CS2 in-game settings, I would stick to fps_max 360 and enable Nvidia Reflex.
The only other alternative which you could use is enable VRR (G-sync) & enable Reflex in-game with fps_max 325 (or 320) FPS cap
The larger cap away from the max refresh rate is to avoid the probability to jump out of G-sync range.

Hope this helps.

Mobo is Seasonic Focus 750 gold
8k hz op1 mouse, also tried with old g502 mouse
Lenovo Legion y25g-30 gsync 360hz
HyperX quadcast microphone
DT770 pro headset

Latest GPU and windows version.

I am running with gsync vsync fast, reflex + boost and cap of 440 fps in game. I have tried a lot of combinations as well as cap of below 360 but my averages were just lower.

Zen timings screenshot along with d2 benchmark map run

[kyube]

Mobo is Seasonic Focus 750 gold
8k hz op1 mouse, also tried with old g502 mouse
Lenovo Legion y25g-30 gsync 360hz
HyperX quadcast microphone
DT770 pro headset

Latest GPU and windows version.

I am running with gsync vsync fast, reflex + boost and cap of 440 fps in game. I have tried a lot of combinations as well as cap of below 360 but my averages were just lower.

Zen timings screenshot along with d2 benchmark map run

Thank you for including a CapframeX screenshot of the CS2 D2 benchmark map.
You seem to have cropped the voltages section of the ZenTimings screenshot you've provided, could you include those?
By latest drivers & Windows, do you mean R575 (such as 576.52) and 24H2?
The newest Windows version & Nvidia GPU drivers are report to cause severe stuttering & performance regression in a few games.
I strongly advise that you swap to pre-R570 branch (566.36 & older) and to Windows 11 23H2.

Your kit seems to be included in the motherboard's QVL list, but it's always best to verify this with a stress-testing app sw such as RamTestPro, HCI Memtest, Ycruncher, OCCT etc.
If you don't have the time / want to put in the effort — running EXPO disabled is the safest & easiest method for troubleshooting purposes since it isn't time-consuming.

As for the syncing technologies dilemma:
There's no need to use FastSync + Gsync, they cannot work at the same time. Only one can be utilised at the time.

You can only run Gsync + Vsync + fps cap below native refresh rate (they are both technologies which work <360fps)
Fast-Sync is only necessary when your frame rate is >360fps.
However, Vsync still adds a latency penalty even if you're on the surface <360fps. I would avoid using it, unless you are sensitive to tearing & want to avoid having it.

I don't think one should even consider using Fast-Sync at all. I've never found it to work adequately..

Try one of the following:
a.) Set display to native refresh rate (360Hz), fps_max 320, Gsync enabled, Reflex in-game enabled, LLM disabled in NVCP, Vsync to auto or disabled in NVCP
b.) Set display to native refresh rate (360Hz), fps_max 360, don't use any syncing technology (no vsync, gsync or LLM)

These should be a few good steps to troubleshoot your issue.

[fullmetalcsg]

Mobo is Seasonic Focus 750 gold
8k hz op1 mouse, also tried with old g502 mouse
Lenovo Legion y25g-30 gsync 360hz
HyperX quadcast microphone
DT770 pro headset

Latest GPU and windows version.

I am running with gsync vsync fast, reflex + boost and cap of 440 fps in game. I have tried a lot of combinations as well as cap of below 360 but my averages were just lower.

Zen timings screenshot along with d2 benchmark map run

Thank you for including a CapframeX screenshot of the CS2 D2 benchmark map.
You seem to have cropped the voltages section of the ZenTimings screenshot you've provided, could you include those?
By latest drivers & Windows, do you mean R575 (such as 576.52) and 24H2?
The newest Windows version & Nvidia GPU drivers are report to cause severe stuttering & performance regression in a few games.
I strongly advise that you swap to pre-R570 branch (566.36 & older) and to Windows 11 23H2.

Your kit seems to be included in the motherboard's QVL list, but it's always best to verify this with a stress-testing app sw such as RamTestPro, HCI Memtest, Ycruncher, OCCT etc.
If you don't have the time / want to put in the effort — running EXPO disabled is the safest & easiest method for troubleshooting purposes since it isn't time-consuming.

As for the syncing technologies dilemma:
There's no need to use FastSync + Gsync, they cannot work at the same time. Only one can be utilised at the time.

You can only run Gsync + Vsync + fps cap below native refresh rate (they are both technologies which work <360fps)
Fast-Sync is only necessary when your frame rate is >360fps.
However, Vsync still adds a latency penalty even if you're on the surface <360fps. I would avoid using it, unless you are sensitive to tearing & want to avoid having it.

I don't think one should even consider using Fast-Sync at all. I've never found it to work adequately..

Try one of the following:
a.) Set display to native refresh rate (360Hz), fps_max 320, Gsync enabled, Reflex in-game enabled, LLM disabled in NVCP, Vsync to auto or disabled in NVCP
b.) Set display to native refresh rate (360Hz), fps_max 360, don't use any syncing technology (no vsync, gsync or LLM)

These should be a few good steps to troubleshoot your issue.

Thanks for the information. I downgraded my Nvidia drivers, can't say i noticed any difference other than higher power consumption. I go around 140 when before i was about 120W.
I also run some CS2 benchmarks without EXPO and it was the same results.

I attached a lot of tests. All off is no gsync vync or reflex. All on is gsync + vsync fast or on, i didn't notice any difference.
The highest 0.1% I can achieve is with everything on, and uncapped fps as you can see on the last screenshot. Even on that high demand scenario, my power consumption is 140W and the GPU usage is not above 90%.

In general results with no gsync/vsync stutter a lot , even on 360 cap and very low GPU usage. It seems like the GPU is chilling and then suddenly stutters for no reason.

My 3Dmark benchmarks are really good so I am pretty sure it is not a problem with the GPU. I am also running a mild overclock on the GPU, i have benchmarked without it as well.

[fullmetalcsg]

Mobo is Seasonic Focus 750 gold
8k hz op1 mouse, also tried with old g502 mouse
Lenovo Legion y25g-30 gsync 360hz
HyperX quadcast microphone
DT770 pro headset

Latest GPU and windows version.

I am running with gsync vsync fast, reflex + boost and cap of 440 fps in game. I have tried a lot of combinations as well as cap of below 360 but my averages were just lower.

Zen timings screenshot along with d2 benchmark map run

Thank you for including a CapframeX screenshot of the CS2 D2 benchmark map.
You seem to have cropped the voltages section of the ZenTimings screenshot you've provided, could you include those?
By latest drivers & Windows, do you mean R575 (such as 576.52) and 24H2?
The newest Windows version & Nvidia GPU drivers are report to cause severe stuttering & performance regression in a few games.
I strongly advise that you swap to pre-R570 branch (566.36 & older) and to Windows 11 23H2.

Your kit seems to be included in the motherboard's QVL list, but it's always best to verify this with a stress-testing app sw such as RamTestPro, HCI Memtest, Ycruncher, OCCT etc.
If you don't have the time / want to put in the effort — running EXPO disabled is the safest & easiest method for troubleshooting purposes since it isn't time-consuming.

As for the syncing technologies dilemma:
There's no need to use FastSync + Gsync, they cannot work at the same time. Only one can be utilised at the time.

You can only run Gsync + Vsync + fps cap below native refresh rate (they are both technologies which work <360fps)
Fast-Sync is only necessary when your frame rate is >360fps.
However, Vsync still adds a latency penalty even if you're on the surface <360fps. I would avoid using it, unless you are sensitive to tearing & want to avoid having it.

I don't think one should even consider using Fast-Sync at all. I've never found it to work adequately..

Try one of the following:
a.) Set display to native refresh rate (360Hz), fps_max 320, Gsync enabled, Reflex in-game enabled, LLM disabled in NVCP, Vsync to auto or disabled in NVCP
b.) Set display to native refresh rate (360Hz), fps_max 360, don't use any syncing technology (no vsync, gsync or LLM)

These should be a few good steps to troubleshoot your issue.

Thanks for the information. I downgraded my Nvidia drivers, can't say i noticed any difference other than higher power consumption. I go around 140 when before i was about 120W.
I also run some CS2 benchmarks without EXPO and it was the same results.

I attached a lot of tests. All off is no gsync vync or reflex. All on is gsync + vsync fast or on, i didn't notice any difference.
The highest 0.1% I can achieve is with everything on, and uncapped fps as you can see on the last screenshot. Even on that high demand scenario, my power consumption is 140W and the GPU usage is not above 90%.

In general results with no gsync/vsync stutter a lot , even on 360 cap and very low GPU usage. It seems like the GPU is chilling and then suddenly stutters for no reason.

My 3Dmark benchmarks are really good so I am pretty sure it is not a problem with the GPU. I am also running a mild overclock on the GPU, i have benchmarked without it as well.