CS2 What Is The Best Setup

[hEnzeee]

Gsync does work, the excessive fps are shown in graph, but the render displays only as much fps as its refresh rate. The render chooses the more actual frame if it has a choice (typically, if you have 500fps and 360hz monitor, the extra 140 frames are not displayed as they are superseded by another 140 more actual frames)

Hmm sounds quiet interesting. Well for me. If i use "fast sync" + Freesync + gsync, and play with uncapped fps ingame, i get aound 600+ then there will be alot of sscreen tearing. If i cap to 357, there is no screen tearing. i have benq xl2566k 360 hz monitor.

[Riggedinho]

Counter Strike has always been very special and it's even more so with CS 2.

Same principle as CS:GO you do not need Nvidia Reflex or G-SYNC, it is absolutely necessary to purify... in order to have the lowest possible latency.
The main benchmark for this is your "MS" in the net_graph, the lower and more stable it is, the better and more fluid your game will be.

“MS” are linked to a symbiosis between your maximum FPS ceiling and your refresh rate.

Unfortunately CS 2 being greedy it requires a very good computer to have the lowest MS possible.

the maximum threshold which you are sure that your FPS never falls below, in short the threshold of your constant max fps.
And so to have 600 constant FPS in any situation smoke nade etc... in this game you have to pay for an expensive computer...

Personally with an i5 13600KF and rtx 3060 in native resolution 1920x1080 in 5on5 on a Faceit server on Mirage I have an average of 3.5 MS and on Ancient 6 of course because not same FPS stability

This game is still a bit of a pay2win, like Twistzz who said that he felt much better with an RTX 4090 rather than an RTX 4070.

[Carioca1]

In my experience (100+ hours of benchmarking cs2), the BEST setup for cs2 is less related to the hardware requirements, but more to do on optimising every inch of performance on your hardware. Of course, you need a certain minimum for highly competetive play, say an i7-14700K or Ryzen 7 5800X3D combined with at least an RTX 3060 or better.

However, FAR MORE IMPORTANT is actually deeply understanding your system in terms of how it performs and behaves when changing in-game graphics settings. In my experience, you want to find the graphics settings at a given resolution that lead to IDEAL values, aim at:
- GPU usage | avg: 70-80% | max < 96%
- CPU core thread usage | avg: 65-75% | max < 92%
- 1% low average fps > refreshrate | lower Hz if not
- frametime consistency | horizontal line, no major spikes.

Now, let me explain WHY I recommend these usage values, not just for cs2 but for any fast paced FPS.

To low GPU usage significantly handicaps frametime consistency because you shift higher load spikes (98%+) to the CPU, thus, achieving a CPU bottleneck scenario much faster. Also, you are under-using your GPU unnecessarily, when it could provide more performance. To high GPU usage is also bad, because this can cause a pike up in the render queue, where too many pre-rendered frames arrive from the CPU, creating a GPU bottleneck potentially. This can add notable input lag, however, REFLEX was designed for such scenarios.

To low CPU usage, leads to performance loss because the CPU isn't working to its full potential. To high usage is perhaps the worst experience you can have, since that would constitute a CPU bottleneck (99%+ usage). Among all bottlenecks, the one you want to avoid most is the CPU one. It introduces terrible microstutters and lag (FAR FAR worse than the GPU Bottleneck case).

Benchmark your competive match using CapeframeX (free software) for 10 minutes on different maps. Note down your 1% low average, 0.1% low average and enable the GPU-Busy metric on the WORST of all benched maps. Now tweak your cs2 graphics settings such that not only your 0.1% and 1% low increases, but also that you achieve BEST harmony between CPU and GPU, which is determined by the GPU-Busy metric.

The GPU-Busy metric measures how much time the GPU spends rendering per frame (ms), so you can compare it to the frametime (=CPU render time in ms), the closer these two are together/aligned, the more ideal you are sharing workloads between the two (avoiding sleep times). In CapeframeX, GPU-Busy is give as a deviation metric, that is called "GPU-Busy Deviation" measured in percentage. The lower the percentage, the less the GPU and CPU render times deviate from another (= the better). I recommend a GPU-Busy Deviation between 15-25% for CS2, not higher. Lower than 15% would be good, but it is likely you exceed 96% GPU max usage by trying, so be careful.

In short, to find the best setup in CS2 is not only hardware related, but also finding those graphics settings that balance your system at hand perfectly. Pay special attention to 0.1%, 1% lows and GPU Busy Deviation. Also, if your 1% lows < max. Refreshrate, lower your Refreshrate to: Hz = 1% low fps - 5 . This will guarantee refresh compliance at all times (increases smoothness). For example your worst 1% low averaged 207 fps (on a 240Hz monitor), then go an make a custom resolution with a Refreshrate of 200Hz.

[Slender]

In my experience (100+ hours of benchmarking cs2), the BEST setup for cs2 is less related to the hardware requirements, but more to do on optimising every inch of performance on your hardware. Of course, you need a certain minimum for highly competetive play, say an i7-14700K or Ryzen 7 5800X3D combined with at least an RTX 3060 or better.

However, FAR MORE IMPORTANT is actually deeply understanding your system in terms of how it performs and behaves when changing in-game graphics settings. In my experience, you want to find the graphics settings at a given resolution that lead to IDEAL values, aim at:
- GPU usage | avg: 70-80% | max < 96%
- CPU core thread usage | avg: 65-75% | max < 92%
- 1% low average fps > refreshrate | lower Hz if not
- frametime consistency | horizontal line, no major spikes.

Now, let me explain WHY I recommend these usage values, not just for cs2 but for any fast paced FPS.

To low GPU usage significantly handicaps frametime consistency because you shift higher load spikes (98%+) to the CPU, thus, achieving a CPU bottleneck scenario much faster. Also, you are under-using your GPU unnecessarily, when it could provide more performance. To high GPU usage is also bad, because this can cause a pike up in the render queue, where too many pre-rendered frames arrive from the CPU, creating a GPU bottleneck potentially. This can add notable input lag, however, REFLEX was designed for such scenarios.

To low CPU usage, leads to performance loss because the CPU isn't working to its full potential. To high usage is perhaps the worst experience you can have, since that would constitute a CPU bottleneck (99%+ usage). Among all bottlenecks, the one you want to avoid most is the CPU one. It introduces terrible microstutters and lag (FAR FAR worse than the GPU Bottleneck case).

Benchmark your competive match using CapeframeX (free software) for 10 minutes on different maps. Note down your 1% low average, 0.1% low average and enable the GPU-Busy metric on the WORST of all benched maps. Now tweak your cs2 graphics settings such that not only your 0.1% and 1% low increases, but also that you achieve BEST harmony between CPU and GPU, which is determined by the GPU-Busy metric.

The GPU-Busy metric measures how much time the GPU spends rendering per frame (ms), so you can compare it to the frametime (=CPU render time in ms), the closer these two are together/aligned, the more ideal you are sharing workloads between the two (avoiding sleep times). In CapeframeX, GPU-Busy is give as a deviation metric, that is called "GPU-Busy Deviation" measured in percentage. The lower the percentage, the less the GPU and CPU render times deviate from another (= the better). I recommend a GPU-Busy Deviation between 15-25% for CS2, not higher. Lower than 15% would be good, but it is likely you exceed 96% GPU max usage by trying, so be careful.

In short, to find the best setup in CS2 is not only hardware related, but also finding those graphics settings that balance your system at hand perfectly. Pay special attention to 0.1%, 1% lows and GPU Busy Deviation. Also, if your 1% lows < max. Refreshrate, lower your Refreshrate to: Hz = 1% low fps - 5 . This will guarantee refresh compliance at all times (increases smoothness). For example your worst 1% low averaged 207 fps (on a 240Hz monitor), then go an make a custom resolution with a Refreshrate of 200Hz.

low settings + enable only p0 state gpu

[Carioca1]

low settings + enable only p0 state gpu

Honestly, this will depend highly on your specific CPU + GPU combination. On my system (CPU: i9-13900K, GPU: RTX 4080), I achieved best results with MSAA 8x, Shadows High, Ambient Occlusion Medium, HDR Quality and all other settings on Low. This combination gives me on competitive NUKE @ 1632x1080 (as an example):
- CPU usage | avg: 66% | max: 90%
- GPU usage | avg: 78% | max: 94%
- GPU Busy Deviation = 18%
- Total avg = 562 fps
- 1% lows = 231 fps
- 0.1% lows = 194 fps
- frametimes | min: 0.8 ms | avg: 2.1 ms | max: 7.8 ms

Of course I was curious and tested MSAA 2x, with ALL settings on Low as well. Results were:
- CPU usage | avg: 89% | max: 98%
- GPU usage | avg: 44% | max: 67%
- GPU Busy Deviation = 43%
- Total avg = 656 fps
- 1% lows = 213 fps
- 0.1% lows = 184 fps
- frametimes | min: 0.6 ms | avg: 1.8 ms | max: 16.4 ms

As you can see, on high-end and powerful components, setting everything to LOW will only improve average FPS. However, much more important are the 0.1% and 1% lows which suffer quite a bit, thus, reducing perceived smoothness. Also, this REDUCTION in the fps LOWS is primarily caused by high GPU-Busy Deviation as a result of overloading the CPU and underloading GPU. Also, frametime (ft) consistency (= max ft - min ft), gets worse, DESPITE the average being slightly lower (1.8 ms vs 2.1 ms). Heavy fluctuations and spikes are felt really bad, since it can cause small microstutters (fluctuations between 0.6 ms and 16.4 ms are horrible, 0.8 ms to 7.8 ms is much smoothner).

Anyways, this post is NOT meant to enforce or suggest my graphics settings, but rather to share my experience about the game behaviour on given hardware. It is key to find the appropriate graphics settings ON AN INDIVIDUAL BASIS, every system is different. Ultimately, it will boil down to personal preference. Some people are more, some less sensitive to fluctuations abd inconsistencies.

In any case, I do hope this information can be of help.

[Slender]

low settings + enable only p0 state gpu

Honestly, this will depend highly on your specific CPU + GPU combination. On my system (CPU: i9-13900K, GPU: RTX 4080), I achieved best results with MSAA 8x, Shadows High, Ambient Occlusion Medium, HDR Quality and all other settings on Low. This combination gives me on competitive NUKE @ 1632x1080 (as an example):
- CPU usage | avg: 66% | max: 90%
- GPU usage | avg: 78% | max: 94%
- GPU Busy Deviation = 18%
- Total avg = 562 fps
- 1% lows = 231 fps
- 0.1% lows = 194 fps
- frametimes | min: 0.8 ms | avg: 2.1 ms | max: 7.8 ms

Of course I was curious and tested MSAA 2x, with ALL settings on Low as well. Results were:
- CPU usage | avg: 89% | max: 98%
- GPU usage | avg: 44% | max: 67%
- GPU Busy Deviation = 43%
- Total avg = 656 fps
- 1% lows = 213 fps
- 0.1% lows = 184 fps
- frametimes | min: 0.6 ms | avg: 1.8 ms | max: 16.4 ms

As you can see, on high-end and powerful components, setting everything to LOW will only improve average FPS. However, much more important are the 0.1% and 1% lows which suffer quite a bit, thus, reducing perceived smoothness. Also, this REDUCTION in the fps LOWS is primarily caused by high GPU-Busy Deviation as a result of overloading the CPU and underloading GPU. Also, frametime (ft) consistency (= max ft - min ft), gets worse, DESPITE the average being slightly lower (1.8 ms vs 2.1 ms). Heavy fluctuations and spikes are felt really bad, since it can cause small microstutters (fluctuations between 0.6 ms and 16.4 ms are horrible, 0.8 ms to 7.8 ms is much smoothner).

Anyways, this post is NOT meant to enforce or suggest my graphics settings, but rather to share my experience about the game behaviour on given hardware. It is key to find the appropriate graphics settings ON AN INDIVIDUAL BASIS, every system is different. Ultimately, it will boil down to personal preference. Some people are more, some less sensitive to fluctuations abd inconsistencies.

In any case, I do hope this information can be of help.

run fps benchmark

[Carioca1]

I am already running with only p0 enabled, since "prefer maximum performance" in NVIDIA CONTROL PANEL will run GPU in P0 mode, which means it wont use lower P states to reduce power draw.

[Slender]

I am already running with only p0 enabled, since "prefer maximum performance" in NVIDIA CONTROL PANEL will run GPU in P0 mode, which means it wont use lower P states to reduce power draw.

no, that not

[Carioca1]

Yeah I figured it out. I forced P0 over registry tweak as recommended, however, performance is EXACTLY to same as locking the GPU Clock of my RTX 4080 to a fixed OCd clock of 2950 MHz (+120 MHz OC) @ 12000 MHz effective memory clock (+750 MHz OC) in Afterburner (at 1150 mV). The advantage of Afterburner is that you can save it to a profile and switch it off after finishing games. In terms of performance the Clock locking in Afterburner is identical to the registry tweak of forcing P0, I run several Benchmarks to confirm.

[Irvinboy]

In my experience (100+ hours of benchmarking cs2), the BEST setup for cs2 is less related to the hardware requirements, but more to do on optimising every inch of performance on your hardware. Of course, you need a certain minimum for highly competetive play, say an i7-14700K or Ryzen 7 5800X3D combined with at least an RTX 3060 or better.

However, FAR MORE IMPORTANT is actually deeply understanding your system in terms of how it performs and behaves when changing in-game graphics settings. In my experience, you want to find the graphics settings at a given resolution that lead to IDEAL values, aim at:
- GPU usage | avg: 70-80% | max < 96%
- CPU core thread usage | avg: 65-75% | max < 92%
- 1% low average fps > refreshrate | lower Hz if not
- frametime consistency | horizontal line, no major spikes.

Now, let me explain WHY I recommend these usage values, not just for cs2 but for any fast paced FPS.

To low GPU usage significantly handicaps frametime consistency because you shift higher load spikes (98%+) to the CPU, thus, achieving a CPU bottleneck scenario much faster. Also, you are under-using your GPU unnecessarily, when it could provide more performance. To high GPU usage is also bad, because this can cause a pike up in the render queue, where too many pre-rendered frames arrive from the CPU, creating a GPU bottleneck potentially. This can add notable input lag, however, REFLEX was designed for such scenarios.

To low CPU usage, leads to performance loss because the CPU isn't working to its full potential. To high usage is perhaps the worst experience you can have, since that would constitute a CPU bottleneck (99%+ usage). Among all bottlenecks, the one you want to avoid most is the CPU one. It introduces terrible microstutters and lag (FAR FAR worse than the GPU Bottleneck case).

Benchmark your competive match using CapeframeX (free software) for 10 minutes on different maps. Note down your 1% low average, 0.1% low average and enable the GPU-Busy metric on the WORST of all benched maps. Now tweak your cs2 graphics settings such that not only your 0.1% and 1% low increases, but also that you achieve BEST harmony between CPU and GPU, which is determined by the GPU-Busy metric.

The GPU-Busy metric measures how much time the GPU spends rendering per frame (ms), so you can compare it to the frametime (=CPU render time in ms), the closer these two are together/aligned, the more ideal you are sharing workloads between the two (avoiding sleep times). In CapeframeX, GPU-Busy is give as a deviation metric, that is called "GPU-Busy Deviation" measured in percentage. The lower the percentage, the less the GPU and CPU render times deviate from another (= the better). I recommend a GPU-Busy Deviation between 15-25% for CS2, not higher. Lower than 15% would be good, but it is likely you exceed 96% GPU max usage by trying, so be careful.

In short, to find the best setup in CS2 is not only hardware related, but also finding those graphics settings that balance your system at hand perfectly. Pay special attention to 0.1%, 1% lows and GPU Busy Deviation. Also, if your 1% lows < max. Refreshrate, lower your Refreshrate to: Hz = 1% low fps - 5 . This will guarantee refresh compliance at all times (increases smoothness). For example your worst 1% low averaged 207 fps (on a 240Hz monitor), then go an make a custom resolution with a Refreshrate of 200Hz.

Hi I am trying to follow these steps, in the end you say that we should make a custom resolution with a custom refresh rate. Can I leave the refresh rate as it is and just cap my fps ? Or what should I use to set that custom refresh rate ?