CS:GO (And other games) Stutter/Lag Fix

[Puff]

Hi!
I've been scouting this forum for a while and found some interesting info on many subjects, so wanna say thanks to you all.
It is time once and for all to leave CS:GO stutters subject (and other games for that matter), since I've seen quite a lot of posts and no concrete fixes.
Today I dedicated half a day to running benchmarks in CS:GO with CapFrameX to showcase different behavior in the game with different Nvidia settings.
I've run the tests 3 times, every time exiting the game between the runs. There were 2 scenarios with a workshop benchmark map and playing against bots. Against bots scenario:
Dust2 map, T side, Buy deagle, shoot couple bullets, reload, start benchmark, drop mid, kill a bot, run down short, plant bomb, look to ct spawn for the rest of the time (each capture is 60s).
Subjective experience: Uncapped and no V-sync just feels unplayable.
Nvidia panel Fps cap is I guess the next best thing if you don't want to use V-sync, but shooting a bot mid is still bad, but better.
All the fixes combined - you wouldn't even tell that there is any stuttering, game is butter smooth, V-sync added latency is negligible, since Fps is capped below sync rate. It truly feels like 9/10 experience (there are unavoidable mini stutters couple times in a game).
Here is the data to back it up. Use it and enjoy the games, Cheers!

CSGO.JPG (53.17 KiB) Viewed 2690 times

TLDR: V-sync on, G-sync on, Fps cap -3 Fps below screen's refresh rate In Nvidia panel(in game settings = uncapped fps and V-Sync off for these runs).

[Slender]

Hi!
I've been scouting this forum for a while and found some interesting info on many subjects, so wanna say thanks to you all.
It is time once and for all to leave CS:GO stutters subject (and other games for that matter), since I've seen quite a lot of posts and no concrete fixes.
Today I dedicated half a day to running benchmarks in CS:GO with CapFrameX to showcase different behavior in the game with different Nvidia settings.
I've run the tests 3 times, every time exiting the game between the runs. There were 2 scenarios with a workshop benchmark map and playing against bots. Against bots scenario:
Dust2 map, T side, Buy deagle, shoot couple bullets, reload, start benchmark, drop mid, kill a bot, run down short, plant bomb, look to ct spawn for the rest of the time (each capture is 60s).
Subjective experience: Uncapped and no V-sync just feels unplayable.
Nvidia panel Fps cap is I guess the next best thing if you don't want to use V-sync, but shooting a bot mid is still bad, but better.
All the fixes combined - you wouldn't even tell that there is any stuttering, game is butter smooth, V-sync added latency is negligible, since Fps is capped below sync rate. It truly feels like 9/10 experience (there are unavoidable mini stutters couple times in a game).
Here is the data to back it up. Use it and enjoy the games, Cheers!
CSGO.JPG

TLDR: V-sync on, G-sync on, Fps cap -3 Fps below screen's refresh rate In Nvidia panel(in game settings = uncapped fps and V-Sync off for these runs).

Stable timer 0.5ms / 1ms (TSC tick) + fps_max 401 / 501 get you stable frametime without external lag-limiter and vsync (omg!)

[F1zus]

Stable timer 0.5ms / 1ms (TSC tick) + fps_max 401 / 501 get you stable frametime without external lag-limiter and vsync (omg!)

Why exactly 401 or 501 frames, and not 300/400/500?
I use the amd radeon chill function and limit the fps to 300. This gives the most stable framerate, but slightly higher latency than the in-game limiter.

[Slender]

Stable timer 0.5ms / 1ms (TSC tick) + fps_max 401 / 501 get you stable frametime without external lag-limiter and vsync (omg!)

Why exactly 401 or 501 frames, and not 300/400/500?
I use the amd radeon chill function and limit the fps to 300. This gives the most stable framerate, but slightly higher latency than the in-game limiter.

because it source engine.
just try to set 0.5ms timer, and lock fps.

[F1zus]

Stable timer 0.5ms / 1ms (TSC tick) + fps_max 401 / 501 get you stable frametime without external lag-limiter and vsync (omg!)

Why exactly 401 or 501 frames, and not 300/400/500?
I use the amd radeon chill function and limit the fps to 300. This gives the most stable framerate, but slightly higher latency than the in-game limiter.

because it source engine.
just try to set 0.5ms timer, and lock fps.

I don't like the 0.5ms timer, on Windows 11 this is a little more difficult to do. The islc program constantly accesses the kernel of the system literally every second and causes additional delay. I do not like it. The difference between 0.5ms and 1ms is small, the mouse is a little bit sharper, this does not give an advantage. I'd rather try 301 fps (my video card is not capable of delivering 400 stable).

[Slender]

Stable timer 0.5ms / 1ms (TSC tick) + fps_max 401 / 501 get you stable frametime without external lag-limiter and vsync (omg!)

Why exactly 401 or 501 frames, and not 300/400/500?
I use the amd radeon chill function and limit the fps to 300. This gives the most stable framerate, but slightly higher latency than the in-game limiter.

because it source engine.
just try to set 0.5ms timer, and lock fps.

I don't like the 0.5ms timer, on Windows 11 this is a little more difficult to do. The islc program constantly accesses the kernel of the system literally every second and causes additional delay. I do not like it. The difference between 0.5ms and 1ms is small, the mouse is a little bit sharper, this does not give an advantage. I'd rather try 301 fps (my video card is not capable of delivering 400 stable).

if you wont use 0.5ms timer (but why? islc is worse, why you not use service timer / timertool 3 or another timercall?), you can use quake values:
125, 250, 333
fps_max 126 / 256 / 334-343

[Puff]

Hi!
I've been scouting this forum for a while and found some interesting info on many subjects, so wanna say thanks to you all.
It is time once and for all to leave CS:GO stutters subject (and other games for that matter), since I've seen quite a lot of posts and no concrete fixes.
Today I dedicated half a day to running benchmarks in CS:GO with CapFrameX to showcase different behavior in the game with different Nvidia settings.
I've run the tests 3 times, every time exiting the game between the runs. There were 2 scenarios with a workshop benchmark map and playing against bots. Against bots scenario:
Dust2 map, T side, Buy deagle, shoot couple bullets, reload, start benchmark, drop mid, kill a bot, run down short, plant bomb, look to ct spawn for the rest of the time (each capture is 60s).
Subjective experience: Uncapped and no V-sync just feels unplayable.
Nvidia panel Fps cap is I guess the next best thing if you don't want to use V-sync, but shooting a bot mid is still bad, but better.
All the fixes combined - you wouldn't even tell that there is any stuttering, game is butter smooth, V-sync added latency is negligible, since Fps is capped below sync rate. It truly feels like 9/10 experience (there are unavoidable mini stutters couple times in a game).
Here is the data to back it up. Use it and enjoy the games, Cheers!
CSGO.JPG

TLDR: V-sync on, G-sync on, Fps cap -3 Fps below screen's refresh rate In Nvidia panel(in game settings = uncapped fps and V-Sync off for these runs).

Stable timer 0.5ms / 1ms (TSC tick) + fps_max 401 / 501 get you stable frametime without external lag-limiter and vsync (omg!)

Proof? I tested your claim and it doesn't hold any water.

[Traveler]

..., V-sync added latency is negligible,..

sub 1ms latency added?