CS2 Frametimes question

[BumFlannel]

What are you using to cap your frame rate?

The in game frame cap sucks and always ends up giving you a frame rate lower than your cap that bounces around causing shit frame times. In my case with a 7950X and 7900XTX in a 5v5 official match with fps uncapped I can push an average of about 600fps depending on the map. The frame times are nasty though because the game is dropping fps by large amounts multiple times per second and it makes the game look low fps and micro stuttery.

My fps in 5v5 maps never drops below my refresh rate of 360hz, but if I cap to 360 using the in game cap my average will now be jumping between 320 and 360 giving again bad frame time and stutter, even though uncapped my fps will never drop below 360.

If I uncap in game and use the cap in the AMD driver I will get a solid 360fps with rock solid frame times.

The main issue with this game though is it's incredibly difficult to get smooth performance at high fps. You need a PC capable of high enough average fps that the huge frame time spikes won't drop you below your refresh rate. So if you have a 240hz monitor and want smooth gameplay you need a PC that can handle like 400 to 500fps average. You basically have to brute force like twice the actual fps you want to target to get a smooth experience, which is ridiculous.

However there's another phantom issue where your fps is high as usual, you're using the driver cap, frame times are solid, yet you can see and feel the game chugging. Frame rate looks low, mouse is sluggish, motion clarity is like looking through a frosted glass window (even on my qd OLED). The game just randomly does this is there eia no explanation for it.

I use only in game fps_max ,everything else gives me more input lag. Now im testing with fps_max 145 ,because it doesn't drop below 142/3 and i can get P1 144fps even in deathmatch so i can test things. Thats why i think fps its not the problem,something with the game is wrong ,because if you are alone or 2-3 people in the server its smooth even with 100fps.When there are more people in the server - screen tearing appear ,mouse start to feel heavy and this is with exactly the same FPS and 1%fps.

Today i tried valorant its so smooth with 144 and 300fps nothing like cs.

Capping in the driver rather than in game will give you the most stable frame times. The in game cap just sucks for some reason. But you're right about the game being inconsistently sluggish despite having solid frame times after capping. That seems to be a totally separate issue vs just raw frame times/general performance that no one can seem to figure out. I've had this for so many years now starting way back in cs go. Sometimes in CS:GO you would even see the whole screen jittering as you tried to move and turn your mouse, strafing and air strafing, but not when you stood still and turned your mouse.

I believe there's 2 things going on. First there is the general performance issues, then there is processing issues somewhere under the hood screwing up the timings of the simulation but not the rendering. So while you're seeing high fps but having sluggish gameplay, your PC is pushing the frames it should, but the actual simulation of game mechanics (movement, input, animation, recoil, etc) is all inconsistent and jittery causing the game to LOOK low fps even though it's not.

Some evidence for this. When this would happen in cs go it looked like a raw low fps/performance issue/64fps because the game wasn't interpolating properly and you were seeing things happen at the tick rate while frames were being rendered at, well, your frame rate. In cs2 you see the game interpolating/rubber banding you during or followed by frame time spikes but you see no issues with network telemetry. It's almost as if the local simulation of game logic is lagging and being interpolated to correct for that lag and is being visualised in telemetry as frame spikes, even though they aren't frame spikes in rendering, they're spikes in simulation and what is being rendered looks laggy because its rendering everything based on that laggy simulation.

TLDR, It appears to be simulation jitter/jitter on the actual processing of game logic, not rendering or network jitter. What causes it I have no idea.

[joseph_from_pilsen]

I discovered simple thing working on simple math...

The annoying frametime spikes are equal to a count of...

The hardware render time (as everybody would expect)...
...and... actual network jitter

Yes, its simple. When your network jitters in ping spike 30ms, your frametime spike will be 4ms+30ms = 34ms. It works, tested with laggy network. Its always the count of these 2 things, the shitty game doesnt even have render and input independent on networking. There is no smoothening like in valorant, where e. g. an occasional jitter wont freeze your aim or mess with your movement. Also the "Buffering to smooth over packet loss" doesnt work at all, no matter if i set it to 0 1 or 2 tick, the frametime lag is still the same, dunno what it does, but im sure what it does NOT, it doesnt smoothen the gameplay over the jitter even if 15.6ms 1 tick buffer is WAY more than avg 5ms jitter of my internet.

I have 2 internet connections at 2 addresses.
One broken fibre with 5-10ms permanent jitter where i have given the ISP last chance to fix it this week until i switch it to a working alternative (permanent red warning about 9-10ms framespikes), and one wireless which got broken down too this winter and fix will come with a fibre next month and it has a tiny download jitter but with occasional spikes up to 25ms randomly where i discovered that the frametime is equal to the actual jitter + render time as the frametime graph copies netgraph whenever the connection encounters worse times.

Yeah and bad news, this game really works only under LAN conditions and at really perfect networks with 0-1ms jitter.

This is a plotter of connection which somehow works and jitter is very rare (but everything what is not a straight line will cause frametime spike for me and the game behaves poor at this moment).



This is still OK as the jitter is 1-2ms (working PON fibre):



This is already unplayable with permanent 12ms framespike warning and jitter error (broken PON fibre):



the result:



The situation is similar no matter which computer i use, the only difference is that at computer with weaker CPU the spikes hit harder and the lag is more impacting the gameplay as it takes longer until it resynces and framerate drops lower.
In Valorant at the same connection it only shows some minor ping increase in netgraph and the game runs normally without any chance to notice.

[JimCarry]

congrats dont worry volvo dont care,why optimize the game when you have can make a new case ?