Post
by joseph_from_pilsen » 23 Dec 2023, 13:15
Great job, yes its normal that fps is stable 340then, at 280hz i get 260fps as the drivers do the limit by themselves a bit below refresh rate. You dont need to set any limit manually, the driver overrides it by its own pre-set value. Interesting fact: if you start the game with -vulkan, the driver limit is not being applied and you must use your own or use non-limit values (like vsync fast). But the game with -vulkan runs always a bit worse (at least at nvidia cards).
It also confirms that ULMB/vsync is a good trade-off between responsibility and visibility and motion persistence, for scanning environment for opponents is still critical to have best motion clarity, best persistence and still low input lag. Its nice to have 5ms input lag instead of 8ms but if you dont spot opponent as soon as it appears on your monitor or the game is feeling like 60hz slideshow any time you move your mouse cursor, its a bit useless... I tested all the settings (i have 280hz monitor), vsync off is a big no in cs2, totally unsmooth like 60hz, and ulmb is a bit discutable, OFF is more crisp when the picture is static or when walking slowly, ON is much better when looking anywhere than at my cursor as the background doesnt unsharp + whenever i move faster. The slight delay doesnt matter, the game servers are so trashy that even with ULMB2 is the client ahead from server A LOT (counter step at client = still moving at server etc.).
It confirms that cs2 has low input lag and the input lag is not the reason why aim and game itself is feeling so often so much off, the reason is OUTPUT lag which is up to 60ms and you cant affect it by hardware, is server sided (a delay between input and any effect on opponents, not environment - if you shoot an enemy, there is a delay between input and effect (hit animation), so multiple kills are a pain in ass as you waste time with output lag (you cant switch to another ttarget before you see whether you have got the first one or not).
Minimal achieavable input lag in csgo was around 11ms and it wasnt possible to reduce it by any ways as there was a fixed input lag done by obsolete game engine, so input lag in cs2 is up to 3 times lower and maybe with NASA hardware and benq 2586q (540hz new version) we will see even slightly better results. So oldschool csgo players are used to play with 11ms input lag nevertheless, valorant has AFAIK 7ms. So even with vsync on + ULMB its possible to transfer very easy.
Again, with single digit ms input lag, the main issue and bottleneck is the server, sadly cs2 client performance is affected by server (same like it was in csgo). When server variance is high (and its alwayss high, valve only obfuscated this value to hide it from players as they are crooked and fraudulent company), the input lag plays no role, the aim/movement will suck as the server (bad) response and performance behaves like server sided input lag (dont mess it with output lag, its another thing, output lag doesnt have impact on your input, but server variance and performance does have, it simply fu..ks up your movement and aim).
And last, important notice, STABLE input lag added always matters less than any jittery value. With 11ms vs 8ms vs 5ms stable input lag you will adapt during few minutes and get good results at any value (unless its too high, ofc.). However any jitter will cause you headache, sadly this is cs2 case - at valve official servers its unstable server performance (= stuttery response from server) and unstable networking - mostly at valve official servers = > packetloss => some packets are being dropped and re-sent => choppy gameplay. Unfortunatelly, cs2 is heavy on networking and CPU performance generally (not only on client but server side) so even faceit servers are not smooth, with cq_netgraph 1 (try also value 3) you will never see a permanent empty red straight line. Whatever else you will see in netgraph = issues. And i dont remember many servers with permanent red straight line in netgraph, yes they do exist, no they are not common.