Servers are a mess for me

[GoonFartz8888]

After playing l4d2 in a while out of nowhere official servers near me and even servers i made (official, best, local etc) have been awful. I use no mods to test it even local by myself and its horrible. I legit tried every server type and 40-100 ping and 200-400 ping when hoards. Even with no ping its awful. I've noticed once someone joins my server it does to good to horrible. I noticed my internet has got a tiny bit slower but it shouldn't be unplayable for me. I have a download speed or something of 370mb's and 5G compatible router and my PC is a legit beast. If this is a problem for others I doubt valve will do anything so help me please.

[GoonFartz8888]

its all servers

[Tool.]

left4dead is in general like this compared to other source games, because it's locked to a 33 tickrate. this means you only get upto 33 game updates per second on normal servers, and it gets worse..
the official servers run so many instances of game servers that during hordes you can expect this number to go down to 3-5 ticks per second as their machines struggle to allocate resources, which is going to feel really laggy.

all this means that in the best case scenario, you'll have a latency of 30ms + distance to server, as the game has to wait 30ms anyways for the next tick with gameinfo.. in worst case scenario, 3 ticks/sec is a latency of 333ms + distance to the server.. :(

you can see all of this by enabling your developer console (right click in steam -> properties -> add -toggleconsole to launch options) and typing net_graph 2 in it. the numbers you wanna look out for will be in the bottom right corner of your screen when playing on a server.

this is why most modded servers (including mine) utilize a tickrate unlocker so that the server and clients can exchange more info faster (usually 100-tick, mine are upto 256 ticks/sec to help out players with bad wifi)

if you wanna give it a go, here's a link to the tickrate unlocker's github:
https://github.com/accelerator74/Tickrate-Enabler/releases/tag/build
beware that this mod only helps serverside when i.e. hosting a local game, it does not give magical performance when playing on someone else's server.

[Blaquicat]

Originally posted by Tool.:

all this means that in the best case scenario, you'll have a latency of 30ms + distance to server, as the game has to wait 30ms anyways for the next tick with gameinfo.. in worst case scenario, 3 ticks/sec is a latency of 333ms + distance to the server.. :(

All true, except this part.

You can have 1 tick and 5ping with no big deal, the real problem is the tick also limits how much processing the server can do..
https://steamcommunity.com/sharedfiles/filedetails/?id=1151702553
https://steamcommunity.com/sharedfiles/filedetails/?id=1151702642

So.. while server thread usage (cpu usage on stats) is less than 100% the game was smooth with 5 ping, after a while the cpu usage reaches 100% and then the game starts to lag like hell..
On the second pic i was using 125tick, but if you notice the net graph the server was struggling to keep 50updates/s..
Tested on a virtual machine with linux on it.

[Tool.]

Originally posted by Blaquicat:

You can have 1 tick and 5ping with no big deal

Check this out, I conducted a little test using my neighbour's wifi for a less local result.

91 ticks/sec, 27ms average latency: https://i.imgur.com/p1AGC0S.png
5 ticks/sec, 268ms average latency: https://i.imgur.com/w6EUuf7.png

You can test this yourself by using Sourcemod's sm_cvar setting, and restricting fps_max. I did sm_cvar fps_max 5, which restricts my server to just 5 ticks/sec, and the latency completely dies despite nothing else going on.

Granted net_graph shows each individual package is still fast (~35ms all the time), the average latency shown in the playerlist is what represents how I will really feel the gameplay, and at 5 ticks/sec I can't even stand still and jump without it feeling laggy and weird.

If I remember right it has to do with the lerp time.. a big lerp time will allow a big difference between player and server info, which can help with stuff like movement.. but also means that you can shoot players without aiming at them. Most prefer a low lerp time for the accuracy, I have mine set to 100ms which I think is considered average or high.. still, if my latency goes above my lerp time (100ms) the server will constantly try to correct me, making my game also feel laggy.

[Blaquicat]

Originally posted by Tool.:

Originally posted by Blaquicat:

You can have 1 tick and 5ping with no big deal

Check this out, I conducted a little test using my neighbour's wifi for a less local result.

91 ticks/sec, 27ms average latency: https://i.imgur.com/p1AGC0S.png
5 ticks/sec, 268ms average latency: https://i.imgur.com/w6EUuf7.png

You can test this yourself by using Sourcemod's sm_cvar setting, and restricting fps_max. I did sm_cvar fps_max 5, which restricts my server to just 5 ticks/sec, and the latency completely dies despite nothing else going on.

Granted net_graph shows each individual package is still fast (~35ms all the time), the average latency shown in the playerlist is what represents how I will really feel the gameplay, and at 5 ticks/sec I can't even stand still and jump without it feeling laggy and weird.

If I remember right it has to do with the lerp time.. a big lerp time will allow a big difference between player and server info, which can help with stuff like movement.. but also means that you can shoot players without aiming at them. Most prefer a low lerp time for the accuracy, I have mine set to 100ms which I think is considered average or high.. still, if my latency goes above my lerp time (100ms) the server will constantly try to correct me, making my game also feel laggy.

Well, you have a 100(?) tick server, running at 5fps, obviously its not going work, if you want to test that properly, run the server with 5 tick, and leave fps alone.

Tickrate is how many game stages the server produces per second, the default fps is not 33, but 300, you can have two different frames playing along on the same tick, thats why the default updaterate is 60, much more than the 33 default tickrate.

"A multiplayer client will typically need to render three or more frames with each server update it receives (assuming 60fps+ and cl_updaterate 20)."
Tickrate > Server update > Frames

[Tool.]

Originally posted by Blaquicat:

Well, you have a 100(?) tick server, running at 5fps, obviously its not going work, if you want to test that properly, run the server with 5 tick, and leave fps alone.

Tickrate is how many game stages the server produces per second, the default fps is not 33, but 300, you can have two different frames playing along on the same tick, thats why the default updaterate is 60, much more than the 33 default tickrate.

"A multiplayer client will typically need to render three or more frames with each server update it receives (assuming 60fps+ and cl_updaterate 20)."
Tickrate > Server update > Frames

Restricting the dedicated server's fps was the easiest way to restrict the server's tickrate without restarting it with new launch options.. :(

It was on a dedicated server so it doesn't generate "frames" per seconds anyways as it doesn't have a graphical output besides just console text, so 1 fps = 1 tick sent to clients basically.

In my screenshots you can validate that setting fps_max 5 results in exactly a 5.0 tick/sec inbound speed.

The server's fps can be higher than the tickrate (i.e. bad client connection) but the tickrate can never be higher than the server's fps, which makes it an effective way of testing what happens when the server doesn't send enough ticks per second.

By default the fps_max is set to whatever default refresh rate the operation system runs at, so usually 60hz or 60. This is perfect for a game like CSS that doesn't limit tickrate, and where the default tickrate is also 60 I think?.. but on Left4Dead with the 33tick/sec restriction, almost half of those server fps are wasted by only listening to clients and not sending out new info.

This means that even if you do like some CS 1.6 servers do, and modify a linux kernel to run at smth like 1000 fps.. Left4Dead servers would perform as bad as always until they got unlocked.

[Blaquicat]

Originally posted by Tool.:

Originally posted by Blaquicat:

Well, you have a 100(?) tick server, running at 5fps, obviously its not going work, if you want to test that properly, run the server with 5 tick, and leave fps alone.

Tickrate is how many game stages the server produces per second, the default fps is not 33, but 300, you can have two different frames playing along on the same tick, thats why the default updaterate is 60, much more than the 33 default tickrate.

"A multiplayer client will typically need to render three or more frames with each server update it receives (assuming 60fps+ and cl_updaterate 20)."
Tickrate > Server update > Frames

Restricting the dedicated server's fps was the easiest way to restrict the server's tickrate without restarting it with new launch options.. :(

It was on a dedicated server so it doesn't generate "frames" per seconds anyways as it doesn't have a graphical output besides just console text, so 1 fps = 1 tick sent to clients basically.

In my screenshots you can validate that setting fps_max 5 results in exactly a 5.0 tick/sec inbound speed.

The server's fps can be higher than the tickrate (i.e. bad client connection) but the tickrate can never be higher than the server's fps, which makes it an effective way of testing what happens when the server doesn't send enough ticks per second.

By default the fps_max is set to whatever default refresh rate the operation system runs at, so usually 60hz or 60. This is perfect for a game like CSS that doesn't limit tickrate, and where the default tickrate is also 60 I think?.. but on Left4Dead with the 33tick/sec restriction, almost half of those server fps are wasted by only listening to clients and not sending out new info.

This means that even if you do like some CS 1.6 servers do, and modify a linux kernel to run at smth like 1000 fps.. Left4Dead servers would perform as bad as always until they got unlocked.

Nop, you are just choking the server, thats why you have so many latency.
https://steamcommunity.com/sharedfiles/filedetails/?id=2804949805

The server wont allow less than 30 tick, however you can "emulate" less than that using less updaterate, as show on my screenshoot (which shows both cmdrate and update at 5 and 10, the low as it can get)
Lerp was automatically set at 200, cant be set lower either, because as mentioned above, the game needs to render few frames for each update, with less updates it needs more time to interpolate those extra frames..
Anyway, it does not mean lag, as you can see.

[Senpaija]

Hordes can make the ping drop, but it's usually because the maximum amount at one time has been raised. Perhaps your device is overworking despite the vanilla amount.

[Blaquicat]

Originally posted by Senpaija:

Hordes can make the ping drop, but it's usually because the maximum amount at one time has been raised. Perhaps your device is overworking despite the vanilla amount.

The server can do only so much per tick, meaning if it gets overwhelmed, the latency will start to grow as the server gets takes to much time to process things.

Originally posted by Blaquicat:

https://steamcommunity.com/sharedfiles/filedetails/?id=1151702553
https://steamcommunity.com/sharedfiles/filedetails/?id=1151702642

This custom map for example, its unplayable on 33 tick, no matter what you do, many things happening at once, highly overwhelm the server tickrate.
Even the "good" ping pic, you can notice the server struggling to keep 50tick. (125 tick set on server)

[GoonFartz8888]

I just wanna say thanks for all the help and Scootertrix I hope your dad gets drunk enough one day to beat you into a coma dips***

[GoonFartz8888]

Originally posted by ScooterTrix:

does it look like anyone cares? get some more hours and change that profile pic, nasty lathargic child grooming jackass

Dont put your own insecurities on me buddy

[AustrAlien2010]

I cannot help you much, because the last time I tried helping anyone with their video game on the Steam forums, I literally ended up being reported for that.

I recommend playing single player, because then you have no latency at all.