Recent Lag ... and what's being done.

[pooty]

Yeah if you set your netspeed > 10000 it will uncap the fps lock. At 10000 and under its capped at 85fps. After that if you increase netspeed you can get over that. There's chart somewhere with what netspeed uncaps what. I've set the max netspeed for the server at 13600, which can support 200 fps.

Looking here: https://docs.google.com/document/d/1Kni ... 2qgWc/edit
and here https://docs.google.com/document/d/1ahb ... BXhWA/edit
"Firstly, be aware that the client-side frame rate is capped for a reason: to reduce bandwidth to and from your computer. The higher your frame rate, the more times per second the game is updating your screen, which means more data. That said, only do this tweak if your internet connection is pretty solid. It will hurt you more than benefit you to enable this on something like a 32-man ONS pub, while it is great for something like 1on1 DM. "
Now that was written maybe a decade ago and well have 10-100x faster internet now as well as faster server hardware.... but still

Why 13600, it enables most to have 140 or so FPS.

BUT I wonder if we should just cap it at 10000? Why?
The game itself, having dug through some of the code, runs tick() function (think of it as the function that updates things like vehicles firing etc) runs a 100Hz (100 times a sec).

What's the point of having it update at say 140 frames per sec if the game maxes out at 100 updates a sec? This also matters on your monitor refresh rate ( I have one that support 144hz refresh rate, older ones don't update that fast, and the newer ones are faster). No point in having 180fps if your display doesn't support it...it won't redraw fast enough to matter.
More updates means more packets sent faster, greater chance for client packetloss. If everyone went netspeed 20000 we might swamp the engine (likely CPU, but not the network on the machine).

Prior to UTComp (which allows for default setting of netspeed over 10000), you had to manually set netspeed higher. I can't say I've ever noticed any difference at 85fps vs 140fps. Which makes me say are we making server work harder for really nothing.

Snarf beat me to it... fire weapons cause a client/local frame rate drop. I've experienced it, and I have plenty of Graphics power for this game (1080 card)

[captainsnarf]

13600 netspeed cap is acceptable. Please don't cap us at 10000. I did not buy 240hz monitors to run them at 85hz. I can certainly tell the difference.

[pooty]

Yeah I tried it a bit last night setting my own netspeed to 10000 and 85fps cap vs. 140fps. I could tell the difference for sure, especially on Masterbath.
So 13600 it is, I do think anything over 200 fps gets dicey at least according to that doc.

[pooty]

So I am still seeing lag in spots. And we are spiking up over 50% with 25+ players. Maybe its just me but felt way better without any Enhanced Netcode. I think we should try a week or so without it enabled. I also want to see if that's what causing these random crashes.

[pooty]

So I have my UT client set up to get 145fps. I get that easily locally and on a lightly loaded server.

12/09/2022 Config no ENC
Valarna 26 Players, bottomed out around 80-90 fps.
AratheiaComplex 24 Players, no sustained loss of FPS (only around the firebug and flame tank.. )

[pooty]

So I upped my client to 165 fps.
12/10/2022 Config no ENC, UTComp on (these are all lowest observed rates)
DJY 30 Players 160 fps
RP 24 P 160 fps
NVM 29 P 125 fps (DRaco maybe?)
Minus KS 24 P 85 fps (We all know this one is trouble)

[pooty]

So tonight, not huge player numbers and other than the normally laggy KitchenSink map, lag was no really an issue.

However, I noticed on the NFO stats, we were generally 35% or lower, BUT the overall machine was higher than normal 28-30% (across all cores). So I wonder if a bit of this lag is just noisy neighbor. I've seen where our game server goes 40-50%, and overall machine is lower like 20%.

Given costs of dedicated server, probably have to live with it...but if I see it too often I'll open a ticket with NFO.

[pooty]

So the last two days, the OLStats has been removed/off (this is the mutator that gathers the stats for the statsDB page -- not the balancer). Is it just me or has it felt like the lag was much, much better? Seemed like last night we had a pretty full server (26+) for almost 2 hours cpu just touched 48% overall much lower... the only map I really noticed lag was Minus-KS. Dria seemed good. I wonder if given that it records pretty much every kill, damage type etc...if the collect and writing the stats is contributing? Stats logs on Disk are 600K max depending -- given UT is single core keeping these stats might bog things.

That being said several people find them useful, so I can turn them back on. Not sure if its related but seems like the crashes went away too.

[infinitecat]

Some of the data collected has always perplexed me, like Special Events, Weapon/Suicide Specific Information, Killing Sprees, Weapon Accuracy, and Vehicle and Turret Specific Information. These burden data transfer, have no effect on the rankings and are of little educational value.

UT comes down to kills, frags, nodes built/destroyed/etc, and time it takes to get it all done.

[Canon Fodder]

Last night, the only map I had some lag was one of the Arctic Junkyards. It lasted about a minute. It was towards the end of the match. But that was it in about 2 hours of play.