Still not got to the bottom of why my ping goes high for a minute 40 mins into a race. What app does @RasmusP suggest I use to monitor network activity?
Sadly I never had to analyse network stuff so I have no experience about any tools
From a theoretical point of view, you'd need to find out where the issue lies exactly.
- router to server
- home network
- your PC to router
- inside your PC
- inside AC
My FritzBox has some priority settings, where I've set my PC to "real time".
So if I'm racing and my partner happens to want to stream on the TV, the iPad and her phone at the same time while synchronizing some 100GB cloud files on the laptop:
MY connection won't see any impact at all and I'll have to listen to complaints about slow download speeds and the videos not switching to HD quality
That would only be visible in your router, I think.
Again my FritzBox shows Downstream + Upstream in 5 seconds ticks.
Sadly not for every connected device specifically...
It might also be your PC doing something aside from AC every half an hour or so and you only see it when it happens during a race.
Like you got suggested: onedrive syncing, for example.
That would be visible in the Windows own resource monitor that can be opened from the performance tab in TaskManager.
However it doesn't have a history, which is super annoying!
There's the "perfmon" from Windows but that only tracks aaaalllll the things known to the universe WITHOUT telling you which application did what.
However I've found a little gitHub tool:
Bug fixes Fixed a crash Fixed the application saving data usage info under the wrong date.
github.com
After extracting and running the exe, Windows will tell you not to run it. Click on the "details" or whatever the button in the left bottom says in English and then run it anyway.
Most GitHub creators won't pay for the Windows signature stuff so you'll have to trust the creator if you wanna run it.
If you don't have .NET 6.xx, you'll get a little error message asking you if you'd like to download it.
I clicked "yes" and chrome opened a Microsoft page where I could download it.
Installed it and started the net monitor via the exe again.
And it worked
It seems to do the job for a first investigation!
Sadly the bandwidth/time graph is
not scrollable so you won't be able to find a potential spike, mark the time frame around that spike and filter for applications within that timeframe.
BUT at least it can LOG the usage for all processes!
I started CM, updated all the cars & tracks configs+vao files and then connected to a 32 cars Nordschleife Tourist server for a minute:
Maybe let this run during the next race and then check if anything looks unusual?
It was diagnosed at time as a netcode issue, whatever exactly that means I don't know.
Game netcode mostly does:
- data receiving (we send data to the server)
- estimation (server estimated trajectories based on the received data)
- data sending (server sends out the estimated result to all clients)
- tick rate (how often these 3 steps are done per second)
In AC's case it seems that the "estimation" and "data sending" will run infinitely with full collision physics no matter if "data receiving" still works like it should.
So if the car is going over a crest and one client stops sending data to the server, the "receiving" will stop but the "estimation" will continue, causing the car to fly off like a jet.
For whatever stupid reason, there's no filter / checking algorithm that puts the car into "ghost mode" when the receiving data makes no sense.
When you see a car rubberbanding, the "data receiving" part still gets data, but with gaps and way too much latency, causing the "estimation" constantly correcting.
Depending on how well these 3 steps are coded, it's "good netcode" or "bad netcode".
The tick rate changes the theoretical accuracy. Too low and the estimation becomes difficult, too high and it gets difficult to keep things synchronized.
