Trying to figure out my best option here regarding playing multiplayer online with some buddies. I'll outline my issue(s), and explain my proposed solution, however, I'm still having a hard time wrapping my head around this. Ok, so here's the issue. Whenever I play this game (or project cars 2), if I'm hosting a P2P session, they start to experience extreme rubberbanding of AI cars on the track. In particular, the closer they get in physical proximity to ME, the worse the AI seems to perform (AI randomly heading off the track, smashing into each other, jumping around, etc). When I hear people complaining about rubberbanding, the first thing I think of is latency. I've done a number of tests on my latency in general, however, nothing looks out of sorts - usually ranging from 40-65ms ping time. However, I AM on a wireless connection in the room I run my rig on, so I figured, maybe that's part of the issue. So, here was my idea to solve the issue and try to stabilize the connection between everyone. I stood up a dedicated server on another machine on my network, that's connected via wire directly to my router (500MB up/500MB down. The hope being, that the game could be "hosted" on the server, and everyone could connect to the server, rather than making a P2P connection, hopefully keeping latency lower and connection quality higher. However... I noticed something a moment ago when testing this all out, which is why I'm hoping for some clarification. I noticed that when I connected to the server (on my own) using the "-selectDS" method, that the AI ping times were all showing as "0", which is where my confusion is. If the game session is being "hosted" on a "dedicated server", wouldn't the dedicated server be hosting the AI as well? I guess my question is brought on by the fact that when a ping says "0", that usually means it's running locally (on my machine), rather than on the server itself. I guess I'm having a hard time figuring out what the benefit is of having an AMS2 dedicated server from a performance perspective. Does my inquiry make sense? Can anyone shed some light on this?
If you are the host, AI is still AI’ing on your local machine. The DS is a rebroadcaster, meaning all the networking happens on it, but it isn’t the session host. Only the host is. In other words, it’s normal that the AI shows 0ms for you. You are still hosting the AI and sending the networking info to the DS for it to rebroadcast.
Interesting. Ok, is there a performance advantage to using a rebroadcaster/dedicated server? How much overhead does it actually alleviate as opposed to a P2P session?
Yes but it is negligible. The advantage comes in the shape of the basic admin tools. You can access the admin panel to kick people, you can add people’s steamids to a blacklist and dependingg on how you set things up (in this case without allowing game to control lobby/-selectds), you can set up a rotation configuration or even change lobby settings in the panel (but unless you give others remote access to your admin panel you’d always need to be present). It isn’t optimal but it will see improvements this year still (i hope).
So, I was able to do a little bit of testing today with regards to connection quality and rubber-banding. While my gaming rig is on wireless, the dedicated server is wired in to the router. When using the "-selectDS" method, it cleared up ALL of the rubber-banding and other weird symptoms.
