Re:Deterministic Lockstep for Networked Games
by gafferongames on Saturday December 27, 2014 @09:17PM
(#8804890)
This is not always true. Some amount of packet loss exists over WiFi and 3G and in this case this packet loss is not indicative of congestion.
[ Parent
| on HN ]
Re:Deterministic Lockstep for Networked Games
by dllthomas on Saturday December 27, 2014 @09:33PM
(#8804930)
Even if there are spurious signals from other sources, it's still true that packet loss is how the internet signals congestion (absent ECN, which my understanding is basically undeployed on the internet proper).
Re:Deterministic Lockstep for Networked Games
by gafferongames on Sunday December 28, 2014 @12:41PM
(#8806797)
In the presence of packet loss that does not signal congestion, treating it as if it is congestion provides a lower quality result. http://1024monkeys.wordpress.com/2014/04/01/game-servers-udp...
Re:Deterministic Lockstep for Networked Games
by dllthomas on Sunday December 28, 2014 @03:48PM
(#8807415)
I don't disagree, but that converse is typically even more true. The strong defense of your approach here is "it's a tiny amount of data, period" (which - paraphrasing - you've said elsewhere). I am 100% behind the notion that one should be careful when responding to packet loss with increased traffic.
Re:Deterministic Lockstep for Networked Games
by gafferongames on Sunday December 28, 2014 @04:29PM
(#8807547)
Yes. With great power comes great responsibility.
Re:Deterministic Lockstep for Networked Games
by gafferongames on Friday January 09, 2015 @03:33PM
(#8864327)
This looks promising: http://modong.github.io/pcc-page/