Page 1 of 1
TAS Client - Lag Meter to Game Host
Posted: 06 Jun 2007, 20:36
by monohouse
I think this feature is needed don't you ?
...say average of 10 ? or better yet, UDP ping ?
Posted: 06 Jun 2007, 20:40
by Neddie
Connections fluctuate. Computer activity fluctuates. I don't know if this would be accurate or useful.
Posted: 06 Jun 2007, 20:46
by Lippy
Well at the moment it's not possible anymore according to AF:
http://spring.clan-sy.com/phpbb/viewtop ... c&start=86
Posted: 06 Jun 2007, 21:10
by AF
The process of pinging in itself would cause a huge connection lag for all users as the number of pings would be n!
So 200 users online would generate 200x199x198x197x196... = 7.8865786736479050355236321393219e+374
Each user would recieve at least 199 pings.
The only option would be a central server. At which point you then have to make sure its a good server and then there're the issues neddierow pointed out.
Posted: 06 Jun 2007, 21:24
by Kloot
n users gives n(n - 1) / 2 pairs of people pinging each other, not n!
Posted: 06 Jun 2007, 21:26
by Peet
AF wrote:The process of pinging in itself would cause a huge connection lag for all users as the number of pings would be n!
So 200 users online would generate 200x199x198x197x196... = 7.8865786736479050355236321393219e+374
Each user would recieve at least 199 pings.
The only option would be a central server. At which point you then have to make sure its a good server and then there're the issues neddierow pointed out.
All that is unnecessary...all that would be required is for each client in a battle to ping the host with a UDP packet on the port that the game would be hosted on. This would test both latency and whether NAT traversal was done properly.
Posted: 06 Jun 2007, 22:27
by PicassoCT
neddiedrow wrote:Connections fluctuate. Computer activity fluctuates. I don't know if this would be accurate or useful.
Me think it should be possible to simulate the Traffic and Endgame Cpu & Ram Usage ... depending on Mapsize and Playernumber..
Posted: 06 Jun 2007, 23:08
by AF
Its never that simple peet.
Posted: 06 Jun 2007, 23:26
by rattle
AF wrote:The only option would be a central server. At which point you then have to make sure its a good server and then there're the issues neddierow pointed out.
Yeah, like what is a good server... there's a reason why there are western and eastern US, asian and also european servers available for online play for a lot of games.
Posted: 07 Jun 2007, 00:26
by Tobi
You can already test your ping to the central server (/ping in lobby, or heck you could even do ping
http://www.google.com on the commandline). Not that this ping has any useful correlation with your in game ping though...
Peet seems right though, theres no need to make every client ping every client (plus the scalability problems..), simply pinging the battle host when in battle should be good enough.
Posted: 07 Jun 2007, 00:47
by monohouse
All that is unnecessary...all that would be required is for each client in a battle to ping the host with a UDP packet on the port that the game would be hosted on. This would test both latency and whether NAT traversal was done properly.
yes this is a good fix to the subject - indeed I mean latency to the host, not the main server, the main server is useless in determining game lag/latency.
also, fluctuations will be refrected in the average, ping frequency is irrelevant as it's the players in the game and the host are the only thing affecting that.
the main server is also of corse irrelevant, only the lag to the host of the current game, so like an in-battle lag meter to the host only.
this should put an end to players that lag in games and don't know why, before the game is started the lag will already be known.
yhea, "pinging the battle host when in battle should be good enough" only battle host, the rest is irrelevant
Posted: 10 Jun 2007, 01:11
by Complicated
Just give spring more chance to lag then k?