While I have experienced what is being described in this thread I do not believe the problem to be hit-detection OR network latency.
After reading the following thread:
[url]http://forums.tripwireinteractive.com/showthread.php?t=68693&highlight=tickrate[/URL]
I can see the argument that sometimes it might be "cover" getting in the way somehow, but mostly I now mostly believe the problem has to do with the default server's MAXIMUM tickrate setting (server ROEngine.INI), the player's configured internet speed setting (player ROEngine.INI) and total player-count on the servers.
For instance.
We have tested the following on our server:
sf. Shoot First [Tactical] - TE & CD
24.77.33.94
SERVER HARDWARE:
Core 2 E8500 3.16Ghz, 8GB RAM, 3MBps upload bandwidth
SERVER CONFIG:
- Changed the default MAXIMUM (20) tickrated on the server to (40). Warning! This DOUBLES the CPU usage on the server.
[IpDrv.TcpNetDriver]
MaxInternetClientRate=20000 (Default = 10000)
NetServerMaxTickRate=40 (Default = 20)
- Changed the player count from 32 to 24 players just to be safe for testing.
CLIENT CONFIG:
- Changed our client "configured internet speed" settings in the local player ROEngine.INI's
[Engine.Player]
ConfiguredInternetSpeed=20000 (Default = 10000)
After almost a week of testing we believe that making these changes has made a
significant difference in what can be described as "bullet travel time", "hit markings/effects" and in general what has been described in this thread as "hit detection loss".
Feel free to join our server anytime to test this theory for yourselves. Also, while those of us members located in the same province in the server noticed no change until the server filled up some the members located in California (with a usual 110 ping to the server) noticed an immediate improvement.