How do I build a RW compatible Torque Engine?
by Bryan Turner · in General Discussion · 03/25/2002 (4:30 pm) · 5 replies
Hello,
I'm tring to get some networking statistics from a running game, but no one is playing the 'fps' mod right now, so I figgured I'd just start up my engine with the RW mod and catch the action there.
Unfortunately it crashes before even displaying the logo screen! I'm using Release_1_1_1, what diffs do I need to pick up to be compatable with the RW mod?
Thanks!
--Bryan
I'm tring to get some networking statistics from a running game, but no one is playing the 'fps' mod right now, so I figgured I'd just start up my engine with the RW mod and catch the action there.
Unfortunately it crashes before even displaying the logo screen! I'm using Release_1_1_1, what diffs do I need to pick up to be compatable with the RW mod?
Thanks!
--Bryan
#2
Thanks! Works great.
Now if I can just find a server with > 3 people playing. :)
I'm interested in gathering data to look at bandwidth optimization strategies. The data I just collected ran like this:
5 Player Game:
Avg Packet size: 172.3 bytes
Avg #Ghosts: 10.0
3 Player Game (after adding some more stats):
Avg Packet Size: 45.2 bytes
Avg #Ghosts: 3.4
Avg #PlayerGhost Updates: 3.1
Avg Player Update Size: 12.6 bytes
Looks like optimizing the player update would make the biggest difference. It appears that the player update dumps 12-18 bytes for just the position, velocity and view orientation vectors (which coincides nicely with the average player update size).
--Bryan
03/25/2002 (6:19 pm)
Matt,Thanks! Works great.
Now if I can just find a server with > 3 people playing. :)
I'm interested in gathering data to look at bandwidth optimization strategies. The data I just collected ran like this:
5 Player Game:
Avg Packet size: 172.3 bytes
Avg #Ghosts: 10.0
3 Player Game (after adding some more stats):
Avg Packet Size: 45.2 bytes
Avg #Ghosts: 3.4
Avg #PlayerGhost Updates: 3.1
Avg Player Update Size: 12.6 bytes
Looks like optimizing the player update would make the biggest difference. It appears that the player update dumps 12-18 bytes for just the position, velocity and view orientation vectors (which coincides nicely with the average player update size).
--Bryan
#3
How did you collect it? I haven't looked into any of the networking debug features of the engine, but I really see how this can be useful when trying to allieviate latency issues.
03/28/2002 (9:31 am)
Wow, very cool information!How did you collect it? I haven't looked into any of the networking debug features of the engine, but I really see how this can be useful when trying to allieviate latency issues.
#4
03/28/2002 (11:14 am)
Yeah, were you using a protocol analyser or are you doing something in the engine to dump these stats?
#5
Then use notepad to delete the non-protocol data, and import it as a CSV into Excel for post-processing.
I have a recording of an entire game which maps every player's position on the landscape (looks neat in an Excel XY-Plot).
There may be some built-in protocol trace functions, but I didn't see any obvious ones (only error-tracking, not protocol sniffing).
--Bryan
03/29/2002 (11:15 am)
Actually, I dropped to the old-skool printf() method. Each packet that is recieved I store some info on as it gets processed (size, # ghosts, size of each ghost update, etc..) then printf() the data with commas between it.Then use notepad to delete the non-protocol data, and import it as a CSV into Excel for post-processing.
I have a recording of an entire game which maps every player's position on the landscape (looks neat in an Excel XY-Plot).
There may be some built-in protocol trace functions, but I didn't see any obvious ones (only error-tracking, not protocol sniffing).
--Bryan
Torque Owner Matt W
You're using an older version of source, so get the HEAD version and try again. That should fix that problem.