Torque networking
by Nathan · in General Discussion · 01/16/2006 (5:57 pm) · 8 replies
I've heard that torques networking code is good, but how good is it really? Would someone with a 24k connection be able to play multiplayer with torque?
#2
Yes, Torque's networking is "good". It is "really good". Good enough to be used in a number of commercial titles and splitnered off into a network layer for intergration into other engines. If your ghosting metrics are low, someone with a low-end connection can play just fine. If the amount of data you need to send, check, validate, etc is high, then it is necessarily ramped higher.
01/17/2006 (7:30 am)
Any network model depends on how much information you need to propagate over the network. Asking how good is it and thengiving a generic statistic without any idea of the amount of data you need to keep track of makes no practical sense since it is an unanswerable question except for in the most generic terms.Yes, Torque's networking is "good". It is "really good". Good enough to be used in a number of commercial titles and splitnered off into a network layer for intergration into other engines. If your ghosting metrics are low, someone with a low-end connection can play just fine. If the amount of data you need to send, check, validate, etc is high, then it is necessarily ramped higher.
#3
Ive worked on a lot of modern game engines and none of them are a fraction as bandwidth
friendly as Torque, not even close.
I can get tons of bots running around, a bunch of players(litterally like 500 players, 500 bots)
and it still works with a 56k modem. Good luck using any comparible engine(quake,doom, unreal, etc).
The networking in torque is very compressed, tight, and miminal. The prediction/extrapolation is still
better then what I see in other engines.
Its almost as if today programmers are spoiled with dsl and cable modems and so write slow, fat, heavy
networkworking code.
But when torque was originally written back in the tribes 1 days, modems was really all you had. Now we
can benefit by that limitation they once had.
Sure it does come with some limits, but you can overcome them pretty easily (like the number of attributes on a player).
In my last few tests awhile back I had 1000 bots and 800 players running around a map and still only used minimum bandwidth.
contraints are a wonderful thing :) i think people forget there are still a lot of people on dial up.
01/17/2006 (8:00 am)
I would go a step further and say there networking, specifically for fps, its remarkable.Ive worked on a lot of modern game engines and none of them are a fraction as bandwidth
friendly as Torque, not even close.
I can get tons of bots running around, a bunch of players(litterally like 500 players, 500 bots)
and it still works with a 56k modem. Good luck using any comparible engine(quake,doom, unreal, etc).
The networking in torque is very compressed, tight, and miminal. The prediction/extrapolation is still
better then what I see in other engines.
Its almost as if today programmers are spoiled with dsl and cable modems and so write slow, fat, heavy
networkworking code.
But when torque was originally written back in the tribes 1 days, modems was really all you had. Now we
can benefit by that limitation they once had.
Sure it does come with some limits, but you can overcome them pretty easily (like the number of attributes on a player).
In my last few tests awhile back I had 1000 bots and 800 players running around a map and still only used minimum bandwidth.
contraints are a wonderful thing :) i think people forget there are still a lot of people on dial up.
#4
01/17/2006 (8:44 am)
Well, I tried think tanks and it didn't work at all on my 24k connection.
#5
01/17/2006 (9:04 am)
It worked well on my "not-quite" 56k connection that I get for free through the university. Lore had a little more problem, but it also has more going on. ZAP!, using TNL worked very well.
#6
well just because the networking is great, doesnt mean people build games with it and keep the lean and mean nature of the networking.
I know my tests worked with thousands of units running around, but I didnt touch the actual networking layer, I used with as is.
once you start to modify it you have to be very careful, its really easy to add something that greatly increases the bandwidth required.
the other thing is if you 24k connection has a lot of packet loss nothing is going to work well.
01/17/2006 (12:26 pm)
Quote:
Well, I tried think tanks and it didn't work at all on my 24k connection.
well just because the networking is great, doesnt mean people build games with it and keep the lean and mean nature of the networking.
I know my tests worked with thousands of units running around, but I didnt touch the actual networking layer, I used with as is.
once you start to modify it you have to be very careful, its really easy to add something that greatly increases the bandwidth required.
the other thing is if you 24k connection has a lot of packet loss nothing is going to work well.
#7
I've stayed out of the rest of the discussion, because David and TheMartian are making excellent points, and nothing more needed to be said really!
01/17/2006 (12:52 pm)
Just for reference, ThinkTanks actually does use the networking layer properly. 24k modem really is particularly low however, even for Torque: it was developed with roughly a min of 48k modem speeds, and anything less is going to be less than optimal.I've stayed out of the rest of the discussion, because David and TheMartian are making excellent points, and nothing more needed to be said really!
#8
01/17/2006 (1:59 pm)
Ok Stephen, but can it produce 1.21 Gigawatts? 
Torque Owner Mark
Default Studio Name
But as for connection speeds go, really, that's more dependant on the connection I'd have to say..