How much is will it take? (to bog system down)
by Charlie Higdon · in General Discussion · 01/15/2009 (8:42 pm) · 3 replies
I'm not much of a game creator, and reading that TGE (which I bought less than a year ago) is going to be dropped, I may not have the chance.
I'm wondering what is the limit on placing bots, vehicles, buildings, players, etc, before a game made will be unplayable.
For example, take one 10 story building with all the doors, furniture, windows, props and such. That doesn't sound too bad. Make a small neighborhood of those and it starts to add up. Introduce players, vehicles, and bots, and I imagine it could get bad quick.
I know part of it depends on the amount of polygons used in objects (buildings, player models, etc). I was just wondering if it's entirely possible to construct an entire small town with roads, police station, fire dept, city hall, some moderate sized 5-10 story buildings, cars all about, that kind of stuff...without the game becoming unplayable.
I'm wondering what is the limit on placing bots, vehicles, buildings, players, etc, before a game made will be unplayable.
For example, take one 10 story building with all the doors, furniture, windows, props and such. That doesn't sound too bad. Make a small neighborhood of those and it starts to add up. Introduce players, vehicles, and bots, and I imagine it could get bad quick.
I know part of it depends on the amount of polygons used in objects (buildings, player models, etc). I was just wondering if it's entirely possible to construct an entire small town with roads, police station, fire dept, city hall, some moderate sized 5-10 story buildings, cars all about, that kind of stuff...without the game becoming unplayable.
About the author
#2
01/15/2009 (11:26 pm)
And don't forget about the contributing factor of the hardware that you run it on. Your test map may run fine as you describe it on one computer, but not another, or it could possibly be even better. This is one of those questions that gets asked a lot. The only safe answer is: Test, test, tweak, and test some more. Results will vary according to every factor already mentioned.Quote:I'm not much of a game creator, and reading that TGE (which I bought less than a year ago) is going to be dropped, I may not have the chance.Don't sell yourself short. Learn as much as you can, set some goals and work towards them. But have fun doing it. Do as much as can and if you find yourself wanting or needing that extra graphical oomph or increased performance curve you can always update to TGEa. Even though GG may drop TGE themselves doesn't mean you have to, it just wont be updated anymore (it hasn't been updated in over a year and a half anyway). Keep in mind that you've already paid for your license and they can't make you not use it -- you can still make as many games as you wont to so long as you don't abuse or exceed the EULA requirements. I doubt that the community for TGE is going away anytime soon either. Someone will probably be around to answer questions, submit resources, or write fixes for TGE for a while yet, and that was always it's greatest value to be found.
#3
Object creation and destruction appears to use a lot of network bandwidth (master-server bandwidth in the case of single player games). So, this can be tweaked to some extent by upping the packet size/rate - but keep in mind this will tax the server more. In a single player game this means more CPU time being utilized.
I think I was finally able to get about 100 bots shooting 1 projectile each second without too many problems. My personal solution to this problem was rewriting the projectile class to be 'lighter' as well investigating ways to compress or optimize the packetizing.
To make a small town like that you might consider limiting the "think" rate of objects once they get farther away from the player. You can remove pathfinding and jump things from place to place. Hell, just about all the optimizations people invented for Neverwinter Nights MMO Modules would apply here ;)
Some other ideas are limiting the ghosting distance (I saw a few resources on this) and making judicious use of Level of Detail in your models. Just a few suggestions. Good luck with things.
01/17/2009 (8:43 pm)
The biggest limiting factor in Torque for me, is the number of object creations and destructions per second. For example, having 100 units walking about is manageable. But with the default network settings, if just 30 of my bots tried to shoot a projectile you'd see maybe 10 projectiles fire, then 10 more appear halfway through the firing sequence, and the last 10 would not appear. Object creation and destruction appears to use a lot of network bandwidth (master-server bandwidth in the case of single player games). So, this can be tweaked to some extent by upping the packet size/rate - but keep in mind this will tax the server more. In a single player game this means more CPU time being utilized.
I think I was finally able to get about 100 bots shooting 1 projectile each second without too many problems. My personal solution to this problem was rewriting the projectile class to be 'lighter' as well investigating ways to compress or optimize the packetizing.
To make a small town like that you might consider limiting the "think" rate of objects once they get farther away from the player. You can remove pathfinding and jump things from place to place. Hell, just about all the optimizations people invented for Neverwinter Nights MMO Modules would apply here ;)
Some other ideas are limiting the ghosting distance (I saw a few resources on this) and making judicious use of Level of Detail in your models. Just a few suggestions. Good luck with things.
Associate Jaimi McEntire
King of Flapjacks