- Joined
- Jul 27, 2006
- Messages
- 9,484
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
Battlefield 2 - 256 Player Combat Test
- Thread starter Cerulean
- Start date
Mystic-Micro
Gawd
- Joined
- Aug 23, 2006
- Messages
- 759
1 Word. HOW!??!?!?!?!?!?
Guitarrasdeamor
2[H]4U
- Joined
- Apr 5, 2008
- Messages
- 3,622
This isn't MAG!
GushpinBob
2[H]4U
- Joined
- Dec 11, 2007
- Messages
- 2,721
It looks to be quite the lagfest, though I'm surprised that the guy's computer isn't in flames. I wonder if the same thing could be done in BF1942 (FUN!!!
).
Too bad BF2 isn't multithreaded. It would be awesome to have epic 500+ player rounds running on a Core i7.
edit: oh man, think of being the commander and having a militia of 200 troops.
).Too bad BF2 isn't multithreaded. It would be awesome to have epic 500+ player rounds running on a Core i7.
edit: oh man, think of being the commander and having a militia of 200 troops.
- Joined
- Jul 27, 2006
- Messages
- 9,484
Both the Host and Client machines. The host machine is my computer, and the client machine is my brother's computer.
On the host machine you would have the CPU doing two things:
1. Controlling each individual AI bot
2. Helping the video card with graphical rendering and gameplay effects/systematics
This was the primary FPS killer. If I had a clone of this machine and used the clone as a client machine, the client machine would have most likely have had a very playable to excellent (totally smooth) framerate. How can I say this? Just look at the client machine used in this experiment. All it had to do was receive the data from the host machine and render the graphics -- this is so much less work than the Host machine has to do; and as you can see, the framerate for the client machine wasn't too bad, and the client machine is pretty inferior to the specs of the host machine too.
Second, the lagging. Lagging as in the bots warping from point to point. The latency throughout the game did not exceed 5ms. This leads to the theory that either the network 100mbit pipe is maxed out or that the server is hitting it's maximum data transfer rate (ie. in Half-Life you can adjust the maximum amount of data the server is allowed to transfer in a second). On the Host machine, the bots did not "lag" like this.
Third, I have repeated this experiment on some larger maps (custom and official) and have found that performance on the host machine is significantly greater than on smaller maps where you have all/majority of the bots meeting at 1 single point in the map. Actually, the framerate on larger maps is significantly better because the bots are scattered, do not all meet in one single point (and otherwise would be a killer to the video card + CPU in combination, as you see in the video), and a lot of the bots aren't in the rendering field/point of view of the player. Remember the last part of the video where I was messing around as a commander? Before I zoomed in 3x, the framerate was perfect and excellent, and all bots were coordinating smoothly. This applies to this "larger map" technique. For this matter, if a map were large enough and AI waypoint/path trajectory generations were decent, my machine could actually handle a whole lot more bots than 256 (I am firmly confident in this claim).
Fourth, UAV would initiate a server crash, even with 128 players. I have a feeling that there is some sort of limit set on the UAV or that the UAV systemics weren't programmed to easily support some number of players between 64 and 128.
Fifth, there is a limit of like 9 squads. Around 72 to 75 bots were not assigned to any squads in the "256" test. Sucks to be you if you are the last one to join.
I could have used 255 bots, or even 256. It was pretty simply to find the maximum hardlimit number of player slots; 256 worked and 257 didn't work.
Very true. But where are you going to find 256 players, all having high-end machines and super good internet connections, and a large enough map to support all this? First guess would be a LAN party arena, but then just imagine the costs of running such a huge party.
I find all this impressive because from conclusions, the host machine (mine) could easily handle a lot more bots as a dedicated server, and that if everyone had an HD4850 and a matching processor to keep from bottlenecking the video card, it would actually be totally playable and a reality (technically it IS a reality, just that nobody actually takes advantage of this). But you also have to keep in mind that the engine wasn't necessarily coded to support 256 players, therefore lots of optimizations (that could exist) are not at their best. By this experiment, Battlefield 2 would make a very nice all-out-war MMO for LAN parties.
The only thing missing is the maps specially made for 128, 196, and 256 player combat.EDIT ::
Oh come on, this was nothing to my CPU. Keep in mind that the proper way to do this would be to run it as a dedicated server, not a listen server -- or where you host and play on the same machine. If BF2 were multithreaded, performance (framerate-wise) would definitely increase.
I forgot to mention, but some of the additional tests I ran with 254 bots had the bot difficulty increased to 1.0 (maxed difficulty). There was no noticeable performance loss either.
The Q9450 is one strong baby. I would imagine that the i7 could handle triple if not quadruple the number of bots I ran in this test (and actually a whole lot more than that too, since I firmly concluded that my processor could easily handle a bunch more bots).