Increase the Limit of Simultaneous Enemies
- Thread starter Templarion
- Start date
well i played yesterday with kf, 75 specimens max on westlondon, absolutely no change in fps on my quadcore .....
Upvote
0
I quite understand what rallfo meant.
The problem isn't really with the clients, they can always buy better CPU and graphics card but its much more difficult with servers hardware.
30 players and 100 max runtime zombies really stresses the CPU of our server, and like rallfo said, i believe this is one of the reasons why there are aren't many 6+ servers.
I don't understand the unreal engine and the code as rallfo does but i believe squeezing some more efficiency in the AI code should be definitely possible.
The problem isn't really with the clients, they can always buy better CPU and graphics card but its much more difficult with servers hardware.
30 players and 100 max runtime zombies really stresses the CPU of our server, and like rallfo said, i believe this is one of the reasons why there are aren't many 6+ servers.
I don't understand the unreal engine and the code as rallfo does but i believe squeezing some more efficiency in the AI code should be definitely possible.
Last edited:
Upvote
0
Exactly, especially given that Tripwire chose to use a non-multithreaded game engine in 2009, our server has 7 cores just sat there doing sweet FA. And yes we had to obtain an 8-core Opteron as it's the only thing with the necessary clock speed server side to make 32 players / 108 zombies playable which quite frankly is ridiculous given the same machine that used to be part of a supercomputing grid now struggles to move a few zombies about.
The theme in this thread from the admins that have commented is universal and it is on the largest servers that an issue that affects all servers is most apparent.
Currently the cheap Celeron affairs within most clan/admin budgets sometimes struggle to run standard 6-player servers for this very same reason and so this is a very pertinent issue from the point of view of general gameserver market penetration.
Besides even if 6+ player servers are not officially sanctioned, myself and mastertheknife alone have seen many thousands of players through our collective doors who may simply turn around and return to the other zombie game they probably came from as a result of these performance issues.
Some Tripwire input is very much needed now, if only to assure us we aren't banging our collective heads against a vBulletin brick wall here...
The theme in this thread from the admins that have commented is universal and it is on the largest servers that an issue that affects all servers is most apparent.
Currently the cheap Celeron affairs within most clan/admin budgets sometimes struggle to run standard 6-player servers for this very same reason and so this is a very pertinent issue from the point of view of general gameserver market penetration.
Besides even if 6+ player servers are not officially sanctioned, myself and mastertheknife alone have seen many thousands of players through our collective doors who may simply turn around and return to the other zombie game they probably came from as a result of these performance issues.
Some Tripwire input is very much needed now, if only to assure us we aren't banging our collective heads against a vBulletin brick wall here...
Last edited:
Upvote
0
![[TW]Ramm-Jaeger](/data/avatars/s/0/2.jpg?1568638684){kind=avatar}
First off, let me say that anyone that claims their server's CPU is struggling on a standard 6 player game is just full of it. We've gotten feedback from server hosting companies running hundreds of KF servers as well as other game's servers. The feedback we've gotten is that the CPU usage of stock KF is actually less than that of L4d. With even low end CPUs you can run several instances of KF per core. So CPU usage for stock KF is not an issue for servers.
Regarding adding the ability to add more zeds per player, this is something that we are considering. There are a couple of issues here though to think about. One is gameplay balance. Adding more zeds per player throws off the gameplay balance and perks earning, and will require extensive beta testing at all different levels (32 per player, 50, etc) to ensure the game is still balanced and fun. I know someone will reply to this thread and say "we don't care if its balanced or fun" and well, sorry, but when you are in the business of selling games you have to care.
Regarding CPU usage for increased monster counts/player counts, the Unreal 2.5 is what it is. We're not going to invest massive man months (or years) worth of effort to take the engine multi core, nor are we going to rewrite Epic's base AI code from scratch (which is what would be required to optimize it more). Epic's AI code in 2.5 is pretty heavyweight. That is just a fact of using the engine.
We designed the base game to run well on a broad range of machines, and we accomplished this goal. We also gave the ability for people to go crazy in sandbox mode and run whatever they want.
So to sum up, we're looking at opening up some increases in zombie counts for the standard gametype, but don't have any sweeping changes planned to go crazy like 80, 100, 1000 zeds at once.
Regarding adding the ability to add more zeds per player, this is something that we are considering. There are a couple of issues here though to think about. One is gameplay balance. Adding more zeds per player throws off the gameplay balance and perks earning, and will require extensive beta testing at all different levels (32 per player, 50, etc) to ensure the game is still balanced and fun. I know someone will reply to this thread and say "we don't care if its balanced or fun" and well, sorry, but when you are in the business of selling games you have to care.
Regarding CPU usage for increased monster counts/player counts, the Unreal 2.5 is what it is. We're not going to invest massive man months (or years) worth of effort to take the engine multi core, nor are we going to rewrite Epic's base AI code from scratch (which is what would be required to optimize it more). Epic's AI code in 2.5 is pretty heavyweight. That is just a fact of using the engine.
We designed the base game to run well on a broad range of machines, and we accomplished this goal. We also gave the ability for people to go crazy in sandbox mode and run whatever they want.
So to sum up, we're looking at opening up some increases in zombie counts for the standard gametype, but don't have any sweeping changes planned to go crazy like 80, 100, 1000 zeds at once.
Upvote
0
Good to hear.
Thanks for replying. It's good to see devs commenting in idea threads.
Upvote
0
You chose to run a non-standard configuration on your server, so if you have performance issues, it's your own damn fault.
The game was not created to have more than six players at a time, from neither a game design nor a performance standpoint.
My server is running on a single-core, several year old machine in the other room running Slackware Linux and it does just fine with no performance issues at all.
Upvote
0