RaGEZONER
- Joined
- Sep 25, 2009
- Messages
- 637
- Reaction score
- 398
Ive always wondered about why the server uses the cpu when nothing is happening.
This is what I figured out so far. (Sry for the mess, copy pasted notes bellow)
The cause is this,
WorldTimer=1:100,2:150,3:3000,6:60000,8:2000,9:10000,13:150,14:2000,15:30000,16:200,20:200,21:200
Setting them to,
WorldTimer=1:1000,2:1500,3:3000,6:60000,8:2000,9:10000,13:1500,14:2000,15:30000,16:2000,20:2000,21:2000
will reduce the cpu usage by quite a lot, since the binary updates/check's tick will now be done every 2-3 seconds instead of every 150 milliseconds.
There is a downside to raising the time as you might imagine it.
I will try to document these, if any one is interested. :
Notes:
------------------------------------------
ReactorTimeOut=100 smaller value, like 1 for example will cause near 100% cpu usage
ProcessDataLimit=1024, Limits how much data it can process at once, thus lower cpu usage when under high load, but more delay when performing ingame actions because of the pending data?
This setting can/should be increased if the CPU is not hitting 100%
Setting both closer to 1 will increase the cause of ACE asert when restarting the world service.
The high cpu usage of ReactorTimeOut=1 and a limit of processing of only 1 data per tick?
Usage with default settings i7 2600k vmware 2 cores:
ram 12.2% 444MB max
33-44% cpu
After
3-4% cpu
The binary's time check for the mob and other spawn settings.
The lower the value the more accurate it is, the more CPU it uses if a lot of data/settings.
WorldTimer=1:100,2:150,3:3000,6:60000,8:2000,9:10000,13:150,14:2000,15:30000,16:200,20:200,21:200
ID?:milliseconds
1:100, (mob idle move, attack back) cpu23%
2:150, (is checking for mob spawn) cpu15%
13:150, (?) cpu15%
16:200, (?) cpu9%
20:200, (?) cpu9%
21:200, (?) cpu8%
Example:
Nipperlug respawn every 10 seconds, set in mobs.scp?
2:150 The frequency of binary checks/updates the defined setting for when the Nipperlug should respawn?
This is what I figured out so far. (Sry for the mess, copy pasted notes bellow)
The cause is this,
WorldTimer=1:100,2:150,3:3000,6:60000,8:2000,9:10000,13:150,14:2000,15:30000,16:200,20:200,21:200
Setting them to,
WorldTimer=1:1000,2:1500,3:3000,6:60000,8:2000,9:10000,13:1500,14:2000,15:30000,16:2000,20:2000,21:2000
will reduce the cpu usage by quite a lot, since the binary updates/check's tick will now be done every 2-3 seconds instead of every 150 milliseconds.
There is a downside to raising the time as you might imagine it.
I will try to document these, if any one is interested. :
Notes:
------------------------------------------
ReactorTimeOut=100 smaller value, like 1 for example will cause near 100% cpu usage
ProcessDataLimit=1024, Limits how much data it can process at once, thus lower cpu usage when under high load, but more delay when performing ingame actions because of the pending data?
This setting can/should be increased if the CPU is not hitting 100%
Setting both closer to 1 will increase the cause of ACE asert when restarting the world service.
The high cpu usage of ReactorTimeOut=1 and a limit of processing of only 1 data per tick?
Usage with default settings i7 2600k vmware 2 cores:
ram 12.2% 444MB max
33-44% cpu
After
3-4% cpu
The binary's time check for the mob and other spawn settings.
The lower the value the more accurate it is, the more CPU it uses if a lot of data/settings.
WorldTimer=1:100,2:150,3:3000,6:60000,8:2000,9:10000,13:150,14:2000,15:30000,16:200,20:200,21:200
ID?:milliseconds
1:100, (mob idle move, attack back) cpu23%
2:150, (is checking for mob spawn) cpu15%
13:150, (?) cpu15%
16:200, (?) cpu9%
20:200, (?) cpu9%
21:200, (?) cpu8%
Example:
Nipperlug respawn every 10 seconds, set in mobs.scp?
2:150 The frequency of binary checks/updates the defined setting for when the Nipperlug should respawn?