04-13-2009, 10:23 PM (This post was last modified: 04-13-2009, 10:45 PM by christophf.)
Hi all,
I'm new here and would like to introduce myself. Im concerned already for a very long time with the topic Kernel Optimatizion etc.
My resualts are, that my Server got very stable FPS without some libhacks etc.
But, not absoulutely stable. So now i hope u can help me to fix my problem to get 100% 1000FPS.
I add a picture from my server
Edit: using Realtime Kernel with realtime processing on hlds and on timers (softirq-hrtimer)
Do you have a multi core server? if so do you have htop installed? cna you tell me when running the idler which core does the idler max out? is it the same core your srcds is running on sometimes the idler will run on a different core than the srcds server which will have no affect at all. I found the idler runs best on the core of the server you wish eliminate the "bumps" from. Try using taskset to force the idler onto a certain core.
*Windows lack of output*
You: Hey, I want to run this program!
Windows: Ok.. It crashed... Now what? Give up?
You:...wtf...
*linux output helpful?*
You: ./My_program
Linux:...Failed!...oo kitties!
You:...wtf...
afterhoursgaming Wrote:sometimes the idler will run on a different core than the srcds server which will have no affect at all.
in my experience, the idler does not need to run on the same core as the server. in fact, I have found a single idler to be effective even for all servers running on the root.
have you tried everything that is written in the FAQ / trouble shooting section of the howto? and please make a longer measurement with players on the server (use real players, not bots). the one measurement with players has only a single drop, you cannot derive much information from that...
Hello again,
now my Server FPS are constantly on 1000FPS. I use a Realtime Kernel and some fixes, but when 10 players are on the FPS drops something to 500FPS and in the next second up to 1000FPS how can i solve this problem.
Is there any "trick" in the Kernel options to fix it.
Ilder has no effect, i dont use Chrt.
06-14-2009, 08:52 PM (This post was last modified: 06-14-2009, 08:55 PM by christophf.)
In my Opinion the Chrt is fine for a fast result. But my tests without Chrt, and other kernel configurations produce smooother gameplay.
here are a short measurement. Sometimes the FPS are constantly on 1000FPS u see in the middle of the measurement, but i dont really know how why.
my cyclictest show the kernel delay on 2-3ms
and this is a list of my interrupts, i hope someone can help me
I've tried different kernels on many different systems. RT is not always good, but in the great majority it delivers great performance - if correctly used (see my howto).