Yes, I have also noticed stability/performance is extremely map dependent. Tried 128-tick Arms Race server on ar_chess and the var was constantly over 2.0 as soon as there were more than 10 players (must be all those floating teacups!).
Same hardware can handle 24 slots on other maps and gamemodes. That's a fairly big gap. Regarding tickrate drop on round start: I've pretty much come to accept it since the performance-destroying update. What I want to know is why we took a performance hit after that update. That's all I want to know: just why is the CPU working so much harder than before? At least if I knew it was going to good use (optimised hit reg maybe), then I could move on and work with it, but at the minute, I don't think anyone saw any subjective improvements from that update. Looks like major inefficiency in the code. > > Message: 4 > Date: Wed, 26 Mar 2014 19:04:59 +0200 > From: ics <[email protected]> > To: [email protected] > Subject: Re: [Csgo_servers] Server performance problems > Message-ID: <[email protected]> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > Interesting results you got. I'm running 128 tick 24 slot server on > Q9559 (2.83GHz) and it used to be fine before Valve's update that > brought performance down. So right now i suffer from unstable var and > tickrate dropping on round starts. Some maps are the worst, for example > Overpass really eats CPU a lot more than Dust2, Mirage or even cbble. > > -ics
_______________________________________________ Csgo_servers mailing list [email protected] https://list.valvesoftware.com/cgi-bin/mailman/listinfo/csgo_servers
