<div class="IPBDescription">Has it droped for you?</div> We use a rental company for our server so there is no way for me to see what our usage is. Has the CPU usage droped alot for you with 2.0?
Yes and No, the default server footprint seems to have gone up - for example I have a test machine running an AMD k6-II 500. Previously this used 0-1% CPU when empty. It now uses 15% CPU when empty.
However the additional footprint that each player adds is now significantly reduced as is the impact of additional entities throughout the game (although there is still an impact). This means that CPU usage creep is much less of an issue.
Actuall examples from servers I run:
Celeron 800 - previously could handle 12 players, now handling 10 players (CPU load 80 - 89%). AMD 2100+ - previously could handle 16 players, now handling 24 players (CPU load 70-85%). AMD K6-II 500 - not tested with more than 2 players - I wouldn't expect much ;P
In my opinion the newer NS 2.0 now scales much better but there is a higher minimum spec (at a guess a 1Gig Athlon) that is needed - I would exclude Durons and Athlons - I don't know how the lack of on die cache affects things but I wouldn't think it helps.
<!--QuoteBegin--haven+Aug 4 2003, 08:41 AM--></span><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>QUOTE</b> (haven @ Aug 4 2003, 08:41 AM)</td></tr><tr><td id='QUOTE'><!--QuoteEBegin--> Yes and No, the default server footprint seems to have gone up - for example I have a test machine running an AMD k6-II 500. Previously this used 0-1% CPU when empty. It now uses 15% CPU when empty. <!--QuoteEnd--> </td></tr></table><span class='postcolor'> <!--QuoteEEnd--> That's probably because you're using the new HLDS I assume? If you go back to the old one and apply the unofficial patch you should see it go back to 0-1% cpu usage when empty.
Currently using 3.1.1.1d with the NS 2.0 compatibility patch and metamod 1.16.1. So whats this unofficial patch and why would I want to run it ? I'd be grateful for any details - otherwise just a name will do and I'll go trawl for it myself.
CPU utilization has dropped dramatically for me. Win2K server, HLDS 4111d, P3/800 512MB. Used to have really high CPU usage at around 12-14 players, and it got really bad if the game went for more than 30 minutes. Which meant a lot of lag. Since 2.0, I've been able to increase the maxplayers from 16 to 18, and CPU usage has been almost a non-issue. Still see some heavy usage during firefights with a lot of players, but it is hardly noticeable to clients at all. We'll see a little increase in ping times, but it's not unplayable. I'm even considering increasing to 20 players.
Kudos to Flayra and all the mapmakers for addressing the performance issues and making this one incredible release.
My CPU usage hovers around 30% average with hera and bast going up to 50%.
Server Specs: AMD Athlon 1900+ 512MB ram, RedHat 7.3 Game Configuration: 12 player server, HLDS_L 3.1.1.0 with Boffix patch, AdminMod 2.50.56, Metamod 1.16.1, about a dozen plugins between metamod and Adminmod
Overall CPU and player graph: <a href='http://www.vadakill.com/rrdtool/Vadakill-2.0-other-day.png' target='_blank'>CPU Usage Stats - 24 hour</a>
CPU creep is defiately better but it is still quite consuming per user slot comparitively. As a hosting provider I won't be lowering my NS rates anytime soon.
CPU usage on my server decreased from 0.2% to 0.1% its very noticable! <!--emo&:D--><img src='http://www.unknownworlds.com/forums/html/emoticons/biggrin.gif' border='0' style='vertical-align:middle' alt='biggrin.gif'><!--endemo-->
We have lowered our rates in accordance with this release (check the sig for link) [/advertisement]
It seems like someone put a finger in the leak, if you see what i mean, before it just built up and up and up and eventually saturated ANY Cpu, now it seems a bit more steady and controllable, uses about 60% of a Celeron 1.7 with 16players.
Thats on 3.1.1.0/slackware, not bad if you ask me, comparable to other mods infact <!--emo&:)--><img src='http://www.unknownworlds.com/forums/html/emoticons/smile.gif' border='0' style='vertical-align:middle' alt='smile.gif'><!--endemo-->
I dont know how my server providor does it. But NS always used an insanely low amount of CPU, strangely though pings are high but its very stable. <!--emo&:0--><img src='http://www.unknownworlds.com/forums/html/emoticons/wow.gif' border='0' style='vertical-align:middle' alt='wow.gif'><!--endemo-->
v1.04 14 player server was running Windows 2000 server and getting around 80-90% CPU usage with pings spiking up to the 400's at the end of some maps.
upgraded software too:
v2.0 16 player server, running Windows 2003 server. Now my CPU usage is around 30-50% and pings never spike. They remain stable throughout the game (usually half the people are under 100). I love it <!--emo&:)--><img src='http://www.unknownworlds.com/forums/html/emoticons/smile.gif' border='0' style='vertical-align:middle' alt='smile.gif'><!--endemo-->
My buddy is using Dual Opteron for the server, and its running pretty well, can handle 27 players (one is for HLTV, 30 spectators) w/o any problems /W another 32 players DoD Server running on the same server.
He said he will try to put the NS to 32 Players as well and see what happens. I'll report the results back later on.
Comments
However the additional footprint that each player adds is now significantly reduced as is the impact of additional entities throughout the game (although there is still an impact). This means that CPU usage creep is much less of an issue.
Actuall examples from servers I run:
Celeron 800 - previously could handle 12 players, now handling 10 players (CPU load 80 - 89%).
AMD 2100+ - previously could handle 16 players, now handling 24 players (CPU load 70-85%).
AMD K6-II 500 - not tested with more than 2 players - I wouldn't expect much ;P
In my opinion the newer NS 2.0 now scales much better but there is a higher minimum spec (at a guess a 1Gig Athlon) that is needed - I would exclude Durons and Athlons - I don't know how the lack of on die cache affects things but I wouldn't think it helps.
Those are my experiences, yours may differ ...
20 slots, pings around the 300-400 normal maps. on 1.04 we could run 14-16 players and had a ping of 20-50
and still using a lot of cpu power
redhat 9
(didnt test slackware box yet).
That's probably because you're using the new HLDS I assume? If you go back to the old one and apply the unofficial patch you should see it go back to 0-1% cpu usage when empty.
So whats this unofficial patch and why would I want to run it ?
I'd be grateful for any details - otherwise just a name will do and I'll go trawl for it myself.
I haven't tried the patches for windows, but it works fine for linux.
Game Server Stats: Pentium 4 2.53Ghz, 1GB DDR RAM, Windows 2000 Advanced Server
Kudos to Flayra and all the mapmakers for addressing the performance issues and making this one incredible release.
Server Specs: AMD Athlon 1900+ 512MB ram, RedHat 7.3
Game Configuration: 12 player server, HLDS_L 3.1.1.0 with Boffix patch, AdminMod 2.50.56, Metamod 1.16.1, about a dozen plugins between metamod and Adminmod
Overall CPU and player graph:
<a href='http://www.vadakill.com/rrdtool/Vadakill-2.0-other-day.png' target='_blank'>CPU Usage Stats - 24 hour</a>
XP2100
1 Gig DDr ram
windows 2003 advanced server
It seems like someone put a finger in the leak, if you see what i mean, before it just built up and up and up and eventually saturated ANY Cpu,
now it seems a bit more steady and controllable, uses about 60% of a Celeron 1.7 with 16players.
Thats on 3.1.1.0/slackware, not bad if you ask me, comparable to other mods infact <!--emo&:)--><img src='http://www.unknownworlds.com/forums/html/emoticons/smile.gif' border='0' style='vertical-align:middle' alt='smile.gif'><!--endemo-->
v1.04 14 player server was running Windows 2000 server and getting around 80-90% CPU usage with pings spiking up to the 400's at the end of some maps.
upgraded software too:
v2.0 16 player server, running Windows 2003 server. Now my CPU usage is around 30-50% and pings never spike. They remain stable throughout the game (usually half the people are under 100). I love it <!--emo&:)--><img src='http://www.unknownworlds.com/forums/html/emoticons/smile.gif' border='0' style='vertical-align:middle' alt='smile.gif'><!--endemo-->
He said he will try to put the NS to 32 Players as well and see what happens. I'll report the results back later on.