I'm not shure exacly on the load you have on the dod servers, but here is a
top copy paste from an athlon x2 @ 4200
top - 06:28:30 up 4 days, 6:54, 1 user, load average: 4.04, 4.16, 4.09
Tasks: 86 total, 8 running, 78 sleeping, 0 stopped, 0 zombie
Cpu0 : 62.0% us, 6.3% sy, 0.0% ni, 25.0% id, 0.0% wa, 0.0% hi, 6.7% si
Cpu1 : 64.7% us, 5.7% sy, 0.0% ni, 29.3% id, 0.0% wa, 0.0% hi, 0.3% si
Mem: 1992952k total, 1975976k used, 16976k free, 5284k buffers
Swap: 2104504k total, 3288k used, 2101216k free, 402104k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ P COMMAND
23729 pub2 15 0 418m 376m 7876 S 32.6 19.4 361:45.33 1 hlds_amd64
17196 pub1 15 0 458m 413m 7828 S 28.6 21.2 646:06.19 0 hlds_amd64
363 pub5 15 0 138m 110m 6556 S 26.3 5.7 5:10.41 1 hlds_amd64
31132 pub4 15 0 133m 109m 6380 S 18.3 5.7 45:06.07 0 hlds_amd64
646 pub3 15 0 186m 158m 6652 R 14.0 8.2 397:34.68 1 hlds_amd64
5496 root 16 0 35876 26m 5060 R 10.0 1.4 405:35.37 0 snort_inline
32401 cws 15 0 86044 55m 6460 R 4.7 2.9 4:19.65 1 hlds_amd64
5633 cws 15 0 122m 93m 6760 R 2.7 4.8 122:55.23 1 hlds_amd64
5598 cw 15 0 86044 57m 6400 R 2.0 2.9 97:17.82 1 hlds_amd64
5685 cws 15 0 96140 66m 6744 R 2.0 3.4 102:53.09 1 hlds_amd64
5699 cws 15 0 93996 65m 6712 R 2.0 3.4 101:55.44 1 hlds_amd64
At the time of the capture the publice serveres where all full (20*5 players
in total, and the clanwar servers were all empty)
Here is another example from a dual-opteron 265 (dual core):
top - 22:27:21 up 25 days, 22:42, 1 user, load average: 4.37, 4.73, 4.87
Tasks: 124 total, 6 running, 118 sleeping, 0 stopped, 0 zombie
Cpu0 : 54.3% us, 5.6% sy, 0.0% ni, 33.1% id, 0.0% wa, 1.7% hi, 5.3% si
Cpu1 : 43.0% us, 2.6% sy, 0.0% ni, 54.0% id, 0.0% wa, 0.0% hi, 0.3% si
Cpu2 : 49.7% us, 1.7% sy, 0.0% ni, 48.7% id, 0.0% wa, 0.0% hi, 0.0% si
Cpu3 : 44.5% us, 1.7% sy, 0.0% ni, 53.5% id, 0.0% wa, 0.0% hi, 0.3% si
Mem: 2058636k total, 2045512k used, 13124k free, 8088k buffers
Swap: 4200988k total, 642132k used, 3558856k free, 449560k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ P COMMAND
13836 pub7 15 0 275m 231m 7992 R 37.1 11.5 275:27.04 3 hlds_amd64
15628 pub1 15 0 154m 119m 7336 R 28.1 5.9 43:20.68 3 hlds_amd64
18173 pub8 15 0 187m 143m 4260 R 21.5 7.1 561:45.70 1 hlds_amd64
18008 pub9 15 0 210m 163m 4220 S 20.2 8.1 489:58.11 2 hlds_amd64
32358 pub3 15 0 311m 201m 4480 R 19.2 10.0 1094:04 2 hlds_amd64
15728 pub4 15 0 136m 103m 7056 S 18.9 5.1 15:45.42 1 hlds_amd64
15001 pub5 15 0 151m 124m 6660 R 18.2 6.2 80:48.70 2 hlds_amd64
15570 pub2 15 0 135m 102m 7152 S 16.9 5.1 41:22.89 0 hlds_amd64
5959 root 5 -10 36012 25m 4824 S 10.3 1.3 2942:24 0 snort_inline
612 cws 15 0 213m 69m 3756 S 5.0 3.4 676:52.52 0 hlds_amd64
626 cws 15 0 128m 38m 3732 S 2.6 1.9 560:21.26 0 hlds_amd64
640 cws 15 0 112m 52m 3368 S 2.3 2.6 535:17.69 0 hlds_amd64
6055 pub6 15 0 114m 56m 3808 S 2.0 2.8 628:23.62 0 hlds_amd64
736 cws 15 0 95120 48m 3572 S 2.0 2.4 499:07.16 0 hlds_amd64
654 cws 15 0 99176 51m 3664 S 1.7 2.6 495:00.01 0 hlds_amd64
12621 cws 15 0 141m 34m 3760 S 1.7 1.7 300:48.94 3 hlds_amd64
All public servers except pub6 are fully loaded 20 player servers (160
players in total).
When you have equally loaded servers, and you are pushing the server under
50% idle, all procesors get enough work to do.
CPU utilization/processors is expressed as usage/core, so an average of 20%
usage/process/core is fair enough.
Besides, being SMP enbled, when map change occurs spiks are not even felt,
compared to equaly used single core processor witch is truly a good thing :)
----- Original Message -----
From: "Andrew Forsberg" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Monday, June 26, 2006 11:59 AM
Subject: Re: [hlds_linux] SMP
That's not strictly speaking correct. Although of the three hlds
processes only one does much work. Here's a dod example:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ PPID P COMMAND
5245 dod 0 -20 57736 1268 944 S 0.0 0.0 0:00.01 1 0 hlds_run
32381 dod 0 -20 79548 67m 8092 S 0.0 2.2 2:20.49 5245 3 hlds_amd
32409 dod 0 -20 79548 67m 8092 S 0.0 2.2 0:00.00 32381 2 hlds_amd
32411 dod 0 -20 79548 67m 8092 S 0.0 2.2 0:01.98 32409 2 hlds_amd
Where P (second from the right) is the last used CPU ID. Like you say --
when you've got quite a few servers running simultaneously, SMP helps in
a big way.
On Mon, 2006-06-26 at 11:05 +0300, GoD2.0 wrote:
Hello
Don't know about scrds but hlds doesn't... but really talking serios...
it
doesn't need to. If you have a mid level-dual core you can hold up to
about
2x20 player servers/core :) So yes smp helps when you're holding many
servers :)
_______________________________________________
To unsubscribe, edit your list preferences, or view the list archives, please
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds_linux
