|
|
#1 |
|
Banned
Join Date: Nov 2009
Reputation: 325
Posts: 2,046
|
please help me test multiple game instances on a single core
please help me test multiple game instances on a single core and have some fun
look out for the following dedicated servers and join them if you can... [Dedicated] Mad House - ff is off and hardcore is off [Dedicated] House Of Pain - ff is on and hardcore is on [Dedicated] lsm + tdm - ff is off and hardcore is off [Dedicated] lsm + tdm hardcore - ff is on and hardcore is on amd p2 x4 955 (3.2 ghz quad) 8 gb 1066 ddr2 2 x 500 gb 7200 rpm sata ii raid 0 windows 7 64-bit fios 30/20 mbps if you run into lag or other issues please post details. if it runs great also post something... by the way the server is located in the north east usa near nyc, so if you are thousands of miles away it may lag a bit, regardless it would be good to know post your specs and how you are connected to the internet and what ping you see for my server... also post sort of where you are, i dont want to know your home address, please dont post nor send me, just like the area of country and world, like north east near nyc... Last edited by i_2_i: 02-07-2011 at 01:29 PM. |
|
|
|
|
|
#2 | |
![]() Join Date: Jun 2010
Reputation: 274
Posts: 1,838
|
Quote:
Thanks man. Would be great if this works out. |
|
|
|
|
|
|
#3 |
|
Banned
Join Date: May 2005
Reputation: 4
Posts: 464
|
I'll give it a go, but when I used just one on our dedi it would max out the entire dedi in matter of seconds, just a note
![]() Xeon X3220, 4GB of ram. Gonna give it a go right now. |
|
|
|
|
|
#4 |
|
Banned
Join Date: May 2005
Reputation: 4
Posts: 464
|
Just been on it for a few secs, apart from the ping of 120 (I'm in the Netherlands, so it might just be a distance problem), I found no problems.
Have not been able to playtest with actual people. |
|
|
|
|
|
#5 |
|
Banned
Join Date: Nov 2009
Reputation: 325
Posts: 2,046
|
so far its running great people say no lag
all on a single core of a amd p2 x4 955 (3.2 ghz quad) cpu usage 14-17%, <1, <1, 5-6% instances using 225,250,300,235 mb respectively using right now 300-700 Kbps |
|
|
|
|
|
#6 |
|
Banned
Join Date: Nov 2009
Reputation: 325
Posts: 2,046
|
|
|
|
|
|
|
#7 |
|
Banned
Join Date: May 2005
Reputation: 4
Posts: 464
|
Just looked at it, it seems they've improved on the dedi A LOT.
Might be worth trying to run one again! |
|
|
|
|
|
#8 |
|
Banned
Join Date: Nov 2009
Reputation: 325
Posts: 2,046
|
the latest release is good
the old version tied up the cpu even noone was playing the new version shows <1% cpu utilization when noone is playing |
|
|
|
|
|
#9 |
|
Banned
Join Date: May 2005
Reputation: 4
Posts: 464
|
Still using 30% idle here... Gonna see what it does with some more players on.
|
|
|
|
|
|
#10 | |
|
Banned
Join Date: Nov 2009
Reputation: 325
Posts: 2,046
|
Quote:
try it configure 4 instanced and set them to run on the same core then run process explorer and see cpu usage honesty < 0-1% if idle no instance has gone much over 20% i do acknowledge if you give it a lot of cpu it will use it... you have to restrict it, it runs much like a cache if you are not using it then it uses it... physx thing i'm figuring, but they scale... again 4 16 player instance on a single core cheap amd 955 cpu, old 1066 ddr 2 memory if you have the client and if you see my test servers and if the ping is slow, im north east usa near nyc then join one of them and check it out... |
|
|
|
|
|
|
#11 |
|
Guest
Posts: n/a
|
|
|
|
|
#12 |
|
Banned
Join Date: Nov 2009
Reputation: 325
Posts: 2,046
|
check out
http://www.breachgame.com/forums/dow...file.php?id=12 please run process explorer http://technet.microsoft.com/en-us/s.../bb896653.aspx check out cpu usage then configure and start multiple instances on the same core for each instance dont forget to change the port number in breach.ini and server.ini base directory bgs bgs\bgs1 bgs\bgs2 bgs\bgs3 bgs\bgs4 start /affinity 2 bgs.bat 1 start /affinity 2 bgs.bat 2 start /affinity 2 bgs.bat 3 start /affinity 2 bgs.bat 4 bgs.bat: cd bgs%1 :top breachserver goto top |
|
|
|
|
|
#13 |
![]() Join Date: Jan 2010
Reputation: 170
Posts: 1,690
|
Oh Mad House is you? I played there a few times today. Had good experience except for a couple of rubber banding people - but I'm sure that's down to their own connections.
|
|
|
|
|
|
#14 |
![]() Join Date: Jan 2011
Reputation: 0
Posts: 48
|
Have to say - using the dedicated server you download via steam. I am pulling 20-30% even idle. Are you perhaps using another version?
|
|
|
|
|
|
#15 | |
|
Banned
Join Date: Nov 2009
Reputation: 325
Posts: 2,046
|
Quote:
http://technet.microsoft.com/en-us/s...rnals/bb896653 please understand if you just run the server it will use about that. but what i did was stick 4 16 player instances on a core by forcing affinity so on a quad core core 0 is os, core 1 and 2 were available and core 3 has the game instances running i then monitored the games, very few people complained about any lag... most overwhelmingly said it ran great please understand a game is not like other applications, the game loop is always doing stuff and just looping will tie up the cpu... think of it no one is playing the game message loop is running at full thruttle since the server is headless, no display, it could in threory interating a great many more times than it's needed. as the more and more players join and the instances fill up will those players will utilize the cpu so the message loop will iterate far less, think of it like fps, when noone is on we are like getting 1000 fps, when 10 player join then we are getting far fewer fps, but noone really notices until the iterations or in our example fps falls below what is noticably acceptable. the thing is when the server was empty of players it didn't need to loop that many times heating up the cpu but it does and you messure it as the 20-30%. but this does not matter because the server while not able to loop at many times when it is full it still looping enough to keep players happy and playing smoothly.. also on the server side without hardware support physx make significant demains on the cpu but again just because the cpu may show 100% utilization, it does not mean it's iterating the game message loop too few times... infact the server while running always hows 100% even on an instance running on a core, but as i added instances the game play continued to be acceptable. because the server was iterating too many times... well why did they code things that way, because that is how game engines are coded a single thread message loop. sure you can have other threads for a great many things, but just like in a windows application there is only a single threaded message loop. could they have implemented a yield and locked down the iterations to some maximum amount so as to not tie the cpu, sure they could have and they may very well in the futue. but from the simplicity of the logic and the fact that it does not interfere any noticable amount or really limit the number of instances they didn't need to... they could get away with it... when you write a windows forms application do you put a yield no, someone did for you well someone didn't for them cause they wrote the engine... |
|
|
|
|
![]() |
|
||||||
|
|