BL1P

I have set up all required forwarded ports on my router or so I think ?. UDP : 2300-2305, UDP: 8766, UDP : 27016 I use ZoneAlarm which doesn't allow port forwarding. So for testing I disabled ZoneAlarm and Windows Firewall. Double checked the allowed ports on my router but still the server will not show via steam. Thanks for the suggestions though. I suppose there must be something else blocking my incoming or something but to be honest this is way too much work considering how much work is needed to get a server going in the first place adding another level of complexity is well.... very BIS :) I don't suppose your using BT Infinity Terox and if so would care to share how you set up your router ?

Even when using remote access my server and a friend of mines server still will not show. I was wondering if you knew that if you don't show via steam you still wont show via remote or at least why this is so ?

Is it case sensitive ? -cpuCount not -cpucount ? ---------- Post added at 06:15 PM ---------- Previous post was at 05:24 PM ---------- Same test ran with the following .bat on PERF1 (cpuCount and maxmem) NOTE :- This is NOT my community server this is my HOME server, Dwarden that is why the results are not in the 750s System Specs @echo off echo Launching BL1PPYS Arma3Server... start "" /wait /high "C:\Program Files (x86)\Steam\steamapps\common\Arma 3\arma3server.exe" -port=2302 "-config=D:\bl1pserver\bl1p\BL1P_config.cfg" "-cfg=D:\bl1pserver\bl1p\BL1P_basic.cfg" "-profiles=D:\bl1pserver\bl1p" -name=bl1pserver -cpuCount=6 -maxmem=3071 -mod=@ASM RPT :- ASM and Taskmanager image http://i.imgur.com/isZWzs1.jpg

The system i ran the test on is AMD As far as i know -enableHT has no effect on AMD systems ? could be wrong usually am :) Did run the same test on stable with -cpucount=6 but it made no difference to the server results.

FeedBack comparison for Stable and PERF1. Ran on home PC as Dedicated Server with client disconnecting after mission start. System specs in sig. Win 7 home. Mission ran is AIServerBenchmark Link = https://github.com/BL2P/AIServerBenchmark/releases/tag/3 Start params :- ==========TEST 1========== Test 1 :- (STABLE) Original output filename: Arma3Retail_Server Exe timestamp: 2014/06/18 11:59:50 Current time: 2014/06/24 06:55:15 Type: Public Branch: Stable Version: 1.22.125300 RPT results Test 1 :- ASM and Taskman performance Image for Test 1 :- http://i.imgur.com/ZyQlaOr.jpg (225 kB) ==========TEST 2========== Test 2 :- (PERF1) Original output filename: Arma3Retail_Server Exe timestamp: 2014/06/18 17:46:27 Current time: 2014/06/24 07:35:49 Type: Public Branch: Stable Version: 1.22.125313 RPT results Test 2 :- ASM and Taskman performance Image for Test 2 :- http://i.imgur.com/cLj42sx.jpg (222 kB)

Updated first post with the new version. New version has some small value changes in it from me. A diag_log function by Fluit which allows admins to set the log level so as to be able to set the amout of reports you get sent to the rpt. Also has an fsm by Fred41 which will smooth fps detection by the scripts and increase wait times for each unit creation the closer you get to the fps Limit. https://github.com/BL2P/AIServerBenchmark/archive/3.zip Happy Benchmarking.

+1 for this

https://github.com/BL2P/SeverAIBenchmark Fluit is currently writing a function to set log levels on the .rpt reports so that server admins can turn on or off the reports or lower the amount of reports.

I have a cunning plan ! So... the mission is on Github now and you are a collaborator :)

wow ;) sleep (abs(50.0 - diag_fps) * abs(50.0 - diag_fps) / (50.0 - FPSLIMIT)) I have no idea wtf that does :) It looks like its some sleep value taken from an absolute value of (50 - the servers fps) * (50-server fps) /(50-server fps) ? Im crap at this m8 I might need help sorry :( isnt 50-server fps a bad thing as 50-45 is 5 ? so 5 * 5 / 5 = 5 ? ---------- Post added at 12:51 AM ---------- Previous post was at 12:39 AM ---------- hello m8. the only thing i can think of which might stop the mission from working is the server or hc is already below 45fps. To test that i would turn off the fps limiter in params and run it like that with just an AI amount limit

Although I am very pleased you like this benchmark m8. I have some counters to your suggestions :( The reason i have such a large time between batches spawning is based on the responses Dwarden and a dev gave in similar thread that the AI need time to run certain scripts and path finding things. Spawning the AI too fast will give a very detrimental effect on the cps and fps because of the exponential blablabla :) I prefer the batch to single unit but... I am open to a reason why I should use single unit over batch and if a dev gives a good reason ill definitely swap the way the script works. the reason i like batch over single is that each group would need time and using batch lowers the overall time. I can do a single unit version with no problems but not over the weekend :) as im playing and drinking now till monday :) Just a note. I am really pleased your taking an interest in this Fred as I like all the work you do any suggestions are more than welcome. I am just a stubborn old bastard that needs shit explaining too me before ill do it :)

@sari Pritty cool how you start off at 45fps and gradually increase! as ai are added then slowly decrease as it gets near the end. What .exe did you use on your pc the Server.exe or client .exe ? also what start up did you use if it was server any -cpucount lines or anything ? we use a .bat to launch the server start /high /W arma3server.exe -port=2302 -config=Server_HC\server.cfg -cfg=Server_HC\ArmA3_test.cfg -profiles=Server_HC -name=Server_HC -enableHT and the two things we use in there are /high and -enableHT ---------- Post added at 06:38 PM ---------- Previous post was at 05:59 PM ---------- Following your example Sari I ran the test on my Home PC as well :P I only got upto 450ai on my Home pc before it went and stayed below 45 This is my startup .bat for my home pc @echo off echo Launching BL1PPYS Arma3Server... start "" /wait /high "C:\Program Files (x86)\Steam\steamapps\common\Arma 3\arma3server.exe" -port=2302 "-config=D:\bl1pserver\bl1p\BL1P_config.cfg" "-cfg=D:\bl1pserver\bl1p\BL1P_basic.cfg" "-profiles=D:\bl1pserver\bl1p" -name=bl1pserver -mod=@ASM ASM RPT. System Specs ps. How do I get the image to appear in the spoiler like you did ? for me it just shows a link to the image ?

use the mission supplied in my post and run it on your server an it will give you the results. http://forums.bistudio.com/showthread.php?176492-What-limits-the-server&p=2713453&viewfull=1#post2713453 <--- the post with the mission dl at the bottom :) also use ASM by Fred41 to get a graphical screenshot of the servers performance.

The mission to be used only has one slot. So player count isn't a problem. The mission is testing how many AI your server can get upto before your server goes below and stays below 45fps. not how many players you can have connected. Hardware should be posted but this isn't a prize competition its for fun and to collect results people are getting and talk about that. so I wont be setting things in stone that people must follow :) Uptime is upto the person running the test if they want to run the server for 24hours then run the test mission thats upto them :) I thought it would be nice to have a benchmark for ai amounts and this seemed like a stable enough mission to use. Maybe someone will come up with some better scripting than I used. thats fine Im no scripting guru. just some fat gamer who does the odd bit of script.

Cool gona give the prof a go after the weekend. see what results that gives :)

I did post a full ASM screen did I not ? Unsure if thats what you mean m8 ?

Ran the version3 test on our community server :) http://i.imgur.com/a0kU8Ul.jpg The gauntlet has been thrown ! FPS LIMIT ON" SPLITAI OFF" HEADLESSON = false" FPSLIMIT = 45" TIME PER UNIT = 0.05" AI PER BATCH = 25" BATCH TIME = 7.5" MAX TOTAL AI = 1000" so 750ai created with fps limit on at 45 beat that :) Server specs:-

weird your screenshot doesn't show any details on the left ? no host name no mission name no fps numbers etc-etc ? so I dont know if its server on top or hc on top how many ai blablabla:) EDIT : I figured that bit out by looking more closely and rereading what you typed :)... looks like you ran 1000 ai nice with the fall offs starting at about 500ai and the server having the worst looking graph.(most spikes). Anyway I have noticed a big increase in the ai my server will produce since the latest stable update. I think I went from 200-250ai to 300-350ai quite a % increase. One weird thing I noticed was that in a new version of the mission I made version3 :) It will let you have the mission place ai on HC upto either the amount specified or the fps limit then it places the ai on the server. So when running that version with it set to place ai until fps <45 for me it would place ai on the hc upto about 250-300 then it would start placing ai on the server. when it placed ai on the server the hc would still degrade in performance even though it was not creating anymore ai. at first I thought this was because server and hc where same pc.(so same cores or same cpu) So I created the server on my home pc. and used my server box as my hc client. It produced the same results!!!. once hc dropped below the specified fps limit the server would take over but the hcs fps would still drop with each group created. strange ! It makes me think that even though the server and hc are using separate threads the ai are not truly local to the threads. and that ai placed on server or ai placed on hc will effect the opposite. meaning that the hc is basically useless atm. Yes you get a performance increase at the lower levels of the ai created but once you pass the max limit your server can handle on its own using a HC or not makes no difference in the overall performance. EDIT Mission version 3 :) http://www.gamefront.com/files/24303699/AItest3.Stratis.rar disclaimer : blips a fool and takes no responsibility for any fkups with this mission/bechmark

You using affinity to set the cores or just using -cpucount in a command line? Because I believe the latter doesn't work. Could be wrong but I think I saw somewhere to use affinity to set cores not -cpucount I also think I saw a post somewhere about -cpucount not actually being used anymore in arma3. Ref 1 : https://community.bistudio.com/wiki/Arma_3_Dedicated_Server#Headless_Client_on_a_Dedicated_Server_Notes

Did you run HC on same cores as the Server ?

Ran some more HC tests Test-1 was 4(all) cores on both Server and Hc default settings ie (25 ai every 30sec upto 400) http://i.imgur.com/jedSLKp.jpg Test-2 was core 1and 2 for Server core 3 and 4 for Hc default settings http://i.imgur.com/GKnorIL.jpg Test-3 Server 4core no HC no Fps limit 25ai every 30 sec upto 400 http://i.imgur.com/tdl2vJx.jpg So yer my server performs better without HC !

Hello Fred41. Yes I do have this line in my server_config.cfg :- steamPort = 2300;//--- doesntfuckingwork steamQueryPort = 2301;//--- doesntfuckingwork hostName = "BL1PLAND STUDIOS"; password = ""; passwordAdmin = "ILOVEFRED41"; reportingIP = "arma3pc.master.gamespy.com"; //logFile = "BL1PSERVER.LOG"; localClient[]={127.0.0.1}; motd[] = {}; motdInterval = 10; maxPlayers = 20; kickduplicate = 1; verifySignatures = 0; requiredSecureId = 2; voteMissionPlayers = 3; voteThreshold = 0.33; disableVoN = 1; vonCodecQuality = 10; persistent = 1; timeStampFormat = "short"; BattlEye = 1; doubleIdDetected = ""; onUserConnected = ""; onUserDisconnected = ""; onHackedData = ""; onDifferentData = ""; onUnsignedData = ""; regularCheck = ""; class Missions { class Mission_1 { template = "AItest2.Stratis"; difficulty = "mercenary"; }; }; looks right to me ? I have just DLed your newer version of Arma Server Monitor thanks for the heads up. I will give your "a simple registry tweak, for increased performance" a try on the server aswel as your "An improved custom memory allocator for arma" If I can figure out how to get the LARGEADDRESSAWARE flag set correctly :) I Hope the mission helps get some results back from different server hosts that in turn helps get some feedback to the devs. Its always nice to have a benchmark type mission. EDIT:- Actually on reading your notes on the server part of the "a-simple-registry-tweak-for-increased-performance" ill leave that alone :)

Thanks very much m8. Whats your server specs btw you had some nice results. O yer I have fiberoptic to the box then copper to the house BT Infinity (UK). PS. This server I am using in these tests is only for mission creation and the like. I rarely get more than 1 or two people testing on it with me. But would be nice to get the headless to actually do something other than cost money ! PPS. What where these "Random Network parameters active" ? from your second test results

Erm thanks :) I still don't get this bit though. "Bandwidth the server is guaranteed to never have (in bps)." so I could put 1milliontrillion(bps) because I know it is guaranteed to never have that ? but you say that mine is way too high which I don't understand.

First i think i uploaded a version with the report of ***FPS LIMIT ON backwards lines 34 and 75 should read if (ENABLEFPSLIMIT > 0) then {diag_log "**** FPS LIMIT ON";}else{diag_log "**** FPS LIMIT OFF";}; and not if (ENABLEFPSLIMIT > 0) then {diag_log "**** FPS LIMIT OFF";}else{diag_log "**** FPS LIMIT ON";}; or maybe not and your results where just misleading me on the second run and your server is recovering from below 45 to create and then dropping below again for the report ? I don't understand "smaller aggregate simulation messages and higher bandwidth" Could you explain that to a layman like me ? Here is my .cfg stuff maybe you could suggest some changes for me? MaxMsgSend = 128; MaxSizeGuaranteed = 512; MaxSizeNonguaranteed = 256; MinBandwidth = 131072; MaxBandwidth = 2097152000; MinErrorToSend = 0.001; MinErrorToSendNear = 0.01; MaxCustomFileSize = 1310720; class sockets{maxPacketSize = 1400;}; My connection is 75MB down and 20MB up http://www.speedtest.net/my-result/3568025481