We have a Dual 3.06GHz Xeon bot running hyper-threading so 2 physical processors and 2 virtual processors. The box also has 4GB RAM We notice at peak 24 players and about 10 vehicles we start to get lagg, massive lagg. Now people say on Mittoes there is 40 players and 20 vehicles and it's not laggy. Well what is the difference then? I saw people say it's a single threaded SRCDS at the moment, that's fine we only see it get to 25% and then it doesn't seem to go beyond that but how is Mittoes server getting 40 players and more vehicles? What are the specs of the box or how has it been tweaked??
You have to: 1. use Windows 2. reduce rate of sending minimap updates 3. limit vehicles 4. pray that OB SDK version will be released soon Server uses 512-1024 MB RAM (I'm not sure) so there will be no big difference between 4GB and 1GB. HDD matters only when changing the map. HT won't help you at all. Game uses only 1 core/virtual processor.
Win or unix-System? I can give u a server.cfg from me! :] ...is set ...set to 10 Vehicle/Team (include CV)
We are currently using Windows. We haven't changed the minimap we currently don't have a problem with bandwidth unless that causes CPU load with massive calculations? We could reduce to 10 vehicles but why is everyone saying Mittoes has 20 vehicles and 40 players? We are lagging with like 10 vehicles total? Send away Lala, I'll check it out, thanks.
Minimap causes CPU load. You may try to use 'vprof*' and 'emp_vprof*' commands to see what takes most time.
Ok, so what are other servers running? What are you talking about with vprof? when entering "vprof*" and "emp_vprof*" in console nothing happens it says unknown command when I do a cvarlist I see: emp_vprof_print : cmd : : Print empires vprof report emp_vprof_record_interval : 30 : , "sv" : emp_vprof_reset : cmd : : reset vprof times emp_vprof_start : cmd : : Start empires vprof emp_vprof_stop : cmd : : Stop empires vprof The server isn't full at the moment and we are on district so no vehicles and "emp_vprof_print" says Vprof is not running?
By 'vprof*' I mean all the commands that match that pattern. vprof, vprof_print and so on. There are 2 profilers: emp_vprof and vprof. You enable/disable using 'vprof/emp_vprof'. It should create a log on harddrive when you disable it. You might also get some data to your console using emp_vprof_print or similar commands. You have to *MANUALLY* start the profiler and manually stop it. If you don't enable it then you won't get any data at all.
It's not a bandwidth problem, we have looked at how much is being used and we are not hitting our capacity.
Don't try to use tick 66 or 100. Stick to 33. Oh, and it will only use 25% CPU due to how windows reports CPU usage with multiple CPUs. When windows says 25% with your setup, it means the program is taking up 1 full CPU. Turn hyperthreading off, so you can utilize 50%
Default tickrate is 66. You have to manually set it to 33 with '-tickrate 33'. HT won't help it can only run slower.
It only says 25% because it counts 100% to be ALL four CPUs firing. Turning off HT will have no effect...it may say 50%, but that's only because windows now thinks it has 2 processors instead of 4.
Hmm, this might help... Default tickrate for most games is 33, didn't know Empires was 66, I'll look into this, thanks.
AFAIK Hyperthreading splits your 3.0 GHz CPU into 2 1.5 GHz CPUs. Turning off hyperthreading on a dual core gives you 2 fewer logical CPUs, but they are twice as fast.
Nope, hyperthreading is just a method to allow better multitasking for the CPU. If there are two threads trying to get CPU time, HT will do its best to staisfy them both; if there is only one thread, it will just get all of the CPU time - there is still only one physical processing unit, afaik.
Yeah, 1 physical, but hyper-threading turns it into 2 logical processors each running at half speed. This is what I have observed from my brother's hyper-threading computer. Turning it off helped single-threaded applications run faster.
Hmm, even dropping the server to 33 tickrate didn't resolve the issues... I would have thought it would seeming as it would of halved the processing power in theory.
Again, negative - it is pretending to be two logical processors, but it does not run two processors at 50% speed. A non-HT capable application will just chew up the entire processor, windows just won't know that the whole processor is in use. Thus only some caching and instance related resources are duplicated.
Your quote doesn't contain anything saying that HT-enabled CPU will run single-threaded app at the same speed as HT-disabled (but capable) CPU. I don't have such CPU so the best thing is to try both and compare performance.