Re: [OT] [hlds_linux] Valve's Response to CPU Usage
You can run bf1942 servers for free, just like just about every other dedicated server title out there. James --- James Couzens My Half-Life Admin http://myHLAdmin.com - Original Message - From: [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, September 23, 2003 4:01 PM Subject: Re: [OT] [hlds_linux] Valve's Response to CPU Usage PS - one last thing all you small time server admins (like myself) out there making all your money off cs servers. 3 words. Class Action Lawsuit. If my business fails because of this and I cant find customers quick enough to make money still then i will look into this. My wife works at a lawfirm. Ok, I was going to stay out of all this, but this statement's just plain ridiculous. Look, you're running a FREE PIECE OF SOFTWARE. FREE. You've paid no money to VALVe to run the *server* (if you wanna grip about the client, that's another story). They are not obliged to provide services to you, just as you're not obliged to stick to a platform with which you're obviously unsatisfied. HLDS is one of the few multiplayer games available where the server is not required to be commercially licensed (i.e., no CD key, no purchases necessary). That VALVe have supported your industry (your industry being server rental, I'll assume) for years now, for ABSOLUTELY NO MONEY, is pretty friggin spectacular. I don't know of another commercial entity out there that has chosen to devote as many resources to a free product. I know they're not doing it out of some sacrificial altruistic motives here, but to claim injury from a free piece of software, and to further suggest litigation over it, is just ludicrous. If you want a leg to stand on, go shell out the $40-50 to run a server instance for Battlefield 1942. -- Tim ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: RE:[OT] [hlds_linux] Valve's Response to CPU Usage
* Simon Garner [2003-09-24 18:06]: I wouldn't expect to see Valve fix it until somebody can pinpoint that the problem is reproducable and quantifiable on X hardware with Y configuration running Z version of HLDS with ABC options. As long as we keep sending mixed messages to them, they won't know where to start looking, and nothing will be done. *sighs* ... maybe the third message I'll actually send (the rest are indefinetly postponed) ... anyway ... Thoughts: - People with P4 and HT, might want to use a kernel = 2.4.17. Why? Because it was the first kernel apparently that actually tried to exploit HT. Idealy 2.6.x will perform better with HT's. - People in general, you will want a kernel with pre-emptive and low-latency patches. Why? Because .. read below. - 2.6.x kernels will (and SHOULD), perform differently. Why? Lots of things changed. Most important is the scheduler. - People with SMP boxes, you'll want = 2.4.20, play with the 2.6.x series though. Why? .. see previous. Since HT is like SMP, you will NOT notice a big difference unless the code you are running, takes advantage of some form of threads. The HL engine does _not_ do this. The SMP debate, has been hashed here many times before [1]. People running multiple servers on a box, should notice a bit of a boost (over UP), assuming you are running a kernel = 2.4.20, again play with 2.6.x. A long standing (at least with 1.5) problem, is latency. Specificaly, latency in syscalls, and u/nsleep. As a solution for 1.4/5, I replaced sleep with yeilds in the server. As a result, pings were significantly lower, but CPU usage reported by programs like TOP, were inaccurate[2]. There are other methods I've thought of since. Given the changes in 2.6.x, it would be in ones best interest to look at those scheduler changes. Specificaly, run-queue per CPU vs. one global run-queue. Previous to this message someone else offered to write a database to start collecting more performance numbers. Curious if valve is doing this internaly already. As a side note, try renicing your servers and see if that helps. If so, it may still be a scheduling/latency issue. For the _client_, changing the priority to real-time, resulted in a good 20fps gain. No, the box is not over loaded, and no, I don't have anything else running in the back. [1] - Ideal code: each client has a thread, thread handles network IO, making copious use of shared mem with main thread. - Existing Code: *afaik* - one loop, client packets are processed as received. No threads. Might explain why HP and LP's don't play well. [2] - Top reports ambiguous data like %90 cpu. Well, what is that 90 percent of? A pii's 90 percent, is a hella lot less cpu then an Athlon XP 2400's. Same thing applies to the load. Bottom line, unless you know what you are really seeing, and not looking for eye candy, utils like top are useless to you. I don't know what HL's own stat uses to measure CPU with, but it could be the same as top and hense inaccurate. ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: RE:[OT] [hlds_linux] Valve's Response to CPU Usage
lol - ok then... - Original Message - From: [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Wednesday, September 24, 2003 12:29 AM Subject: RE: RE:[OT] [hlds_linux] Valve's Response to CPU Usage Our issues arised after seeing hlds 3111d on Intel vs. the AMD 3111d binaries - the difference was like night and day - AMD blew the Intels away - this is before steam was released and I brought it to Alfreds attention. He said he would do some tweaking on the compiling pararms and if you look at 3111d on Intel vs. Steam on Intel - Steam wins. Not my place to even try to say what their development platforms are, not sure if they would want to release that info or not - not up to me. The good in us testing is providing our results to them agains the results of everyone else - so maybe they can see / find whatever the issues are - because I'm sure people are not just pulling all this outta the air. So like I said, if you are not having the same issues as other (i.e. horrid cpu on BOTH amd and intel) why are they using you to test. It's because they do not plan on fixing the hlds binary overall. They are trying to get the intel to the ungodly amd numbers and leave it at that. So they are NOT trying to fix the cpu issues as you previously said. They are trying to fix the intel issues. I seriously doubt all these issues are caused by which compiler they are using. It is a cod issue. If it was a compiler issue the cpu usage wouldn't keep getting worse with each release. There is not enough new features to justify the cpu increase from 3.1.1.0 3.1.1.1 steam. If valve is not willing to rewrite some code then the cpu issues will never be fixed. Jeremy ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: RE:[OT] [hlds_linux] Valve's Response to CPU Usage
--Original Message- -From: [EMAIL PROTECTED] -[mailto:[EMAIL PROTECTED] Behalf Of 0x2e -Sent: Wednesday, September 24, 2003 5:52 AM -To: [EMAIL PROTECTED] -Subject: Re: RE:[OT] [hlds_linux] Valve's Response to CPU Usage . - -As a side note, try renicing your servers and see if that helps. If so, -it may still be a scheduling/latency issue. For the _client_, -changing the -priority to real-time, resulted in a good 20fps gain. No, the box is not -over loaded, and no, I don't have anything else running in the back. probably a dumb question, but how are you changing the client priority? kev ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
Simon Garner wrote: The only problem will be where to get the CPU usage data from. To make the results really meaningful we need to have people posting numbers from older versions of HLDS (3.1.1.0) as well for comparison What does legacy software have to do with it? We're faced with the current version that has the same problems that previous versions had, though to a slightly greater degree. I think we ought to leave the old behind and concentrate on the current, since that's all Valve's really going to care about. -- Eric (the Deacon remix) ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
deacon wrote: Would you please explain that from a thorough and factual basis that I can understand? Nom matter how rudimentarily done, is that even possible? ;) ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
joe_null wrote: Would you please explain that from a thorough and factual basis that I can understand? Nom matter how rudimentarily done, is that even possible? ;) Yes, it is. Quite so :P -- Eric (the Deacon remix) ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
--Original Message- -From: [EMAIL PROTECTED] -[mailto:[EMAIL PROTECTED] Behalf Of -[EMAIL PROTECTED] -Sent: Tuesday, September 23, 2003 6:28 PM -To: [EMAIL PROTECTED] -Subject: Re: [OT] [hlds_linux] Valve's Response to CPU Usage - -How long is valve gonna sit around with their dicks in their -hands thinking -we are either making this up or just not caring? I cant afford to wait 6 -months for valve to fix this. If they are getting low cpu usage on their -hardware with hlds_l then by god give us the specs down to the -make/model of -everything and lets see if they are bs'n us. - -You guys at valve gonna fix this at all? I know you read this list. We are -calling you out. We want an answer on this. - -PS - one last thing all you small time server admins (like -myself) out there -making all your money off cs servers. 3 words. Class Action Lawsuit. If my -business fails because of this and I cant find customers quick enough to -make money still then i will look into this. My wife works at a lawfirm. froma previous email from alfred: These are the 3 machines we actively test with. None have revealed anywhere near the CPU load other people report. Build Machine: gcc -v Reading specs from /usr/local/lib/gcc-lib/i686-pc-linux-gnu/2.95.3/specs gcc version 2.95.3 20010315 (release) uname -a Linux linuxbuild2 2.2.5 #1 Fri Apr 2 16:37:56 MEST 1999 i686 unknown SuSE Linux 6.1 (i386) cat /proc/cpuinfo processor : 0 vendor_id : AuthenticAMD cpu family : 6 model : 8 model name : AMD Athlon(TM) XP 2200+ stepping: 0 cpu MHz : 1795.387510 fdiv_bug: no hlt_bug : no sep_bug : no f00f_bug: no fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr 6 mce cx8 9 sep 12 pge 14 cmov fcmov 17 22 mmx 24 30 3dnow bogomips: 1789.13 GNU C Library production release version 2.0.7 64 bit build machine: gcc -v Thread model: posix gcc version 3.2.2 (SuSE Linux) uname -a Linux 64bitcompiler 2.4.19 #1 Wed Apr 30 15:17:44 UTC 2003 x86_64 unknown UnitedLinux 1.0 (AMD64) cat /proc/cpuinfo processor : 0 vendor_id : AuthenticAMD cpu family : 15 model : 4 model name : Athlon HX stepping: 0 cpu MHz : 1595.496 cache size : 1024 KB fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext lm 3dnowext 3dnow bogomips: 3185.04 TLB size: 1088 4K pages clflush size: 64 address sizes : 40 bits physical, 48 bits virtual power management: ts ttp GNU C Library stable release version 2.2.5 Test1: uname -a Linux alfred_linux 2.4.19-16mdkcustom #3 Fri Feb 21 14:26:04 PST 2003 i686 unknown unknown GNU/Linux Mandrake Linux release 9.0 (dolphin) for i586 cat /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 6 model : 5 model name : Pentium II (Deschutes) stepping: 1 cpu MHz : 400.914 cache size : 512 KB fdiv_bug: no hlt_bug : no f00f_bug: no coma_bug: no fpu : yes fpu_exception : yes cpuid level : 2 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr bogomips: 799.53 GNU C Library stable release version 2.2.5 dual_bereta_r0x wrote: Alfred Reynolds wrote: The stats command uses the value from /proc/pid/stat , which is the same value that top uses. Perhaps the difference you are encountering is due to the sampling intervals (hlds smoothes the usage over a 5 second window but top simply shows the instantaneous value). - Alfred Would you mind to send us what are the base distro used by Valve to test/deploy hlds? I mean, kernel version, glibc, utils, and stuff. Even if we hack all the possible distros and kernels, we could have YOURS as a base value. ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
I dunno, IBM spends a lot of time and effort on Linux these days.. But anyway, Tim is correct. Go read the EULA you accepted when you installed HLDS. Theres a little section in there limiting Valves liability to the cost of the product. i.e, nothing. -EvilGrin -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED] Sent: Wednesday, September 24, 2003 12:01 AM To: [EMAIL PROTECTED] Subject: Re: [OT] [hlds_linux] Valve's Response to CPU Usage PS - one last thing all you small time server admins (like myself) out there making all your money off cs servers. 3 words. Class Action Lawsuit. If my business fails because of this and I cant find customers quick enough to make money still then i will look into this. My wife works at a lawfirm. Ok, I was going to stay out of all this, but this statement's just plain ridiculous. Look, you're running a FREE PIECE OF SOFTWARE. FREE. You've paid no money to VALVe to run the *server* (if you wanna grip about the client, that's another story). They are not obliged to provide services to you, just as you're not obliged to stick to a platform with which you're obviously unsatisfied. HLDS is one of the few multiplayer games available where the server is not required to be commercially licensed (i.e., no CD key, no purchases necessary). That VALVe have supported your industry (your industry being server rental, I'll assume) for years now, for ABSOLUTELY NO MONEY, is pretty friggin spectacular. I don't know of another commercial entity out there that has chosen to devote as many resources to a free product. I know they're not doing it out of some sacrificial altruistic motives here, but to claim injury from a free piece of software, and to further suggest litigation over it, is just ludicrous. If you want a leg to stand on, go shell out the $40-50 to run a server instance for Battlefield 1942. -- Tim ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux --- Incoming mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.520 / Virus Database: 318 - Release Date: 18/09/2003 --- Outgoing mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.520 / Virus Database: 318 - Release Date: 18/09/2003 ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
-Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Matthew Donnon Sent: 24 September 2003 00:27 To: [EMAIL PROTECTED] Subject: RE: [OT] [hlds_linux] Valve's Response to CPU Usage HLDS shows the CPU usage for the processor it currently resides on. (ie just one) damn thats clever, didnt occur thats what it's showing Ive just tested 1.6 on a dual 2.4ghz windows 2000 server : Sys_ticrate 100: CPU InOut Uptime Users FPSPlayers 34.54 88.42 113.58 250 300 92.71 19 33.72 91.46 119.78 250 300 92.25 19 25.00 89.28 117.30 250 300 93.17 19 25.00 85.09 111.37 250 300 93.21 19 28.91 82.59 105.89 250 300 100.39 19 29.53 80.77 103.04 250 300 87.23 19 27.69 80.03 100.04 250 300 85.20 19 Sys_ticrate 1000: CPU InOut Uptime Users FPSPlayers 39.06 86.77 125.99 252 301 254.00 18 40.23 88.21 132.32 252 301 203.00 18 34.38 86.23 131.11 252 301 230.70 18 38.91 87.44 138.51 252 301 346.05 18 50.00 93.13 143.64 252 301 256.06 18 40.10 87.81 129.61 252 301 260.15 18 39.06 81.82 118.45 252 301 206.69 18 Note : there are 3 other 20 player servers running on the box. Two are almost full (18/19 players), the other one just has 8 players. Currently there are 44 players total. Total CPU usage according to taskmgr is 45/50% Regards, James so what we are seeing here is that the standard sys_ticrate (100) is chewing between 25-35% per chip (xeon i assume) for a 19 player server. Once ticrate is pushed to 1000 this rises to 35-50% which is understandable. I would appreciate feedback as to the benefits of increasing ticrate are for the client, and is it that significant? I'm also assuming that the 8 player server is using very little cpu ;-) Now I'm sure the numbers are around here somewhere, but being on my webmail client I cant get to them, so how does a windows 1.6 server compare to a linux/bsd one? Matt ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
Ignore the last mail - clicked send by accident! I would appreciate feedback as to the benefits of increasing ticrate arefor the client, and is it that significant? Generally the higher the FPS of the server the better performance you get ingame. Lower more stable pings, and the infamous bullet reg tends to improve. So how does a windows 1.6 server compare to a linux/bsdone? Here are some stats from a single 2.4 - with nothing else running, and with sys_ticrate 100. Slackware 8.2 with 2.5.75 kernel. CPU InOut Uptime Users FPSPlayers 44.67 52.85 67.47 91283 90.92 14 45.00 51.42 65.07 91283 100.47 14 44.60 51.60 64.11 91283 90.92 14 41.00 52.24 65.36 91283 100.02 14 42.00 51.67 63.74 91283 90.74 14 42.67 50.97 63.41 91283 83.49 14 42.67 51.73 63.90 91283 90.92 14 43.25 51.10 63.03 91283 91.12 14 43.80 50.12 61.70 91283 83.35 14 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Matthew Donnon Sent: 24 September 2003 00:27 To: [EMAIL PROTECTED] Subject: RE: [OT] [hlds_linux] Valve's Response to CPU Usage HLDS shows the CPU usage for the processor it currently resides on. (ie just one) damn thats clever, didnt occur thats what it's showing Ive just tested 1.6 on a dual 2.4ghz windows 2000 server : Sys_ticrate 100: CPU InOut Uptime Users FPSPlayers 34.54 88.42 113.58 250 300 92.71 19 33.72 91.46 119.78 250 300 92.25 19 25.00 89.28 117.30 250 300 93.17 19 25.00 85.09 111.37 250 300 93.21 19 28.91 82.59 105.89 250 300 100.39 19 29.53 80.77 103.04 250 300 87.23 19 27.69 80.03 100.04 250 300 85.20 19 Sys_ticrate 1000: CPU InOut Uptime Users FPSPlayers 39.06 86.77 125.99 252 301 254.00 18 40.23 88.21 132.32 252 301 203.00 18 34.38 86.23 131.11 252 301 230.70 18 38.91 87.44 138.51 252 301 346.05 18 50.00 93.13 143.64 252 301 256.06 18 40.10 87.81 129.61 252 301 260.15 18 39.06 81.82 118.45 252 301 206.69 18 Note : there are 3 other 20 player servers running on the box. Two are almost full (18/19 players), the other one just has 8 players. Currently there are 44 players total. Total CPU usage according to taskmgr is 45/50% Regards, James so what we are seeing here is that the standard sys_ticrate (100) is chewing between 25-35% per chip (xeon i assume) for a 19 player server. Once ticrate is pushed to 1000 this rises to 35-50% which is understandable. I would appreciate feedback as to the benefits of increasing ticrate are for the client, and is it that significant? I'm also assuming that the 8 player server is using very little cpu ;-) Now I'm sure the numbers are around here somewhere, but being on my webmail client I cant get to them, so how does a windows 1.6 server compare to a linux/bsd one? Matt ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
A lot of us are out here trying to help Valve - they only have so many people to tend to the 1's of people that use this software...We're testing new binaries, experimenting with different things to provide them some input. Instead of weathered threats, how about some support. Yeah - I was frustrated as hell myself at all this and yes as a hosting provider - we've invested alot of equipment/bandwidth into this deal... Knowing the software was free... The only choice we have is to either shutup or help. My wife doesn't work at a law firm, but I know many Attorneys and they'd laugh at those 3 words. Think * what can you do to help solve the problems at hand - what input can you provide? They're not doing this to be asses, but to improve and succeed what they've already done... Gotta let go of the old to bring in the new - :-) Like a guy at Valve says: Good things come to those that wait:-) Britt - Original Message - From: [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, September 23, 2003 6:01 PM Subject: Re: [OT] [hlds_linux] Valve's Response to CPU Usage PS - one last thing all you small time server admins (like myself) out there making all your money off cs servers. 3 words. Class Action Lawsuit. If my business fails because of this and I cant find customers quick enough to make money still then i will look into this. My wife works at a lawfirm. Ok, I was going to stay out of all this, but this statement's just plain ridiculous. Look, you're running a FREE PIECE OF SOFTWARE. FREE. You've paid no money to VALVe to run the *server* (if you wanna grip about the client, that's another story). They are not obliged to provide services to you, just as you're not obliged to stick to a platform with which you're obviously unsatisfied. HLDS is one of the few multiplayer games available where the server is not required to be commercially licensed (i.e., no CD key, no purchases necessary). That VALVe have supported your industry (your industry being server rental, I'll assume) for years now, for ABSOLUTELY NO MONEY, is pretty friggin spectacular. I don't know of another commercial entity out there that has chosen to devote as many resources to a free product. I know they're not doing it out of some sacrificial altruistic motives here, but to claim injury from a free piece of software, and to further suggest litigation over it, is just ludicrous. If you want a leg to stand on, go shell out the $40-50 to run a server instance for Battlefield 1942. -- Tim ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
Same here on the CPU usage - we've had no complaints yet. Britt - Original Message - From: Kevin J. Anderson [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, September 23, 2003 6:44 PM Subject: RE: [OT] [hlds_linux] Valve's Response to CPU Usage --Original Message- -From: [EMAIL PROTECTED] -[mailto:[EMAIL PROTECTED] Behalf Of -[EMAIL PROTECTED] -Sent: Tuesday, September 23, 2003 6:28 PM -To: [EMAIL PROTECTED] -Subject: Re: [OT] [hlds_linux] Valve's Response to CPU Usage - -How long is valve gonna sit around with their dicks in their -hands thinking -we are either making this up or just not caring? I cant afford to wait 6 -months for valve to fix this. If they are getting low cpu usage on their -hardware with hlds_l then by god give us the specs down to the -make/model of -everything and lets see if they are bs'n us. - -You guys at valve gonna fix this at all? I know you read this list. We are -calling you out. We want an answer on this. - -PS - one last thing all you small time server admins (like -myself) out there -making all your money off cs servers. 3 words. Class Action Lawsuit. If my -business fails because of this and I cant find customers quick enough to -make money still then i will look into this. My wife works at a lawfirm. froma previous email from alfred: These are the 3 machines we actively test with. None have revealed anywhere near the CPU load other people report. Build Machine: gcc -v Reading specs from /usr/local/lib/gcc-lib/i686-pc-linux-gnu/2.95.3/specs gcc version 2.95.3 20010315 (release) uname -a Linux linuxbuild2 2.2.5 #1 Fri Apr 2 16:37:56 MEST 1999 i686 unknown SuSE Linux 6.1 (i386) cat /proc/cpuinfo processor : 0 vendor_id : AuthenticAMD cpu family : 6 model : 8 model name : AMD Athlon(TM) XP 2200+ stepping: 0 cpu MHz : 1795.387510 fdiv_bug: no hlt_bug : no sep_bug : no f00f_bug: no fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr 6 mce cx8 9 sep 12 pge 14 cmov fcmov 17 22 mmx 24 30 3dnow bogomips: 1789.13 GNU C Library production release version 2.0.7 64 bit build machine: gcc -v Thread model: posix gcc version 3.2.2 (SuSE Linux) uname -a Linux 64bitcompiler 2.4.19 #1 Wed Apr 30 15:17:44 UTC 2003 x86_64 unknown UnitedLinux 1.0 (AMD64) cat /proc/cpuinfo processor : 0 vendor_id : AuthenticAMD cpu family : 15 model : 4 model name : Athlon HX stepping: 0 cpu MHz : 1595.496 cache size : 1024 KB fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext lm 3dnowext 3dnow bogomips: 3185.04 TLB size: 1088 4K pages clflush size: 64 address sizes : 40 bits physical, 48 bits virtual power management: ts ttp GNU C Library stable release version 2.2.5 Test1: uname -a Linux alfred_linux 2.4.19-16mdkcustom #3 Fri Feb 21 14:26:04 PST 2003 i686 unknown unknown GNU/Linux Mandrake Linux release 9.0 (dolphin) for i586 cat /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 6 model : 5 model name : Pentium II (Deschutes) stepping: 1 cpu MHz : 400.914 cache size : 512 KB fdiv_bug: no hlt_bug : no f00f_bug: no coma_bug: no fpu : yes fpu_exception : yes cpuid level : 2 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr bogomips: 799.53 GNU C Library stable release version 2.2.5 dual_bereta_r0x wrote: Alfred Reynolds wrote: The stats command uses the value from /proc/pid/stat , which is the same value that top uses. Perhaps the difference you are encountering is due to the sampling intervals (hlds smoothes the usage over a 5 second window but top simply shows the instantaneous value). - Alfred Would you mind to send us what are the base distro used by Valve to test/deploy hlds? I mean, kernel version, glibc, utils, and stuff. Even if we hack all the possible distros and kernels, we could have YOURS as a base value. ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
What map is this on? Jeremy -Original Message- Ive just tested 1.6 on a dual 2.4ghz windows 2000 server : Sys_ticrate 100: CPU InOut Uptime Users FPSPlayers 34.54 88.42 113.58 250 300 92.71 19 33.72 91.46 119.78 250 300 92.25 19 25.00 89.28 117.30 250 300 93.17 19 25.00 85.09 111.37 250 300 93.21 19 28.91 82.59 105.89 250 300 100.39 19 29.53 80.77 103.04 250 300 87.23 19 27.69 80.03 100.04 250 300 85.20 19 Sys_ticrate 1000: CPU InOut Uptime Users FPSPlayers 39.06 86.77 125.99 252 301 254.00 18 40.23 88.21 132.32 252 301 203.00 18 34.38 86.23 131.11 252 301 230.70 18 38.91 87.44 138.51 252 301 346.05 18 50.00 93.13 143.64 252 301 256.06 18 40.10 87.81 129.61 252 301 260.15 18 39.06 81.82 118.45 252 301 206.69 18 Note : there are 3 other 20 player servers running on the box. Two are almost full (18/19 players), the other one just has 8 players. Currently there are 44 players total. Total CPU usage according to taskmgr is 45/50% Regards, James ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
ouch! so the nix version is using approx 10 MORE% cpu with 25% FEWER players compared to the win32. Thats not such a good advertisement for the linux port on intel hardware. You dont perchance happen to have an AMD based machine to perform the same testing on? I do, but am lacking the pipe to test with. Alfred: All the testing machines you have listed are AMD based, or older intel base. Given that all these architectures are more efficient per clock cycle is HLDS really emphasising the calc per clock cycle disparity between intel/amd cpus? Is there any oppurtunity for Valve testing on an Intel machine? I may be able to organise the loan of a current spec machine if you are still based in Aus. Matt So how does a windows 1.6 server compare to a linux/bsdone? Here are some stats from a single 2.4 - with nothing else running, and with sys_ticrate 100. Slackware 8.2 with 2.5.75 kernel. CPU InOut Uptime Users FPSPlayers 44.67 52.85 67.47 91283 90.92 14 45.00 51.42 65.07 91283 100.47 14 44.60 51.60 64.11 91283 90.92 14 41.00 52.24 65.36 91283 100.02 14 42.00 51.67 63.74 91283 90.74 14 42.67 50.97 63.41 91283 83.49 14 42.67 51.73 63.90 91283 90.92 14 43.25 51.10 63.03 91283 91.12 14 43.80 50.12 61.70 91283 83.35 14 SNIP Ive just tested 1.6 on a dual 2.4ghz windows 2000 server : Sys_ticrate 100: CPU InOut Uptime Users FPSPlayers 34.54 88.42 113.58 250 300 92.71 19 33.72 91.46 119.78 250 300 92.25 19 25.00 89.28 117.30 250 300 93.17 19 25.00 85.09 111.37 250 300 93.21 19 28.91 82.59 105.89 250 300 100.39 19 29.53 80.77 103.04 250 300 87.23 19 27.69 80.03 100.04 250 300 85.20 19 SNIP ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
snip Alfred: All the testing machines you have listed are AMD based, or older intel base. Given that all these architectures are more efficient per clock cycle is HLDS really emphasising the calc per clock cycle disparity between intel/amd cpus? Is there any oppurtunity for Valve testing on an Intel machine? I may be able to organise the loan of a current spec machine if you are still based in Aus. We have tested with other INTEL machines (in particular a dual PIII). Optmisation options at this time seem to be limited to choice of compiler (testing suggests that gcc 3.x doesn't provide any significant increase in performance). Matt So how does a windows 1.6 server compare to a linux/bsdone? Here are some stats from a single 2.4 - with nothing else running, and with sys_ticrate 100. Slackware 8.2 with 2.5.75 kernel. CPU InOut Uptime Users FPSPlayers 44.67 52.85 67.47 91283 90.92 14 45.00 51.42 65.07 91283 100.47 14 44.60 51.60 64.11 91283 90.92 14 41.00 52.24 65.36 91283 100.02 14 42.00 51.67 63.74 91283 90.74 14 42.67 50.97 63.41 91283 83.49 14 42.67 51.73 63.90 91283 90.92 14 43.25 51.10 63.03 91283 91.12 14 43.80 50.12 61.70 91283 83.35 14 SNIP Ive just tested 1.6 on a dual 2.4ghz windows 2000 server : Sys_ticrate 100: CPU InOut Uptime Users FPSPlayers 34.54 88.42 113.58 250 300 92.71 19 33.72 91.46 119.78 250 300 92.25 19 25.00 89.28 117.30 250 300 93.17 19 25.00 85.09 111.37 250 300 93.21 19 28.91 82.59 105.89 250 300 100.39 19 29.53 80.77 103.04 250 300 87.23 19 27.69 80.03 100.04 250 300 85.20 19 SNIP ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
Here you basically say, that you won't bother to actually fix anything, since the hl-engine is about to die anyway? (Since you're only willing to change compiler, not the actual source.) And btw, I assume you use gcc 2.95 now, are you aware that it doesn't even have athlon optimizations? gcc 3.3 with -mcpu=athlon-xp -march=athlon-xp should REALLY do something about performance, give us numbers please :) - Sindre Optmisation options at this time seem to be limited to choice of compiler (testing suggests that gcc 3.x doesn't provide any significant increase in performance). ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
snip Alfred: All the testing machines you have listed are AMD based, or older intel base. Given that all these architectures are more efficient per clock cycle is HLDS really emphasising the calc per clock cycle disparity between intel/amd cpus? Is there any oppurtunity for Valve testing on an Intel machine? I may be able to organise the loan of a current spec machine if you are still based in Aus. We have tested with other INTEL machines (in particular a dual PIII). Optmisation options at this time seem to be limited to choice of compiler (testing suggests that gcc 3.x doesn't provide any significant increase in performance). I wouldn't describe the p3 architecture as anywhere near current and in fact group it with p2/athlon as a more efficient per mhz chip than p4. Specifically, have you tested with the p4 core as that seems to be the one with the most performance issues. I do not think its an optimisation issue with GCC, rather that HLDS is in particular highlighting a weakness in the P4/xeon. The P4 is the architecture which intel sacrificed significant perf per mhz so they could ramp the mhz right through the roof. Looking at many benchmarks (admittedly none that have any real relevance in game serving) the P4 needs a 500-600 mhz boost over 32bit athlon, and nearly a 1ghz boost over 64bit athlon to post similar performances. SSE2 support appears to make the biggest difference in software performance and is something I'm sure hlds lacks. Matt ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
On Wednesday, September 24, 2003 1:57 PM [GMT+1200=NZT], Matthew Donnon [EMAIL PROTECTED] wrote: I wouldn't describe the p3 architecture as anywhere near current and in fact group it with p2/athlon as a more efficient per mhz chip than p4. Specifically, have you tested with the p4 core as that seems to be the one with the most performance issues. I think the poor performance of the P4 is a totally separate issue and there's little Valve can do about this. If you're running P4s then that's your choice. The problem remains that on AMD and P3 systems the current Linux HLDS performs worse than the Windows version, and worse than previous versions. -Simon ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
I would never sue for something provided for free. They could have denied peoples use of their software for the purpose of making money. - Original Message - From: [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, September 23, 2003 6:28 PM Subject: Re: [OT] [hlds_linux] Valve's Response to CPU Usage Valve has seriously f*cked us all. Strong statement but its true. I was able to run like 8 16 player servers on my p4 3.06ghz with HT. I have optimized kernel and all that. now im getting complaints when they fill up and i only have 4 running. I check usage and its at 90% on each processor that top shows. How are we supposed to stay alive on that kinda performance? How long is valve gonna sit around with their dicks in their hands thinking we are either making this up or just not caring? I cant afford to wait 6 months for valve to fix this. If they are getting low cpu usage on their hardware with hlds_l then by god give us the specs down to the make/model of everything and lets see if they are bs'n us. You guys at valve gonna fix this at all? I know you read this list. We are calling you out. We want an answer on this. PS - one last thing all you small time server admins (like myself) out there making all your money off cs servers. 3 words. Class Action Lawsuit. If my business fails because of this and I cant find customers quick enough to make money still then i will look into this. My wife works at a lawfirm. - Original Message - From: James Sykes [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, September 23, 2003 5:57 PM Subject: RE: [OT] [hlds_linux] Valve's Response to CPU Usage Ive just tested 1.6 on a dual 2.4ghz windows 2000 server : Sys_ticrate 100: CPU InOut Uptime Users FPSPlayers 34.54 88.42 113.58 250 300 92.71 19 33.72 91.46 119.78 250 300 92.25 19 25.00 89.28 117.30 250 300 93.17 19 25.00 85.09 111.37 250 300 93.21 19 28.91 82.59 105.89 250 300 100.39 19 29.53 80.77 103.04 250 300 87.23 19 27.69 80.03 100.04 250 300 85.20 19 Sys_ticrate 1000: CPU InOut Uptime Users FPSPlayers 39.06 86.77 125.99 252 301 254.00 18 40.23 88.21 132.32 252 301 203.00 18 34.38 86.23 131.11 252 301 230.70 18 38.91 87.44 138.51 252 301 346.05 18 50.00 93.13 143.64 252 301 256.06 18 40.10 87.81 129.61 252 301 260.15 18 39.06 81.82 118.45 252 301 206.69 18 Note : there are 3 other 20 player servers running on the box. Two are almost full (18/19 players), the other one just has 8 players. Currently there are 44 players total. Total CPU usage according to taskmgr is 45/50% Regards, James ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
Dust2 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED] Sent: 24 September 2003 01:33 To: [EMAIL PROTECTED] Subject: RE: [OT] [hlds_linux] Valve's Response to CPU Usage What map is this on? Jeremy -Original Message- Ive just tested 1.6 on a dual 2.4ghz windows 2000 server : Sys_ticrate 100: CPU InOut Uptime Users FPSPlayers 34.54 88.42 113.58 250 300 92.71 19 33.72 91.46 119.78 250 300 92.25 19 25.00 89.28 117.30 250 300 93.17 19 25.00 85.09 111.37 250 300 93.21 19 28.91 82.59 105.89 250 300 100.39 19 29.53 80.77 103.04 250 300 87.23 19 27.69 80.03 100.04 250 300 85.20 19 Sys_ticrate 1000: CPU InOut Uptime Users FPSPlayers 39.06 86.77 125.99 252 301 254.00 18 40.23 88.21 132.32 252 301 203.00 18 34.38 86.23 131.11 252 301 230.70 18 38.91 87.44 138.51 252 301 346.05 18 50.00 93.13 143.64 252 301 256.06 18 40.10 87.81 129.61 252 301 260.15 18 39.06 81.82 118.45 252 301 206.69 18 Note : there are 3 other 20 player servers running on the box. Two are almost full (18/19 players), the other one just has 8 players. Currently there are 44 players total. Total CPU usage according to taskmgr is 45/50% Regards, James ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
this stats output for a default rh9 install on a dual xeon 2ghz with 1.5 gb of ram. CPU InOut Uptime Users FPSPlayers 48.50 37.65 45.28 40 106 49.84 17 map is aztec jake Matthew Donnon wrote: snip Alfred: All the testing machines you have listed are AMD based, or older intel base. Given that all these architectures are more efficient per clock cycle is HLDS really emphasising the calc per clock cycle disparity between intel/amd cpus? Is there any oppurtunity for Valve testing on an Intel machine? I may be able to organise the loan of a current spec machine if you are still based in Aus. We have tested with other INTEL machines (in particular a dual PIII). Optmisation options at this time seem to be limited to choice of compiler (testing suggests that gcc 3.x doesn't provide any significant increase in performance). I wouldn't describe the p3 architecture as anywhere near current and in fact group it with p2/athlon as a more efficient per mhz chip than p4. Specifically, have you tested with the p4 core as that seems to be the one with the most performance issues. I do not think its an optimisation issue with GCC, rather that HLDS is in particular highlighting a weakness in the P4/xeon. The P4 is the architecture which intel sacrificed significant perf per mhz so they could ramp the mhz right through the roof. Looking at many benchmarks (admittedly none that have any real relevance in game serving) the P4 needs a 500-600 mhz boost over 32bit athlon, and nearly a 1ghz boost over 64bit athlon to post similar performances. SSE2 support appears to make the biggest difference in software performance and is something I'm sure hlds lacks. Matt ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
Then your simply not listening. Try grep -i lag -r logs/ If you haven't seen a change in performance running the same number of servers on the same hardware, then your not paying attention. I been doing this since beta 5.2, I didn't just start this yesterday. What kernel are you running? What hardware? How many ACTIVE servers on that hardware? Pingboost? You say you have seen no difference, show us some of stats. Show us somethign other than, no problem here. - Original Message - From: Britt Priddy (PZGN) [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, September 23, 2003 8:21 PM Subject: Re: [OT] [hlds_linux] Valve's Response to CPU Usage Same here on the CPU usage - we've had no complaints yet. Britt - Original Message - From: Kevin J. Anderson [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, September 23, 2003 6:44 PM Subject: RE: [OT] [hlds_linux] Valve's Response to CPU Usage --Original Message- -From: [EMAIL PROTECTED] -[mailto:[EMAIL PROTECTED] Behalf Of -[EMAIL PROTECTED] -Sent: Tuesday, September 23, 2003 6:28 PM -To: [EMAIL PROTECTED] -Subject: Re: [OT] [hlds_linux] Valve's Response to CPU Usage - -How long is valve gonna sit around with their dicks in their -hands thinking -we are either making this up or just not caring? I cant afford to wait 6 -months for valve to fix this. If they are getting low cpu usage on their -hardware with hlds_l then by god give us the specs down to the -make/model of -everything and lets see if they are bs'n us. - -You guys at valve gonna fix this at all? I know you read this list. We are -calling you out. We want an answer on this. - -PS - one last thing all you small time server admins (like -myself) out there -making all your money off cs servers. 3 words. Class Action Lawsuit. If my -business fails because of this and I cant find customers quick enough to -make money still then i will look into this. My wife works at a lawfirm. froma previous email from alfred: These are the 3 machines we actively test with. None have revealed anywhere near the CPU load other people report. Build Machine: gcc -v Reading specs from /usr/local/lib/gcc-lib/i686-pc-linux-gnu/2.95.3/specs gcc version 2.95.3 20010315 (release) uname -a Linux linuxbuild2 2.2.5 #1 Fri Apr 2 16:37:56 MEST 1999 i686 unknown SuSE Linux 6.1 (i386) cat /proc/cpuinfo processor : 0 vendor_id : AuthenticAMD cpu family : 6 model : 8 model name : AMD Athlon(TM) XP 2200+ stepping: 0 cpu MHz : 1795.387510 fdiv_bug: no hlt_bug : no sep_bug : no f00f_bug: no fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr 6 mce cx8 9 sep 12 pge 14 cmov fcmov 17 22 mmx 24 30 3dnow bogomips: 1789.13 GNU C Library production release version 2.0.7 64 bit build machine: gcc -v Thread model: posix gcc version 3.2.2 (SuSE Linux) uname -a Linux 64bitcompiler 2.4.19 #1 Wed Apr 30 15:17:44 UTC 2003 x86_64 unknown UnitedLinux 1.0 (AMD64) cat /proc/cpuinfo processor : 0 vendor_id : AuthenticAMD cpu family : 15 model : 4 model name : Athlon HX stepping: 0 cpu MHz : 1595.496 cache size : 1024 KB fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext lm 3dnowext 3dnow bogomips: 3185.04 TLB size: 1088 4K pages clflush size: 64 address sizes : 40 bits physical, 48 bits virtual power management: ts ttp GNU C Library stable release version 2.2.5 Test1: uname -a Linux alfred_linux 2.4.19-16mdkcustom #3 Fri Feb 21 14:26:04 PST 2003 i686 unknown unknown GNU/Linux Mandrake Linux release 9.0 (dolphin) for i586 cat /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 6 model : 5 model name : Pentium II (Deschutes) stepping: 1 cpu MHz : 400.914 cache size : 512 KB fdiv_bug: no hlt_bug : no f00f_bug: no coma_bug: no fpu : yes fpu_exception : yes cpuid level : 2 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr bogomips: 799.53 GNU C Library stable release version 2.2.5 dual_bereta_r0x wrote: Alfred Reynolds wrote: The stats command uses the value from /proc/pid/stat , which is the same value that top uses. Perhaps the difference you are encountering is due to the sampling intervals (hlds smoothes the usage over a 5 second window but top simply
Re: [OT] [hlds_linux] Valve's Response to CPU Usage
Our tests with 2.5.74 and 2.5.75 kernels showed improved results over the 2.4 kernel series. I used 2.5 kernels for sometime during 1.5 and beta 1.6. Good performance overall. Thats why we were anxious to start testing the 2.6 series as soon as the first test 2.6 came out. - Original Message - From: James Sykes [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, September 23, 2003 7:57 PM Subject: RE: [OT] [hlds_linux] Valve's Response to CPU Usage Ignore the last mail - clicked send by accident! I would appreciate feedback as to the benefits of increasing ticrate arefor the client, and is it that significant? Generally the higher the FPS of the server the better performance you get ingame. Lower more stable pings, and the infamous bullet reg tends to improve. So how does a windows 1.6 server compare to a linux/bsdone? Here are some stats from a single 2.4 - with nothing else running, and with sys_ticrate 100. Slackware 8.2 with 2.5.75 kernel. CPU InOut Uptime Users FPSPlayers 44.67 52.85 67.47 91283 90.92 14 45.00 51.42 65.07 91283 100.47 14 44.60 51.60 64.11 91283 90.92 14 41.00 52.24 65.36 91283 100.02 14 42.00 51.67 63.74 91283 90.74 14 42.67 50.97 63.41 91283 83.49 14 42.67 51.73 63.90 91283 90.92 14 43.25 51.10 63.03 91283 91.12 14 43.80 50.12 61.70 91283 83.35 14 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Matthew Donnon Sent: 24 September 2003 00:27 To: [EMAIL PROTECTED] Subject: RE: [OT] [hlds_linux] Valve's Response to CPU Usage HLDS shows the CPU usage for the processor it currently resides on. (ie just one) damn thats clever, didnt occur thats what it's showing Ive just tested 1.6 on a dual 2.4ghz windows 2000 server : Sys_ticrate 100: CPU InOut Uptime Users FPSPlayers 34.54 88.42 113.58 250 300 92.71 19 33.72 91.46 119.78 250 300 92.25 19 25.00 89.28 117.30 250 300 93.17 19 25.00 85.09 111.37 250 300 93.21 19 28.91 82.59 105.89 250 300 100.39 19 29.53 80.77 103.04 250 300 87.23 19 27.69 80.03 100.04 250 300 85.20 19 Sys_ticrate 1000: CPU InOut Uptime Users FPSPlayers 39.06 86.77 125.99 252 301 254.00 18 40.23 88.21 132.32 252 301 203.00 18 34.38 86.23 131.11 252 301 230.70 18 38.91 87.44 138.51 252 301 346.05 18 50.00 93.13 143.64 252 301 256.06 18 40.10 87.81 129.61 252 301 260.15 18 39.06 81.82 118.45 252 301 206.69 18 Note : there are 3 other 20 player servers running on the box. Two are almost full (18/19 players), the other one just has 8 players. Currently there are 44 players total. Total CPU usage according to taskmgr is 45/50% Regards, James so what we are seeing here is that the standard sys_ticrate (100) is chewing between 25-35% per chip (xeon i assume) for a 19 player server. Once ticrate is pushed to 1000 this rises to 35-50% which is understandable. I would appreciate feedback as to the benefits of increasing ticrate are for the client, and is it that significant? I'm also assuming that the 8 player server is using very little cpu ;-) Now I'm sure the numbers are around here somewhere, but being on my webmail client I cant get to them, so how does a windows 1.6 server compare to a linux/bsd one? Matt ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
RE: [OT] [hlds_linux] Valve's Response to CPU Usage
Dust2 You should try this on Aztec or similar map because it's the one causing the issues, that and airstrip. Jeremy ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
