Anyone knows how this patch is compares to the SD (Staircase scheduler put into 2.6.18, 2.6.18-ck1) patch?
Best regards, Patrik Arvhult --- Dan E <[EMAIL PROTECTED]> skrev: > -- > [ Picked text/plain from multipart/alternative ] > Np, I just made a few changes again and will try > this one one (the site will be down during that > reboot :P). As far as TSC, it was already enabled; > one of my changes involves removing HPET, but I have > a single CPU/single core machine, so I don't think > it's advisable for most servers out there. It > sounds like even some frequency management utils > (cpufreq) can also throw off the TSC. Hopefully by > tonight I can report whether or not this helped out. > > If any of you have tried this, let's hear how it > worked out for you :) > > Colin Silcock <[EMAIL PROTECTED]> wrote: > Awesome, thanks Dan. > > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On > Behalf Of Dan E > Sent: 03 August 2007 13:30 > To: [email protected] > Subject: RE: [hlds_linux] Possible fix for linux > performance? > > -- > [ Picked text/plain from multipart/alternative ] > Bah, I just decided to throw it up on my host. > http://dansbasement.us/.config > > Again, don't forget to check for your own > requirements (I don't even have > SCSI/RAID/GigE on here), and don't forget to apply > the realtime preemption > patch. :) I'll look into the TSC thing that was > mentioned earlier. > > Colin Silcock wrote: Wouldn't mind one too ;) > > Seems like there's a lot of interest in this, if you > do host it online that > would be cool tsuehpsyde! Otherwise I'd be happy to > host it too. > > Colin > > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On > Behalf Of Tobias Rehn > Sent: 03 August 2007 13:05 > To: [email protected] > Subject: Re: [hlds_linux] Possible fix for linux > performance? > > Hey, > > Can i also get a copy of your config file? Thanks a > lot. > > Kind regards, > Tobi > > Anton Chernev wrote: > > Hi, > > Could I also get a copy of your .config file? > > Thanks. > > > > Anton > > > > > >> -- > >> [ Picked text/plain from multipart/alternative ] > >> Send me a copy and I'll gladly host it online for > the list for testing. > >> [EMAIL PROTECTED] > >> > >> -tsuehpsyde > >> SourceKills.com > >> > >> On 8/2/07, Dan E wrote: > >> > >>> -- > >>> [ Picked text/plain from multipart/alternative ] > >>> Sent. Also, don't forget that you need to apply > the complete/realtime > >>> preemption patch that I mentioned in the > beginning of this thread. > >>> > >>> Ryan Burke wrote: I'd like a copy as well. > >>> > >>> Thanks! > >>> > >>> > >>>> -- > >>>> [ Picked text/plain from multipart/alternative > ] > >>>> Sure, I'll send it offlist. If anybody else > wants it/wants to host > >>>> > >>> it, > >>> > >>>> lemme know. Just keep in mind I did tune this > for my own system > >>>> > >>> (drivers > >>> > >>>> included). > >>>> > >>>> "Yousef A. Marafi" wrote: Can you send me your > .conf? > >>>> > >>>> On 8/2/07, Dan E wrote: > >>>> > >>>>> -- > >>>>> [ Picked text/plain from multipart/alternative > ] > >>>>> I may have actually been able to cure this on > my server o.O > >>>>> > >>> Yesterday > >>> I > >>> > >>>>> was doing lots of tweaks that I could find, > and afterwards, the time > >>>>> > >>> I > >>> > >>>>> spent definitely showed. The server did not > have any hitches, lower > >>>>> pings, no tick drops; it was absolutely > splendid :) And this was > >>>>> > >>> during > >>> > >>>>> a 10man 100tick scrim (two times). While it > may not prove the same > >>>>> > >>> for > >>> > >>>>> you, I thought that I'd share my experiences > with you guys in hopes > >>>>> > >>> that > >>> > >>>>> it can help you. > >>>>> > >>>>> First off, I am running CentOS 5 with a > custom-compiled 2.6.22 kernel > >>>>> (dynamic/tickless, 1000hz (though it doesn't > really matter with a > >>>>> tickless kernel), HPET, CFQ, and low-latency > patch). You can get the > >>>>> low latency patch here: > >>>>> > >>> http://people.redhat.com/mingo/realtime-preempt/ > >>> > >>>>> I think this may have been the trick that > actually did it, but I did > >>>>> > >>> a > >>> > >>>>> lot yesterday so I can't be sure. If you apply > it, make sure that you > >>>>> have the "Complete Preemption" mode selected > under "Processor types > >>>>> > >>> and > >>> > >>>>> features". > >>>>> > >>>>> I added some optimization flags while > compiling the kernel, but I'm > >>>>> > >>> not > >>> > >>>>> sure how much it will do (I'm not *that* big > on development): > >>>>> CXXFLAGS="-march=athlon-xp -O3 -ffast-math" > >>>>> > >>>>> I added the following entries to > /etc/sysctl.conf (easiest to do :P): > >>>>> net.core.rmem_max = 16777216 > >>>>> net.core.wmem_max = 16777216 > >>>>> net.ipv4.tcp_rmem = 4096 87380 16777216 > >>>>> net.ipv4.tcp_wmem = 4096 65536 16777216 > >>>>> > >>>>> > >>>>> I also disabled access times (for both files > and folders) on my ext3 > >>>>> root partition. > >>>>> /dev/VolGroup00/LogVol00 / ext3 > defaults,noatime,nodiratime > >>>>> 1 1 > >>>>> > >>>>> Srcds is running with nice of -10, but I'm not > sure how much that > >>>>> > >>> really > === message truncated === __________________________________________________________ Låna pengar utan säkerhet. Jämför vilkor online hos Yahoo! http://shopping.yahoo.se/c-100390123-lan-utan-sakerhet.html?partnerId=96915014 _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
