Return to BSD News archive
Newsgroups: comp.unix.bsd.freebsd.misc,comp.os.linux Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.uwa.edu.au!classic.iinet.com.au!news.uoregon.edu!usenet.eel.ufl.edu!news.mathworks.com!tank.news.pipex.net!pipex!howland.reston.ans.net!news.sprintlink.net!dorite!not-for-mail From: root@dyson.iquest.net (John S. Dyson) Subject: Re: FreeBSD page better than Linux? Message-ID: <42oud3$ge@dyson.iquest.net> Lines: 27 Sender: news@iquest.net (News Admin) Organization: John S. Dyson's home machine References: <42cnbp$mit@blackice.winternet.com> <NEWTNews.810257869.20278.les@compu.mcs.com> <DEL02r.HnG@info.swan.ac.uk> Date: Fri, 8 Sep 1995 08:20:51 GMT Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:5816 comp.os.linux:58938 In article <DEL02r.HnG@info.swan.ac.uk>, Alan Cox <iialan@iifeak.swan.ac.uk> wrote: >> I have been running Linux for a year or so now. I was always under the >> impression that performance was one of it strengths due to its relative >> small size and simplicity. Lately someone told me that FreeBSD preforms >> a lot better under stress (high page fault rate). This is attractive to >> me because I trying to run XFree86, Netscape, gcc, and XEmacs all on my 8M >> pentium system. Needless to say, it can be somewhat of a dog at times. >> I attributed my problems to a) a lack of memory, b) a cheap disk drive and >> or controller. I didn't consider that part of my problem could be the OS. > >ftp to ftp.presence.co.uk and get a copy of kswap. On an 8Mb machine it >makes one hell of a difference to Linux performance under load. Trying to >run Xemacs+gcc+netscape+X at once on an 8Mb machine is never going to be >stunningly quick, but kswap sure makes it better. > I would be very suprised if the kswap patches improve the page fault performance. All that I could see that they probably do is improve the paging in/paging out performance (which is only part, albiet a major part of the problem.) I'll rerun my benchmarks comparing FreeBSD to Linux soon with the kswap patches in place.... It'll be *interesting*, especially with benchmarks like gausspage that don't just need faster I/O, but better paging statistics. :-). John dyson@freebsd.org