Return to BSD News archive
#! rnews 2942 bsd Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.mel.connect.com.au!news.mira.net.au!vic.news.telstra.net!act.news.telstra.net!psgrain!usenet.eel.ufl.edu!news.mathworks.com!enews.sgi.com!www.nntp.primenet.com!nntp.primenet.com!howland.erols.net!newsfeed.internetmci.com!super.zippo.com!zdc!szdc!szdc-e!news From: John Dyson <dyson@freebsd.org> Newsgroups: comp.unix.bsd.freebsd.misc,comp.unix.solaris,comp.os.linux.misc Subject: Re: Benchmarking different Unix Operating Systems Date: Mon, 09 Sep 1996 09:04:10 -0500 Organization: John S. Dyson's home machine Lines: 37 Distribution: inet Message-ID: <323423DA.41C67EA6@freebsd.org> References: <aak2.842008017@Isis.MsState.Edu> <50p41e$1ie@nnrp1.news.primenet.com> <32305b37.590852758@news.intellistar.net> <aak2.842255270@Isis.MsState.Edu> <Pine.GSO.3.95.960909083306.1195F-100000@mercury.kosone.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.0Gold (X11; U; FreeBSD 2.2-CURRENT i386) Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:26944 comp.unix.solaris:81992 comp.os.linux.misc:128035 Steve Cole wrote: > > What benchmark were you using? Regardless, I can't understand how your > Solaris box could crash as you're saying - we have had our Solaris x86 box > stay up as long as 60 days without incident (P133, 128MB RAM, 12GB, > extremely heavy usage). In fact, considering the PC hardware I'm quite > surprised how well it hangs together. Performance isn't on par with the > Ultra 1 by a long shot, but it's not so bad. Linux on the same box melted > under the load... which is why I'm skeptical <grin> > > We're using a 3Com card, Adaptec SCSI-II controller, Pentium 133. > One should not be too awful swayed by low level benchmarks under very light load, unless that is how the system is being used. Simply looking at lmbench, spec, dhrystone, etc and judging a machine based only on that is a bit silly. But, it appears that those who try to use the machine under real loading conditions are doing the right thing. I usually use benchmarks like lmbench for a first-pass evaluation. Quality is not measured directly by how quickly a process can get in/out of kernel mode, for example. Behavior under loading is another measure of quality. I dare say that FreeBSD/Linux are not always "better" than Solaris, even though some low level benchmarks might try to say so. I spend time making decisions about whether to fix a low level performance problem, but the side effect is to slow down heavy load perf and vice versa. There are also such tradeoffs in reliability. It appears many times that when one makes a judgement about an OS, it can be a difference in the way that it handles hardware or an errant driver. That is part of the game, unfortunately :-(. Sig-11 and system hangs do not necessarily point to an OS bug. John dyson@freebsd.org