Date: Mon, 24 Nov 1997 15:12:04 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: garbanzo@hooked.net Cc: jmb@freebsd.org, freebsd-questions@freebsd.org Subject: Re: performance differences Message-ID: <199711242012.PAA02421@dyson.iquest.net> In-Reply-To: <Pine.BSF.3.96.971123192821.13829B-100000@zippy.dyn.ml.org> from Alex at "Nov 23, 97 11:07:15 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
Alex said: > > > On Sun, 23 Nov 1997, Jonathan M. Bresler wrote: > > > > > > check out : > > > > > http://www.techweb.com/se/directlink.cgi?INW19970901S0125 > > > > > for the result of performance tests on linux freebsd and windowsNT. > > > > > > > > FreeBSD was using 1/2 the memory used by the other systems. > > > > FreeBSD was conservative in determinghte amount of memory > > > > installed. The amount used is reported in the startup messages, > > > > which the reviewers must have missed. > > > > > > > > they did not do the minimum of building a kernel to use > > > > the larger amount of memory available > > > > > > The whole point of this was to test a machine "out of the box". I.E. doing > > > as little customization as possible. If they had tested with 3.0 (a.k.a. > > > -current) which sizes >64M OTH, methinks that FreeBSD would have come out > > > on top. > > > > so they said, yet they also claimed that there was no warning > > message...indicating a change would have been made if they > > had read the boot messages....very confusing....but so are > > the numeric results....thye published 100 users and 3000 user > > for most systems, 10 users for NT, and 200 user for FreeBSD. > > > > i would have preferred numbers for all os'es for hte same number of > > users. > > LOL, good luck getting NT to support 3000 users. > > However, I noticed upon reading this that Linux will automagically detect > SMP systems (and I already know NT can kinda do this), that'd be cool if > FBSD could do this too. > We could probably do that pretty easily by selecting a kernel at bootup (which is close to what NT does.) Maybe eventually, we can run exactly the same kernel, but there is necessarily more overhead in an SMP kernel. (I like the SMP's utilization of the APIC, mitigating some of the limitations of the ISA IRQ scheme. It might be nice to use them on a UP kernel.) -- John dyson@freebsd.org jdyson@nc.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199711242012.PAA02421>