From owner-freebsd-hackers Sun Aug 20 20:10:17 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.FreeBSD.org (8.6.11/8.6.6) id UAA07861 for hackers-outgoing; Sun, 20 Aug 1995 20:10:17 -0700 Received: from blob.best.net (blob.best.net [204.156.128.88]) by freefall.FreeBSD.org (8.6.11/8.6.6) with ESMTP id UAA07851 for ; Sun, 20 Aug 1995 20:10:14 -0700 Received: from geli.clusternet (rcarter.vip.best.com [204.156.137.2]) by blob.best.net (8.6.12/8.6.5) with ESMTP id UAA23772; Sun, 20 Aug 1995 20:10:19 -0700 Received: (from rcarter@localhost) by geli.clusternet (8.6.11/8.6.9) id UAA17917; Sun, 20 Aug 1995 20:06:38 -0700 Date: Sun, 20 Aug 1995 20:06:38 -0700 From: "Russell L. Carter" Message-Id: <199508210306.UAA17917@geli.clusternet> To: dyson@freefall.FreeBSD.org, hasty@rah.star-gate.com Subject: Re: Why Linux? (fwd) Cc: hackers@FreeBSD.org, terry@cs.weber.edu Sender: hackers-owner@FreeBSD.org Precedence: bulk Well, I am getting a lot of questions about why FreeBSD over Linux, and because the data was too sparse I have refused to do it, but I would really like to add this information to http://www.geli.com Nothing about what John reports is news to me. Russell Curious though, I deal with low-latency networking types for a (partial) living, and I have had more than one mention the syscall.s in Linux as the model for syscall overhead. Shouldn't this show up somewhere in user land? |In fact, why don't you get someone or you can do it yourself and |make an html page. This sort of information has to go out and |reach people. | | My sound bits, | Amancio | |>>> John Dyson said: | > I just did a competitive performance analysis of FreeBSD vs. Linux. I | > can't say that FreeBSD was the CLEAR winner, but it was much faster | > in many key areas. The only place that I can see that Linux is | > architecturally faster is in the async filesystem stuff. Generally, | > FreeBSD's networking (TCP) is at least 50% faster. FreeBSD's VM | > stuff is much faster. I cannot see where people say that Linux needs | > less memory either. I ran some memory loading benchmarks on FreeBSD and | > Linux, where FreeBSD was a generic kernel and Linux was a "nice" subset | > V1.3.20. It appears that FreeBSD handles loads much better. Also, | > when running Linux I noticed an "old-friend" -- the bouncy SVR3 feel. | > Things generally appear to run a bit slower on Linux, including sequential | > file I/O. I am sure that Linux can fix up these minor performance nits, | > but it does seem to run ok. | > | > I have results on request -- I just don't want to post them yet. If there | > is a big demand, I will though.. | > | > John | > dyson@root.com | | |