*BSD News Article 48271


Return to BSD News archive

Xref: sserve news.software.nntp:14935 comp.unix.bsd.freebsd.misc:4269 comp.os.linux.advocacy:15963
Path: sserve!euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!lll-winken.llnl.gov!hookup!news.mathworks.com!solaris.cc.vt.edu!uunet!in1.uu.net!svc.portal.com!news1.best.com!shell2.best.com!not-for-mail
From: rcarter@best.com (Russell Carter)
Newsgroups: news.software.nntp,comp.unix.bsd.freebsd.misc,comp.os.linux.advocacy
Subject: Re: Sun/Solaris or Pentium/Linux for new server ?
Date: 4 Aug 1995 19:14:45 -0700
Organization: Best Internet Communications, Inc. (info@best.com)
Lines: 48
Distribution: inet
Message-ID: <3vuk6l$n2t@shell2.best.com>
References: <3vlpgk$rdk@graphite.comco.com> <3vpg3g$q35@shell2.best.com> <3vq3if$jmp@dodgson.math.psu.edu>
NNTP-Posting-Host: shell2.best.com

In article <3vq3if$jmp@dodgson.math.psu.edu>,
Dave Barr <barr@math.psu.edu> wrote:
>In article <3vpg3g$q35@shell2.best.com>,
>Russell Carter <rcarter@best.com> wrote:
>>In article <3vlpgk$rdk@graphite.comco.com>,
>>Doug McLaren <dougmc@comco.com> wrote:
>>>for a bit at home, but am far more familiar with Linux.  From personal
>>>experience, Linux's disk I/O is far faster, but Net/FreeBSD may make
>>>up for it in other areas/
>>
>>Hmm, I don't think so!  You might like to check out:
>>
>>http://www.geli.com/data/disk.perf.html
>>
>>And please point me to anything analogous on the Linux side!
>
>Talk to Larry McVoy <lm@neteng.engr.sgi.com>.  A long-time FFS bigot
>who has seen the light and now-time Linux ext2 bigot.  (and he knows
>his stuff -- he wrote the FFS clustering enhancement while he was
>at Sun)
>
>Linux's ext2 is faster than *BSD FFS, especially in the areas that count
>with respect to news (directory updates and file creation).  Linux's
>context switch times, which used to be slower than *BSD, are now almost
>the fastest of even commercial UNIXes.
>
>But anyway, don't trust me -- Larry's got real numbers.

Are these available on the web, as my data is?  I made a thorough
comparison of NetBSD 1.0, FreeBSD 2.0, and Linux 1.0.* in November
last year, and Linux was the loser by far.  (Identical hardware:
P54C-90, 64 MB, Quantum 1080S, NCR53C810)  Since I committed
to FreeBSD for my workstation cluster operating system, I haven't
publicized my Linux numbers.  I would really like to find publically
available performance data for Linux bandwidth and tps, preferably
on a variety of controllers and with interesting motherboards.  
I'm sorry, small HDPARMS don't qualify.

My FreeBSD data is open and public, and anyone is welcome to dispute
it.  I'd like to see the same for every OS out there.

Regards,
Russell
Geli Engineering		Pentium Workstation Clusters
http://www.geli.com