*BSD News Article 92518


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!news.mira.net.au!news.netspace.net.au!news.mel.connect.com.au!munnari.OZ.AU!news.ecn.uoknor.edu!feed1.news.erols.com!news.idt.net!enews.sgi.com!fido.asd.sgi.com!neteng!lm
From: lm@neteng.engr.sgi.com (Larry McVoy)
Newsgroups: comp.unix.bsd.freebsd.misc,comp.unix.bsd.bsdi.misc,comp.sys.sgi.misc
Subject: Re: no such thing as a "general user community"
Followup-To: comp.unix.bsd.freebsd.misc,comp.unix.bsd.bsdi.misc,comp.sys.sgi.misc
Date: 1 Apr 1997 01:50:32 GMT
Organization: Silicon Graphics Inc., Mountain View, CA
Lines: 53
Message-ID: <5hppl8$9t2@fido.asd.sgi.com>
References: <331BB7DD.28EC@net5.net> <5hn00k$dio@fido.asd.sgi.com> <5hnam9$393@hoopoe.psc.edu> <5hp7p3$1qb@fido.asd.sgi.com> <5hpkr6$32u@hoopoe.psc.edu>
Reply-To: lm@slovax.engr.sgi.com
NNTP-Posting-Host: neteng.engr.sgi.com
X-Newsreader: TIN [version 1.2 PL2]
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:38215 comp.unix.bsd.bsdi.misc:6534 comp.sys.sgi.misc:29580

Peter Berger (peterb@hoopoe.psc.edu) wrote:
: YOUR RESEARCH METHODOLOGY WAS UTTERLY AND COMPLETELY BOGUS.

That's nice.  $5 says you didn't read the paper.  $10 says you haven't 
read any of the papers that reference lmbench.  

: is presenting results to me, I can know that their methodology
: was rigorous.  

Let's see.  I gathered results from a bunch of systems, carefully
documented what the systems were, wrote it up, presented it, got best
paper, went home.  Since then, noone has come forward with any suggestion
that the data was incorrect, that the benchmarks were incorrect, or that
paper was incorrect.  So what, exactly, is your frigging problem?

: Larry, the fact that your assumptions turned out to be 
: close to reality doesn't really matter.  

That's the best you can do?  That's so weak.  Let's see, you accuse me
of skewing the data, when that turns out to be false, you accuse me of
using wildly differing hardware, when that turns out to be false, you
say it doesn't matter.

I say it does matter.  I didn't bother measuring FreeBSD and Linux on
the same hardware because I really didn't care about FreeBSD vs Linux.
I did include P5@120mhz Linux numbers to have a reasonable comparison
to the P5@133 FreeBSD numbers, but that was for FreeBSD's benefit, so
they could see where to do more work.

The paper, in case you didn't notice, measured intel, sparc, alpha,
rios, parisc, etc.  It's hardly an apples to apples comparison, nor was
it intended to be.  You're all upset and whining because that's what
you wanted.  Well, if you want a comparison of all the free operating
systems on the exact same hardware, be my guest.  Go do it.

I wanted a reasonable idea of how fast a bunch of different platforms
could do a bunch of different things that I, in my own opinion, consider
to be useful things to know.  I think that is what other people wanted as
well and I think that's why the paper got picked as best of conference.
Simply because it's a bunch of useful information.

: I'm concerned when someone stands up in front of a group of his peers
: and pretends to be able to draw conclusions from hallucinatory
: research.

I believe that conclusions that I drew were: memory & cache systems are
the most important thing.  In fact, that's exactly what the paper said
and exactly what I said at the conference.  Go look.  And it is exactly
what Aaron and Margo "discovered" after working with a modified verion
of lmbench on a bunch of different platforms.

So I'd be really interested to know, buddy boy, exactly what conclusions
you are attributing to me that you consider "hallucinatory".