*BSD News Article 42694


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msunews!agate!howland.reston.ans.net!news.sprintlink.net!helena.MT.net!nate
From: nate@trout.sri.MT.net (Nate Williams)
Newsgroups: comp.unix.bsd
Subject: Re: Linux vs. BSD?!
Date: 20 Feb 1995 19:48:07 GMT
Organization: SRI Intl. - Montana
Lines: 42
Message-ID: <3iarln$9s0@helena.MT.net>
References: <3i7ar8$ahv@marton.hsr.no> <3i83js$avl@ivory.lm.com> <3iaaai$72u@agate.berkeley.edu> <3iab0s$hjl@ivory.lm.com>
Reply-To: "Nate Williams" <nate@sneezy.sri.com>
NNTP-Posting-Host: trout.sri.mt.net

In article <3iab0s$hjl@ivory.lm.com>,
Peter Berger <peterb@telerama.lm.com> wrote:
>These were run on identical hardware.  The benchmark is Dhrystone 2.1, 
>which is more reliable than the 1.1 version I noticed others using as it
>is less affected by secondary cache.
>
>Here's the table.  1,000,000 runs through Dhrystone in each case.
>
>Machine         OS                      One Ds (musec.)         Ds/sec
>=======         ============            ===============         ======
>Blindman        FreeBSD2.0-R            36.0                    27,790.6
>Blindman        BSD/OS 1.1              27.9                    35,820.9
>Blindman        NetBSD 1.0-R            27.8                    36,014.4

Hmm, using the HZ value mentioned in a previous article, and my trusty
HP calculator and the 25% error rate mentioned due to HZ value
differences in NetBSD and FreeBSD.  (FreeBSD uses a larger number to get
more accurate profiling information)

36.0 * .75 = 27
36014.4 * .75 = 27010.8

Hmm, those look awfully similar to the numbers reported by
FreeBSD/NetBSD if there is indeed a discrepancy (which I am asserting). 
I believe the #'s are 100 HZ vs 128 HZ, but I don't have the information
handy so I can't say for sure.

To be completely honest, when #'s are this far out of whack for
benchmarks that compare CPU performance, you need to suspect that
something is wrong.  You should get 'similar' results on ANY OS running
on the same hardware with whetstone and dhrystone since it's a function
of the CPU, and not of the underlying OS.  SCO, Linux, Solaris,
NeXTStep, DOS, FreeBSD, NetBSD, or BSD/386 shouldn't have that much
difference.


Nate
-- 
nate@FreeBSD.org       | Do you think SRI cares what I say?  They certainly
nate@sneezy.sri.com    | don't accept responsibility for my actions, so I
work #: (406) 449-7662 | obviously don't represent them on Usenet.
home #: (406) 443-7063 | *FreeBSD core member and all around tech. weenie*