From owner-freebsd-commit  Tue Jan  2 10:19:14 1996
Return-Path: owner-commit
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id KAA01617
          for freebsd-commit-outgoing; Tue, 2 Jan 1996 10:19:14 -0800 (PST)
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id KAA01605
          for cvs-all-outgoing; Tue, 2 Jan 1996 10:19:09 -0800 (PST)
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id KAA01598
          for cvs-ports-outgoing; Tue, 2 Jan 1996 10:19:05 -0800 (PST)
Received: from Sysiphos (Sysiphos.MI.Uni-Koeln.DE [134.95.212.10])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id KAA01582
          Tue, 2 Jan 1996 10:18:41 -0800 (PST)
Received: by Sysiphos id AA13237
  (5.67b/IDA-1.5); Tue, 2 Jan 1996 19:18:35 +0100
Message-Id: <199601021818.AA13237@Sysiphos>
From: se@zpr.uni-koeln.de (Stefan Esser)
Date: Tue, 2 Jan 1996 19:18:34 +0100
In-Reply-To: grog@lemis.de (Greg Lehey)
       "Re: cvs commit: ports/benchmarks/bytebench - Imported sources" (Jan  2, 18:27)
X-Mailer: Mail User's Shell (7.2.6 alpha(2) 7/9/95)
To: grog@lemis.de (Greg Lehey)
Subject: Re: cvs commit: ports/benchmarks/bytebench - Imported sources
Cc: CVS-committers@freefall.freebsd.org, cvs-ports@freefall.freebsd.org
Sender: owner-commit@FreeBSD.ORG
Precedence: bulk

On Jan 2, 18:27, Greg Lehey wrote:
} Subject: Re: cvs commit: ports/benchmarks/bytebench - Imported sources
} Stefan Esser writes:
} > 
} > se          96/01/01 14:57:37
} > 
} >   Branch:      benchmarks/bytebench  1.1.1
} >   Log:
} >   The BYTE benchmark suite, very popular under Linux users as it seems.
} >   (See: http://www.silkroad.com/bass/linux/bm.html for results ...)
} >   
} >   This is another benchmark that tries to give a single performance
} >   number, but without giving much thought to proper weighting of the
} >   performance numbers. So: Please handle with care ...
} 
} Is this this nonsense that iX publishes?  Do we really want it?  It's
} one of the most meaningless benchmarks I've ever seen (along with all
} the others that iX publishes).

Yes, it is that useless (from a technical perspective) benchmark,
but it might be of some use for marketing reasons :)

(For example, the 'dc' test completes within 0.16 seconds on my
486DX2/66. And this is mainly startup overhead. For this reason 
many of the Linux results were obtained using a statically compiled 
'dc' binary for a better ranking :-)


As we all know, there are soem misconceptions about *BSD, often
maintained and distributed by the Linux crowd. One of those is,
that FreeBSD comes without any reasonable documentation at all,
and you (Greg) are one of those working on that.

Another one is, that Linux is THE lightweight PC operating system,
and we might as well just send some of our results to the Linux
Benchmark archive site at Solkroad.Com (they also have some old 
Sparc and HP results).

I had tried the Byte benchmark some time ago on my 486DX2/66, and 
found that it competes with Linux DX4/100 results (and low end 
Pentium systems :).

If the transatlantic lines would not have been so fast on the 
new years day, I wouldn't have bothered to make this a port :)


A few of the Byte benchmark components are not that bad. Only 
the use of the "average" to define a ranking is quite silly.
The average is dominated by the single test component that gave
the best (relative) result.

But we might get some nice average results, send them to the 
benchmark archive, and get some good press this way :)

Regards, STefan