*BSD News Article 39066


Return to BSD News archive

Xref: sserve comp.os.386bsd.misc:4330 comp.os.linux.misc:31104
Newsgroups: comp.os.386bsd.misc,comp.os.linux.misc
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msunews!uwm.edu!cs.utexas.edu!swrinde!pipex!news.sprintlink.net!news.indirect.com!wes
From: wes@indirect.com (Barnacle Wes)
Subject: Re: How fast? [was: ... slugish ...]
Message-ID: <D0F827.MAD@indirect.com>
Sender: usenet@indirect.com (Darin Wayrynen)
Organization: Internet Direct, indirect.com
Date: Wed, 7 Dec 1994 03:22:54 GMT
References: <1994Nov28.194617.18912@system9.unisys.com> <3bf6ou$pm7@wup-gate.wup.de> <D0C929.1r2@info.swan.ac.uk>
X-Newsreader: TIN [version 1.2.1 [BP] PL2.1]
Followup-To: comp.os.386bsd.misc,comp.os.linux.misc
Lines: 35

Alan Cox (iialan@iifeak.swan.ac.uk) wrote:
: But many many Linux people run well tuned 386 systems with 4Mb of RAM, and
: often very slow CPU's (SX/25 etc). So a 4Mb benchmark is useful, as is an
: 8Mb benchmark.

The point is that a benchmark on a memory-limited machine will tell you how
the systems perform on a memory-limited machine, but may not tell you much
about how they perform on a machine with adequate RAM.  In oder to guage the
performance on a given machine, you need benchmarks on a machine as close
as possible to your own configuration.

Hypothetical situation:

	386sx25, 4 M RAM, IDE disk, OS "F": 432 sandstones
	386sx25, 4 M RAM, IDE disk, OS "L": 512 sandstones
	386sx25, 4 M RAM, IDE disk, OS "N": 436 sandstones

	386dx40, 8 M RAM, IDE disk, OS "F": 812 sandstones
	386dx40, 8 M RAM, IDE disk, OS "L": 796 sandstones
	386dx40, 8 M RAM, IDE disk, OS "N": 803 sandstones

	486dx2/66, 16 M RAM, SCSI disk, OS "F": 1622 sandstones
	486dx2/66, 16 M RAM, SCSI disk, OS "L": 1476 sandstones
	486dx2/66, 16 M RAM, SCSI disk, OS "N": 1708 sandstones

Note: these are completely fictitous figures, so I expect *the
administration* will be quoting them in articles next week.

For the sake of argument, we see that OS "L" performs best on the
slower machine, OS "N" performs best on the high-end machine, and
OS "F" performs best on *my* machine, for the (again hypothetical)
"sandstone" benchmark.  These figures may or may not correspond in
anyway to actual systems, or to the "granite" benchmark. ;^)

	Wes Peters