Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!usc!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!src.honeywell.com!msi.umn.edu!umeecs!umich!sharkey!cfctech!iwblsys!idayton!jimf
From: jimf@idayton.field.intel.com (Jim Fister)
Newsgroups: comp.sys.intel
Subject: Re: 386/486 SPECmarks (was: 286/386 clock frequencies, ...)
Message-ID: <1990Nov4.153835.19587@idayton.field.intel.com>
Date: 4 Nov 90 15:38:35 GMT
References: <1990Nov1.165539.1359@computing-maths.cardiff.ac.uk> <2832@crdos1.crd.ge.COM> <1990Nov3.191613.26746@watserv1.waterloo.edu>
Organization: Intel Corp./ Dayton Sales Office
Lines: 16

tom@mims-iris.waterloo.edu (Tom Haapanen) writes:
>I keep seeing a wide variance in the 486/386 speed ratio: from about 3x
>all the way down to about 1.5x (in PCMag benchmarks).  Have 386 and 486
>machines been SPECmarked?  If so, how can I find out which manufacturers'
>machines, and what the results were?

I believe that the standard Intel answer is that the 486(TM) is about 1.5 to 2x
faster than a 386(TM) at the same clock rate, depending on code and system
blah blah blah.  The math tends to run about 4x of a 387(TM).  Intel has
SPECs, but I've never seen them.  The best thing to do is to bench your
own code on the machines that you would use.  That's about the only useful
benchmark that I've ever seen.

Greetings from the rocking metropolis.

Jimf