Path: utzoo!censor!geac!torsqnt!lethe!yunexus!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!bonnie.concordia.ca!thunder.mcrcim.mcgill.edu!snorkelwacker.mit.edu!apple!julius.cs.uiuc.edu!usc!samsung!uunet!auspex!guy
From: guy@auspex.auspex.com (Guy Harris)
Newsgroups: comp.arch
Subject: Re: Sun != Open Archtecture
Keywords: 386 sun 486 sparc
Message-ID: <5177@auspex.auspex.com>
Date: 11 Jan 91 18:16:29 GMT
References: <2199@aber-cs.UUCP>
Organization: Auspex Systems, Santa Clara
Lines: 19

>Actually (rumour mongering here) the 386/486 SUNs were killed off by SUN
>because they were technically competitive with the SPARC machines,

"Technically competitive" in what sense?  The 25 MHz 386 machine, at
least on the Stanford integer benchmarks, was about as fast as the
Sun-3/2xx (25 MHz 68020 machine); it was faster on the Stanford
floating-point benchmarks than the 3/2xx on which I tried the
benchmarks, but the 3/2xx machine had only a 68881, not an FPA.  (All
compilation was done with "-O", which at the time invoked just the
peephole optimizer on both machines; the global optimizer wasn't
available on the 386, and it sped things up by about 20% on the 68020.) 
I forget how well the 4/280 on which I tried them did with "-O1", but I
think that even without the global optimizer, it was at least
competitive, and probably better.

I think the 386i also had some performance problems with its disk
controller.

The 486 machine might have been more competitive.