Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!uunet!microsoft!brianw
From: brianw@microsoft.UUCP (Brian Willoughby)
Newsgroups: comp.sys.apple
Subject: Re: Apple's committment to the // line
Message-ID: <8197@microsoft.UUCP>
Date: 24 Oct 89 23:19:01 GMT
References: <8910202319.AA14029@trout.nosc.mil>
Reply-To: brianw@microsoft.UUCP (Brian Willoughby)
Organization: Microsoft Corp., Redmond WA
Lines: 52

In article <8910202319.AA14029@trout.nosc.mil> sysop@pro-generic.cts.com (Matthew Montano) writes:
>Nothing to be said of UNIX, every implementation of UNIX on a 386 is something
>to behold and is a proper use of 386 machines. Although the barrier/boundry
>between MS-DOSed 386's and UNIX based 386 boxes is dissappearing quickly, it
>is still there and in my mind there is no comparison. A 386 is an almost
>perfect architecture to run UNIX off of, but MS-DOS? WHY? UNIX is also an
>operating system that was originally designed for mini computers, and when you
>produce a machine (386 box) that is basically what a mini was a few years ago,
>it makes sense that UNIX runs on it handidly. GS/OS have grown from the
>opposite end of the spectrum... remember that.

"almost perfect architecture" ? how do you mean?  Sure its fast, but as
unweildy as the 386 is, I would hardly say that it has any UNIX oriented
features.  What makes the 386 any better than a number of other currently
available processors?  The only reason the 386 is being used is because
of the sheer number of 386-based machines out there.  MS-DOS sold the x86
to the computer public, and the various UNIX vendors are merely catering
to a large installed base.  Don't confuse the performance of any 386 box
by assuming that UNIX favors the 386, because it doesn't.  Limited, non-
orthagonal registers, backward compatible segment registers, crippled I/O
addressing modes, the inability to use BOTH 16 AND 32 bit data sizes AT
THE SAME TIME without using instruction prefix modifiers, and an
instruction set which was not even entirely designed within the same
decade - all of these "features" make the 386 a poor choice, if one could
only ignore the *sales* of hardware and concentrate on just the
*capabilities* of the hardware.  But no one is designing UNIX systems
from an ideal point of view, so they must adapt to the most "popular"
processor instead of the best suited.  It is true that new 386 machines
are being sold as UNIX boxes (thus seemingly voiding the "existing user
base" argument), but they enjoy greater sales because of the ability to
run the large number of existing and proven MS-DOS applications.
Machines based of processors better suited to UNIX sell in fewer numbers
because they cannot offer multi-operating system support for backward
compatibility.

What does this have to do with the Apple II?  Well, you see what can be
done with *one* processor which isn't particularly suited to UNIX - just
think about what the same amount of effort could do with another large
installed user base!  It's not any more difficult to implement UNIX with
the 65xxx than with the 80x86, it's just that the effort has not yet been
applied.  Of course, there isn't much demand for a UNIX II, but until it
exists, there can't BE any demand.  And demand is the thing that brings
us UNIX on the 386.

These are not the opinions of my employer.
They are not interested in the Apple II line.

Brian Willoughby
UUCP:           ...!{tikal, sun, uunet, elwood}!microsoft!brianw
InterNet:       microsoft!brianw@uunet.UU.NET
  or:           microsoft!brianw@Sun.COM
Bitnet          brianw@microsoft.UUCP