Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!rutgers!brl-adm!adm!EDWIN%TAMODP.BITNET@wiscvm.wisc.EDU
From: EDWIN%TAMODP.BITNET@wiscvm.wisc.EDU
Newsgroups: comp.unix.questions
Subject: RE:Cheap boxes & sanity
Message-ID: <6656@brl-adm.ARPA>
Date: Thu, 2-Apr-87 10:32:09 EST
Article-I.D.: brl-adm.6656
Posted: Thu Apr  2 10:32:09 1987
Date-Received: Sat, 4-Apr-87 13:37:32 EST
Sender: news@brl-adm.ARPA
Lines: 29

I have a couple of comments to the following:

>I'm not very expert on this level of Unix, but I have a thought I have
>to share:  is it SANE to run Unix on 80x86 processors?  The complicated
>archictecture seems oriented towards a much more low-level approach to
>programming than is associated with Unix.  (One Xenix advocate boasted
>that this Unix derviative supports *six* memory models.  I'm impressed,
>but I'm also appalled.)
>
>The only motivation for doing Unix on IBM clones seems to be the fact
>that they *are* IBM clones.
>
>Just a thought.

First, lumping 80x86 chips together in the same boat is not accurate.
The 80386 is quite different than the 80286, as it has a virtual memory
mode as well as the 80286's segmented memory addressing scheme.

Secondly, IBM PC's running Unix offer the best price/performance I
know of. A 10 Mhz AT clone with a 70 MB hard disk and Unix can be
bought for about $3000. For around $5000 one can get an 80386 machine,
and from tests I have seen, an 80386 running Xenix has about the same
processing power as a VAX 11/780 running BSD 4.2 (the main limitation
is the slower I/O on the PC). Considering the price of a VAX, an 80386
machine could be considered much more than a workstation (however,
most of the code now available has not been optimized for the 80386).

Edwin Garrett
EDWIN@TAMODP.BITNET