Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site sdcsvax.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!sdcsvax!davidson
From: davidson@sdcsvax.UUCP (Greg Davidson)
Newsgroups: net.unix-wizards
Subject: Re:  How is \"single-user\" done?
Message-ID: <685@sdcsvax.UUCP>
Date: Thu, 14-Feb-85 04:01:07 EST
Article-I.D.: sdcsvax.685
Posted: Thu Feb 14 04:01:07 1985
Date-Received: Fri, 15-Feb-85 06:22:45 EST
References: <8232@brl-tgr.ARPA>
Reply-To: davidson@sdcsvax.UUCP (Greg davidson)
Organization: EECS Dept. U.C. San Diego
Lines: 27
Summary: 

Before we get too excited about explaining the impossibility of running
UNIX in multiuser mode on the IBM PC, I would like to point out that
XENIX and VENIX are frequently run multi-user on the PC, and have
reasonable multi-user performance when doing, for example, general
office work (mostly editing and light database operations).

The Santa Cruz Operation (XENIX) people explained how they handle the
memory protection aspect:  User programs are limited to 16-bit data
pointers, and user tasks are always given a whole 64K segment to play
with.

It is simply not economical to use a PC with a hard disk single user.
Whether you run UNIX or DOS) the cost/user is uncompetitive.  Of
course, you can do even better if you have enough users to timeshare
a super micro.

BTW, speaking of other known impossibilities, I played with VENIX on a
DG/1 at the trade show in Dallas.  It didn't have a hard disk, yet
ran fairly nicely (noticably more responsive than with HP/UX on the
HP Integral).  I ran both vi & csh (the only big programs on the disk)
and they loaded fast and ran fine.  I've seen VENIX running on floppy
based LSI-11 systems before, so this impossibility is also fairly common.

Still, I fully believe that these feats would be impossible for a bulky
system like PC/IX, let alone my favorite UNIX, 4.2BSD.

_Greg