Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site drivax.UUCP
Path: utzoo!linus!philabs!prls!amdimage!amdcad!amdahl!drivax!alan
From: alan@drivax.UUCP (Alan Fargusson)
Newsgroups: net.arch
Subject: Re: Stack architectures - why not?
Message-ID: <231@drivax.UUCP>
Date: Sat, 14-Sep-85 23:33:01 EDT
Article-I.D.: drivax.231
Posted: Sat Sep 14 23:33:01 1985
Date-Received: Thu, 19-Sep-85 03:16:45 EDT
References: <796@kuling.UUCP> <172@myriasa.UUCP>
Organization: Digital Research, Monterey, CA
Lines: 24

Ever hear of Hewlett-Packard? They have been making machines
with pure stack archetectures for years. I used an HP3000 when
I was in school and became very knowlegable about the machine.
The only register that the user had real access to was an index
register. Data addressing was relative to the base of data (DB),
or the stack pointer (S), or the frame pointer (Q). The machine
cost about the same as a PDP-11/70, and performed about the same
as far as I can tell. It had one advantage over the PDP-11 in
that it was a segmented machine that could have more than one
code segment, so you didn't need overlays. It had shared run time
libraries that HP called Segmented Libraries. In fact most of the
operating system was just a shared run time library. Code segments
had two attributes, preveliged/non-preveliged and callable/non-callable.
The protection of the system was implemented using these features.

I guess I will stop rambeling now. I really liked this machine.
Now if HP had just ported UNIX to it.

I think that the HP9000 systems have a similar archetecture.
-- 

Alan Fargusson.

{ ihnp4, amdahl, mot }!drivax!alan


Brought to you by Super Global Mega Corp .com