Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!henry
From: henry@utzoo.UUCP (Henry Spencer)
Newsgroups: net.arch
Subject: Re: top down vs. bottom up design [was strange sex]
Message-ID: <6968@utzoo.UUCP>
Date: Mon, 21-Jul-86 14:01:15 EDT
Article-I.D.: utzoo.6968
Posted: Mon Jul 21 14:01:15 1986
Date-Received: Mon, 21-Jul-86 14:01:15 EDT
References: <2900019@ztivax.UUCP> <802@tekig4.UUCP>, <166@ima.UUCP>
Organization: U of Toronto Zoology
Lines: 25

> I'd be interested in hearing from the RISC crowd -- how much to they tune
> their designs to the precise technology available today, and how much do
> they expect to carry over into their next generation?

An underemphasized but important point of much of the RISC work is that
near-total use of high-level languages and a semi-standardized operating
system largely eliminates hardware dependencies everywhere except in the
bowels of the kernel and the compilers.  The "architecture" these machines
present to their customers is at a higher level, almost independent of the
instruction set, register configuration, etc.  Hence continuity at the
lowest level is not that important.  (It remains useful, mind you, since
redoing the kernel guts and the compilers is not an overnight job.)

This is increasingly true in the Unix world in general, in fact.  My own
shop is about to convert from a little-endian 16-bit machine using DEC
floating-point format to a big-endian 32-bit machine using IEEE floating-
point.  The two architectures resemble each other only vaguely.  The only
real hassles I expect are from stupid and unnecessary divergence between
the two versions of the operating system.  I really don't *care* that the
underlying machine is changing in just about every way, because at the
level I work at, those changes won't show.
-- 
EDEC:  Stupidly non-standard
brain-damaged incompatible	Henry Spencer @ U of Toronto Zoology
proprietary protocol used.	{allegra,ihnp4,decvax,pyramid}!utzoo!henry