Newsgroups: comp.lang.smalltalk
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!ix.netcom.com!netcom.com!sehyo
From: sehyo@netcom.com (Sehyo Chang)
Subject: Re: What makes Smalltalk better than ..
Message-ID: <sehyoD20J54.BMz@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
X-Newsreader: TIN [version 1.2 PL1]
References: <pdlogan.26.0009779F@orglobe.intel.com> <Pine.A32.3.91.950104163025.39742N-100000@swim5.eng.sematech.org> <sehyoD1yp29.HC2@netcom.com> <789384998marten.marten@feki.toppoint.de>
Date: Sat, 7 Jan 1995 02:05:28 GMT
Lines: 48

Marten Feldtmann (marten@feki.toppoint.de) wrote:
: In article <sehyoD1yp29.HC2@netcom.com> sehyo@netcom.com (Sehyo Chang) writes:
: >William D. Gooch (goochb@swim5.eng.sematech.org) wrote:
: >: On Wed, 4 Jan 1995, Patrick D. Logan wrote:
: >
: >: > ....
: >: > So my question becomes what is the difference between Smalltalk and, say, 
: >: > CommonLisp?
: >
: >: As far as I know, no OS has ever been implemented in Smalltalk.  Lisp 
: >: Machines in the MIT family have their own, Lisp-based OS(s).  This has 
: >: significant implications beyond the mere fact that it has been done - 
: >: when your OS and development environment are the same thing, it means 
: >: that nothing is out-of-bounds (a two-edged sword, BTW).
: >
: >You forgot David Ungar's work in trying to implement Smalltalk in VLSI.
: >Also there was Intel's IXP432 effort.  It was 10 years too early.
: >With today's VHDL and Synthesis technology, it would be easy to design 
: >SMalltalk based RISC chip that would outperform C++.
: >

:  But as history has shown (even in the Smalltalk-history): special
: hardware is not a successfull way ... remember the Dorado, IXP432, SOAR.

:  MF

First of all except SOAR, other project was based on CISC which were
obsolete technology by time they came out.  SOAR was successful because
it led to SPARC design.

Before current generation of CAD technology(VHDL, Synthesis,simulation).
hardware design was very slow process often it took many years which meant
special hardware couldn't keep up with more general CPU.

Nowdays, you could design entire new CPu in matter of month. Chip area
and transistor are now virtuall free.

Also, you don't have to design from scratch.  You could start from well knwon
architecture such as ARM or SPARC or MIPS and add special object support
Look at HP' multimedia support in their RISC chips. 

As Smalltalk become more mainstream, I think case can be made to put critical
logic supporting objects into chips.  Concurrent Smalltalk project is another
good example.


-- Sehyo Chang

