Edit Smalltalk Computers

Title:

Last edited: 10:03:38 am on 1 April 2009 by 143.107.225.16

Many people have heard about the Lisp Machines and even if they know little about them feel they illustrate the futility of language specific computers. I'll leave the following paper to dispute that: *http://pt.withy.org/publications/LispM.html*. A very complete list of Lisp Machines can be found in *http://www.andromeda.com/people/ddyer/lisp/* and more interesting information about this subject in *http://en.wikipedia.org/wiki/Lisp_machine*. There is a modern hobbiest "Lisp Machine in a FPGA" project: *http://www.aviduratas.de/lisp/lispmfpga/*.

What far fewer people are aware of is that there was an even larger effort to develop Smalltalk Machines. This page is meant to gather information about such projects.

*http://basis.desk.org:8080/ToytownStLab/Smalltalk+hardware* is a page by Reinout Heeck with information from comp.lang.smalltalk discussions. A paper in Portuguese has a comparison between several object oriented microprocessors (SOAR, FAIM-1, COM, Sword32 and Hitachi AI32): *http://www.inf.ufrgs.br/procpar/disc/cmp135/trabs/sergio/trab1/MaqOO.html*

Some of the names of the architects for the various machines came from *http://www.cs.clemson.edu/~mark/architects.html*.

The JOP (Java Optimized Processor) project has an interesting list of Java processors at *http://www.jopdesign.com/links.jsp*. While there are significant differences between Smalltalk and Java, there is enough in common that many of the ideas developed for hardware designed for one language are valid for those of the other

!!!Bytecode Machines

The most influential implementation of a computer with bytecoded instructions was the Burroughs B5000 by Bob Barton (actually, it was the B6500 which in 1966 reduced the instruction size from 12 bits to 8). Alan Kay suggested to Ed Cheadle that they use this idea in their 1967-1969 Flex Machine and this experience had a major impact in the design of the Smalltalk virtual machine later on.

*http://ed-thelen.org/comp-hist/B5000-AlgolRWaychoff.html*
*http://www.cc.gatech.edu/gvu/people/randy.carpenter/folklore/v3n5.html*
*http://www.smalltalk.org/smalltalk/TheEarlyHistoryOfSmalltalk_Abstract.html*
!!Alto
- TTL microcoded processor
- 1973
- Xerox PARC
- Chuck Thacker (primary designer), also Ed McCreight, Butler Lampson, Bob Sproull, and Dave Boggs
- internal production, 200 built
*http://citeseer.ist.psu.edu/lampson88personal.html*
*http://www.guidebookgallery.org/articles/thexeroxaltocomputer*
*http://www.pdp8.com/alto.htm*
*http://www.bitsavers.org/pdf/xerox/alto/*
!!Alto2
- TTL microcoded processor
- 1975
- Xerox PARC
*http://www.digibarn.com/collections/systems/xerox-alto/index.html*
!!Dolphin
- TTL microcoded processor
- 1978
- Xerox PARC
- Butler Lampson, Chuck Thacker, and Ron Rider
*http://www.bitsavers.org/pdf/xerox/dolphin/*
!!Dorado
- ECL microcoded processor
- 1979
- Xerox PARC
- Butler Lampson, Chuck Thacker, and Ron Rider
*http://www.iam.unibe.ch/~ducasse/FreeBooks/BitsOfHistory/BitsOfHistory.pdf* (chapter 7, pages 113 to 126)
*http://www.bitsavers.org/pdf/xerox/dorado/* (hardware manual and Smalltalk microcode)
!!Dandelion
- 2900 bitslice microcoded processor
- 1980
- Xerox PARC
*http://www.bitsavers.org/pdf/xerox/dandelion/*
!!Sword32 (Katana32)
- 1984
- University of Tokyo
- Norihisa Suzuki, Koichi Kubota, Takashi Aoki
!!Swamp
- bitslice
- 1986
- University of Toronto
- David M. Lewis, David R. Galloway, Robert J. Francis and Brian W. Thomson
*http://portal.acm.org/citation.cfm?id=960112.28710* (full text only for ACM members)
!!AI32
- 1986
- Hitachi
- T. Nojiri, S. Kawasaki and K. Sakoda
*http://portal.acm.org/citation.cfm?id=17365* (full text only for ACM members)

!!!Custom Processors

Stack machines are hard to parallelize so some efforts tried to mix a more register based design with unique features such as tagged data.

!!SOAR
- NMOS RISC with tags and register windows
- 1983
- Berkeley
- David Ungar, Alan Dain Samples and others
- Some of this work fed into the Sparc architecture
*http://portal.acm.org/citation.cfm?id=28708* (full text only for ACM members)
*http://www.grundu.net/projects/soar/*
*ftp://sunsite.berkeley.edu/pub/techreps/CSD-86-287.html*
!!COM
- 1984
- Caltech
- James T. Kajiya and William J. Dally
*http://caltechcstr.library.caltech.edu/368/*
!!Rekursiv
- set of four custom chips and dedicated memories
- 1986
- Linn (famous for the record deck)
- David Harland
- Not actually a Smalltalk machine, this was intended to run Lingo (but see *Mario's comments>http://www.merlintec.com/old-self-interest/msg00248.html* in 1992 about the other language efforts)
*http://www.brouhaha.com/~eric/retrocomputing/rekursiv/*
!!Mushroom
- TTLs and FPGAs
- 1987
- University of Manchester
- Ifor Wyn Williams, Mario I. Wolczko, and Trevor P. Hopkins
*http://www.wolczko.com/mushroom/*
!!J-Machine
- massively parallel architecture
- 1988
- MIT
- Bill Dally and others
- prototypes, three 1024 node machines were built
*http://cva.stanford.edu/projects/j-machine/*

!!!Stock Processors

After microprocessors started taking off (early 1980s) and before large programmable circuits became cheap (around 2001) it wasn't really practical to create special processor designs. The most cost effective solutions were software based implementations of the Smalltalk virtual machine on mass produced microprocessors.

!!NoteTaker
- dual 8086
- 1978
- Xerox PARC
- Doug Fairbair
- prototype, 10 built
!!Tektronix Magnolia
- 60000 8MHz
- 1981
- Tektronix
- prototype
*http://www.iam.unibe.ch/~ducasse/FreeBooks/BitsOfHistory/BitsOfHistory.pdf* (chapters 4 and 16)
!!Tektronix 4404 (Pegasus)
- 68010 10MHz
- 1984
- Tektronix
- commercial, sold for $14,950
*http://groups.google.com/group/net.works/browse_thread/thread/2b98bfa07089b809/* 
!!Tektronix 4406
- 68020
- 1986
- Tektronix
!!Merlin 2
- 68000 8MHz
- 1986
- Inova
- Jecel Mattos de Assump��o Jr
- prototype, 3 built
*http://www.merlintec.com/lsi/merlin2.html*
!!Active Book- Active Book Company, Cambridge, UK
- ARM2as 8MHz + 1Mb RAM + 2Mb ROM (I think)
- Monochrome 600@400 LCD
- Like a large tablet PC of current era
- 1988-91
- Tim Rowledge did port of Eliot Miranda's BrouHaHa plus many extensions.
- Underlying OS was port of Helios, a unixish OS derived from TriPos. 
- Not sold but almost completed before being killed by ATT to remove competition for their (dreadful) GO/Penpoint/EO system.
- Full Smalltalk-80 v2 system, rather good performance. Used copy-on-write to allow keeping most objects in ROM so that RAM usage was minimized.
- ARM2as was custom variant designed by ABC to be a 'static' part. Stopping the CPU clock would simply halt everything, ready to continue later.
!!Momenta
- 386SX 20MHz
- 1991
- DOS+ Digitalk based Smalltalk system 
*http://www.kamranelahian.com/articles/article.php?id=37*
!!Merlin 4
- ARM2 8MHz
- 1992
- Jecel Mattos de Assump��o Jr
- prototype, 2 built
*http://www.merlintec.com/lsi/merlin4.html*
!!MediaPad
- StrongARM SA110 206MHz
- 1997
- Interval Research Corporation, Palo Alto, CA.
- Tim Ryan (hardware), Bob Hoover, Bob Alkire, Don Charnley, Paul McCullough, Alan Purdy, Tim Rowledge and Frank Zdybel Jr. with help from Craig Latta, Mike Penk, Kerry Lynn and Phil McBride. Jon Hylands did a web browser and the group included other people
- Intended to be a central controller/master remote control for a synchronous network for household media sharing and domestic/industrial control.
- No real underlying OS since the Squeak VM was extended to be able to schedule processes for the interrupt system as well as the image.
*http://www.rowledge.org/tim/squeak/RTOSinSmalltalk.html*
!!Squeak On A Chip
- Mitsubishi M32R/D
- 1997
- Mits
- Curtis Wickman
- technology demonstration
*http://wiki.squeak.org/squeak/5727* and *http://sib-download.ddo.jp/~sib/Squeak_World/Squeak_Swiki/271.html* have a few paragraphs about this
!!Internet Appliance
- Hyperstone RISC+DSP
- 2001
- Soft Computing Technology
*http://wiki.squeak.org/squeak/1836*

- lock this page
- enter new password
- reenter same password

Send e-mail alerts to these addresses upon next save: