DEMPSTER@MARLBORO.DEC.COM ("Joe Dempster, DTN: 336.2252 AT&T: 609.665.8711") (03/16/87)
This message originates from 2 sources:
Les Earnest
Computer Science Department
STANFORD UNIVERSITY
Stanford, CA 94305
415.723.9729
ARPA: LES@SAIL.STANFORD.EDU
Joe Dempster
DIGITAL EQUIPMENT CORPORATION
6 Cherry Hill Executive Campus
Route 70
Cherry Hill, NJ 08002
609.665.8711
ARPA: DEMPSTER@MARLBORO.DEC.COM (MARKET)
The goal of this project is to publish an analysis and history of
the evolution, implementation and use of Digital's 36 bit systems.
This period began with the PDP-6 in 1964 and continues today with
TOPS 10/20 development, which is scheduled to end in 1988.
We are working aggressively to finish the project, and have it
published, by March/April 1988. This will require that the
completed manuscript be ready to go into the publication cycle
by August 1987!
The project will attempt to answer the following questions:
1. In what markets/applications were these systems used?
2. Who were the users of these systems and what impact did
roughly 2,500 TOPS 10/20 systems have on their organizations?
3. Who were the principle system architects of these systems?
What features, and if there had been sufficient time to
implement them, would have significantly improved the
architecture?
4. What impact did the decision to continue to examine design
extensions to the architecture have on the usefulness and
acceptability of these systems. This is in contrast to a
more common practice today to work from a detailed design
specification, sometimes dated, building follow-on systems
which provide increased performance through the use of new
component technologies and packaging techniques.
5. What part of the overall design (TOPS10/20) was technology
dependent and what can still be considered "unequaled" in
relation to other computer architectures still undergoing
active development?
6. What type of development environment (both HW and SW)
supported and contributed to the evolution of 36 bit
systems?
7. What influence did TOPS 10/20 have on other vendors system
development?
This history will undoubtedly be assembled from many sources and
participants. Some information will be anecdotal; there will be
interviews with the people involved (users and developers) and technical
papers will be solicited. Of course there will also be the packaging
and assembly of facts as we see them.
The result will hopefully have sufficient depth to serve as:
1. An introductory or advanced text on system design and
hardware/system software implementation.
2. A analysis of the success and difficulties of marketing
complex systems into a very crowded market of competing
alternatives.
3. A catharsis for those of us who have contributed to the
development and use these systems and who will now move
onto new computing architectures and opportunities.
In addition to interviewing directly 25-50 developers, users and
product managers we will continue to work to identify contributors
and significant events up to when the final draft is submitted to
the publisher. Two "topics" are already under development:
1. Rob Gingell from SUN is working on a paper which looks
at extensions to TOPS 20 which would have enhanced its
capabilities.
2. Frank da Cruz and Columbia are summarizing 10 years of
experience and development of TOPS 20 systems. Some
effort will also be made to detail the process which
lead to their selection of a follow-on architecture to
TOPS 20.
There is a need to develop additional topics which represent the
use and application of the technology (TOPS 10/20) in other areas.
Specific recommendations are welcome as are proposals to develop
them. A short abstract should accompany any such proposal. Every
effort will be made to work with individuals or organizations
interested in making such a contribution.
There will be a standalone (no network connections) DECSYSTEM 2020
(YIPYIP) dedicated to supporting the project. This system has a 3
line hunt group, with all lines accessible from a single number
(201.874.8612).
Both YIPYIP and MARKET will have "public" directories for remote
login (<log>DEMPSTER.PROJECT-10262 <Password>LCGLCG). MARKET can
be accessed by modem (617.467.7437), however disk quota is limited.
MARKET's primary purpose <DEMPSTER.PROJECT-10262> is ARPAnet TELNET
access. YIPYIP is a dedicated PROJECT-10262 system. MAIL can also
be sent to DEMPSTER on either system.
YIPYIP and MARKET will keep a running summary of ideas and comments
up on Columbia's BBOARD software. KERMIT also runs on each system
for uploads.
SAIL.STANFORD.EDU will support ARPAnet transfers to a "public" area:
FTP<ret>
CONNECT SAIL.STANFORD.EDU<ret>
SEND AFN.EXT<ret>
DSK: AFN.EXT [PUB,LES]<ret>
SAIL runs WAITS, an operating system similiar to TOPS 10. File
names are limited to 6 characters and extensions limited to 3.
Implementation details:
1. User input is welcomed and desired from all application
and geographic areas.
2. Input from past and present developers is also desired.
3. Throughout the project a secondary goal will be to build
a list of users/locations (installation date, duration and
disposition) of PDP-6 and KA, KI, KL and KS systems.
Serial numbers, if available, are requested.
4. We anticipate that this project will generate a large
volume of information (which we hope will arrive
electronically). Some information, for any number of
reasons, may not be in line with the project's stated
goals. Therefore, all notes, interview material and
submissions will be donated to the Computer Museum in Boston
at the the completion of the project to be available for
future reference and research.
Ideas, contributions, suggestions and criticism are welcome. As these
36 bit systems were the products of a multitude of people, so too
will be the writing of their history.
--------