o
""l
Vol. 18, No. 13
CD
Minicomputers
(and
Their
Applications)
So you just got a new
Spiras-65 System.
More power to you.
Congratulations. You've picked the right computer for the job. Spiras-55. This compact stored
program controller/processor packs the most powerful instruction set in terms of core
economy and effective speed, to give you unprecedented performance/dollar. That's computer
power. And it's all yours, without major options, because all predictable performance features
are standard with
Spiras: • Bootstrap
loader • Double
precision arithmetic
• Floating point
arithmetic· Hardware
multiply and divide
• DMA • DMC for 52
devices • Multilevel
indirect addressing
to 32K • Relative
addressing ±512
words • Immediate
instructions • Register
copy instructions
\Jicro-programmable
ditional instructions
COIlPU'l SItS ,?,: AUTOHA/rrou
24-word base page
fJSCEHDEH 1£H:;9
1emory expansion to
f
... ;
55K • Remotable I/O
control panel with
octal data-entry
.
keyset and NIXIE®
octal! decimal
display • Extensive software
package including A.S.A.
FORTRAN IV
\j
erals
display fo:
editing which provid~
in terms of visual presentation c
as well as interface with any]
In peripheral equipment, softw
your whole Spiras system, you've g01
A
Spiras Systems, Inc. (Forrr
332 Second Avenue, Waltham, Mal
(517) 891-7300. TV
® Registered Trademark of Burroughs Corporation.
Designate No. 18 on I
~ For Interdata Designate No. 40 on Reader Service C.
ms, Inc.
Iticn
l.etterf) To The Editor
Seeks Accreditation
I read with interest yom cornnlents
concerning accreditation qf prlv8tely
operated data processing school:; in
your October issue (PilQfJ '7). I wuuld
greatly appreciatp. I.·.DIVISION
GENERAL .INSTRUMENT CORPORATION
13040 Southi Ceris~x Avenue, .'. Hawthorne,California 90250
. (213)
679·3~77 /772~2351
TWX 910·325·6203
Oesianate No. 19 on Reader Service Card
MULTI-ACCESS FORUM
MARTIN LUTHER KING MEMORIAL PRIZE CONTEST
- SECOND YEAR
(Please post this notice)
Computers and Automation has received an anonymous gift and announces the annual Martin Luther
King Memorial Prize, of $300, to be awarded each
year for the best article on an important subject in
the general field of:
The application of information sciences and
engineering to the problems of improvement in
human society.
The judges in 1970 will be:
Dr. Franz L. Alt of the American I nstitute of
Physics; Prof. John W. Carr III of the Univ. of
Pennsylvania; Dr. William H. Churchill of Howard Univ.; and Edmund C. Berkeley, Editor of
Computers and Automation.
The closing date for the receipt of manuscripts this
year is April 30, 1970, in the office of Computers and
Automation, 815 Washington St., Newtonville, Mass.
02160.
The winning article, if any, will be published in the
July issue of Computers and Automation. The deci··
sion of the judges will be conclusive. The prize will
not be awarded if, in the opinion of the judges, no
sufficiently good article is received.
Following are the details: The article should be
approximately 2500 to 3500 words in length. The
article should be factual, useful, and ~nderstandable.
The subject chosen should be treated practically and
realistically with examples and evidence - but also
with imagination, and broad vision of possible future
developments, not necessarily restricted to one nation
or culture. The writings of Martin Luther King should
be included among the references used by the author,
but it is not necessary that any quotations be included in the article.
Articles should be typed with double line spacing
and should meet reasonable standards for publication.
Four copies should be submitted. All entries will
10
become the property of Computers and Automation.
The article should bear a title and a date, but not the
name of the author. The author's name and address
and four or five sentences of biographical information
about him, should be included in an accompanying
letter - which also specifies the title of the article
and the date.
"Many people fear nothing more terribly than to
take a position which stands out sharply and clearly
from the prevailing opinion. The tendency of most is
to adopt a view that is so ambiguous that it will
include everything and so popular that it will include
everybody .... Not a few men who cherish noble
ideals hide them under a bushel for fear of being
called different."
"Wherever unjust laws exist, people on the basis of
conscience have a right to disobey those laws."
"There is nothing that expressed massive civil
disobedience any more than the Boston Tea Party,
and yet we give this to our young people and our
students as a part of the great tradition of our nation.
So I think we are in good company when we break
unjust laws, and I think that those who are willing to
do it and accept the penalty are those who are a part
of the saving of the nation."
- From U/ Have a Dream" - The Quotations of Martin Luther King, Jr., compiled
and edited by Lotte Haskins, Grosset and
Dunlap, New York, 1968.
Reverend Martin Luther King, Jr., was
awarded the Nobel Peace Prize in 1964,
when he was age 35.
He was in jail in the United States more
than 60 ti meso
He was assassinated in Memphis, Tennessee,
April 4,1968.
COMPUTERS and AUTOMATION for December, 1969
UNBUNDLING AND THE USERS' GROUP
D. F. Stevens
Mathematics and Computing
Bldg. 50A, Rm. 1135
University of California
Lawrence Radiation Laboratory
Berkeley, Calif. 94720
(The following expresses my position on the question of
how computer users' groups should adjust to the new
separate pricing policies recently announced by the major
computer manufacturers. My qualifications to speak on
such a topic include some experience in SHARE (on the
FORTRAN Project) and some experience in VIM (as
chairman, first of the Software Evaluation Committee, and
currently, of the Operating Systems Committee). The views
expressed in the note are, of course, my own, and may not
reflect official opinions either of my computer installation
or of VIM.)
It is surprising that among the spate of articles prompted
by the possibility (and more recently, the actuality) of
unbundling there has been no speCUlation as to the new role
of the users' group (UG). Perhaps this is due to a feeling
that once the step has been taken, the UG has no choice
but to go along with it, and continue to conduct business as
usual. This would, of course, be a mistake; I would like to
offer a more interesting alternative.
Those manufacturers who have unbundled have effectively entered the open marketplace as sellers of that
software which is not in the public domain. The operative
words in that sentence are open and sellers: in other words,
the hardware manufacturer must now compete with other
interested suppliers. It is in the interests of the UG to see
that this competition is fair and as fierce as possible. This
requires that the hardware manufacturer lose his privileged
position with respect to the UG. The UG should, in fact,
invite and encourage participation by other suppliers in all
appropriate activities. To ensure an orderly presentation of
competitive material, the UG will have to adopt, and
enforce, a new set of procedural rules, possibly resulting in
many more closed sessions.
The question of certification is also interesting. The UG
would seem to be the natural agent, either for the certification of a new product, or for the establishment of suitable
criteria for its certification. (I should think that Certification Committees could count on active participation by a
number of suppliers ... ) this leads rather naturally into a
consideration of the UG as a collective bargaining agent.
Such a concept is particularly appealing when considering
the following thorny points:
• What is a reasonable installation cost for a new
product (in terms of manpower and computer
time)? Who bears this cost when the product is
installed by the user? By the supplier? What
recourse does the user have when the estimated
cost is exceeded? Is the time required for performance testing to be included in the installation cost?
• Who bears the cost of time (man and computer)
lost due to grievous software error? Who defines
"grievous"?
• Is credit to be given to a user who fixes a bug?
How much?
• What level of maintenance is included in the cost
of the product? (This includes response time and
documentation as well as the completeness and
accuracy of the fix.) What should it cost (manpower and computer time) to install a fix?
• Manufacturers are well-known for "fixing" a bug
by changing the documentation. Will this continue
to be permissible?
Should the supplier prove recalcitrant, a purposeful users'
group could institute a boycott of considerable severity ...
this may, of course, be neither a desirable nor a practical
adjustment to the new situation: it is past time for debate
0
on the issue to begin.
PROFESSIONAL ETHICS DEBATED BY THE BRITISH COMPUTER SOCIETY
(Based on a report by Kenneth Owen in October 14, 1969
edition of The Times, Printing House Square, London
E-C-4, England)
The subject of professional ethics for computer people
was debated recently by members of the British Computer
Society. Opin ions represented in the debate ranged from
those who felt there was no need for a professional code at
all, to those who suggested a recommended scale of charges
for professional work.
The main topic of interest was the degree of responsibility assumed by a professional for the consequences of
his work for third parties.
Another hotly discussed issue was the privacy of information held in computer systems. Although the problems
of handling "private" information were well aired; there
was a general consensus of opinion that no satisfactorY'
short-term solution could be offered.
The Society has set forth draft proposals for guidance of
its members. These proposals state that members should
exercise their skill impartially and to the best of their
ability, should not disclose confidential information conCOMPUTERS and AUTOMATION for December, 1969
cerning employers or clients without permiSSion, and
should not accept any position in which their interests
conflict with their duty.
.
The guidelines also state that consultants should declare
interests such as directorships, financial interests and personal relationships to clients before accepting instructions,
and should not invite any employee of clients to consider
alternative employment, except with the client's agreement.
In addition, all advertising should be legal, clean, honest
and truthful, and consultants should provide a written
agreement stating their fee before undertaking an assignment.
Under its current rules, the British Computer Society
can exclude any member who has been, in the opinion of
the Council, "guilty of conduct which renders him unfit in
their opinion to retain his membership". The Council must
hear every case itself; it cannot delegate its powers. And
there is no right of appeal.
Because many members of the· Society feel th is is too
inflexible, the Council is trying to produce an improved
disciplinary procedure based on investigation, disciplinary
and appeals committees.
0
13
THE PERFECT DOCUMENTATION
The program was written, debugged, and tested.
Our hero leaned back with a sigh and rested.
"Thank heaven I'm through with that hairy thingI'll goof off a few days to replenish my zing."
Then in walked the Chief of Operations.
"Say how do you work this conglomeration?"
Operating instructions had slipped his mind,
He went back to his desk to get out of that bind.
In a week and a half, this book he had finished,
His strength and morale completely diminished.
He asked his chief for a vacation,
To renew with his family normal relations.
"Take some time off, it's fine with me."
Said his boss, The Chief of Sand P.
"There's two or three things you first must complete,
That will take a few days, maybe a week."
"Our programming standards we can't ignore.
The guys who did went through that door."
"One program deck is of course not enou-gh We need at least seven for backup and stuff"
"Two flow charts required for every projectA one-on-one detailed and general logic."
"We also insert in our archives and files,
A system description in narrative style."
The end of this story we're happy to tellHow our hero avoided this programmer's hell.
No one in the country now tops his production,
Two thousand a day of symbolic instruction.
One noteworthy item must be added,
His office is now completely padded.
Sidney Golder
Chief, Programing Psychiatric Services
Berkeley Systems
333 Washington Street
Boston, Massachusetts 02108
With apologies to Lou Ellen Davis (See Computers
and Automation/or August, 1969, page 43.)
THE QUALITY OF ELECTRONIC DATA
PROCESSING SERVICES
- ADAPSO POSITION PAPER
ADAPSO (Association of Data Processing Service Organizations, Inc.)
420 Lexington Ave.
New York, N.Y. 10017
The following is a position paper issued by ADAPSO,
the trade association of the computer services industry, on
the "Furnishing of Error-Free Electronic Data Processing
Services. "
The Association of Data Processing 'Service Organizations is deeply concerned that there may be some public
misunderstanding of the nature of the services provided by
the electronic data processing services industry, resulting in
an erroneous belief that perfection is feasible. The consequences of reliance on such a belief could be most unfortu~ate.
14
Perfection in the furnishing of EDP services is no more
possible than in any other industry dependent upon human
frailty and fallible equipment.
The causes of human error are legion. Machine error
results from an equally large number of causes, ranging
from dust in a sensing device to fluctuations in electrical
input. Error may also result from economic considerations,
such as the practical limitations on debugging a program;
thus to be certain of zero error in the trillions of computer
calculations incident to any continuing commercial application, requires such exhaustive testing as to be uncompetitive.
Undoubtedly, part of the public misunderstanding is a
result of the exciting ach ievements of the space program,
made possible by the computer. However, pin-point accuracy in space computer applications has resulted only from
almost interminable testing and the broadest use of back-up
equipment, not justified economically in most commercial
applications.
It may be, however, that to some extent the EDP
industry is itself also to blame, because computer personnel
unfortunately sometimes fail to recognize the ignorance of
the lay public in technical matters. When an EDP professional says that his program is perfect or that his service
never makes an error, he of course excludes that essential
minimum which cannot practically be avoided; he intends
no more misrepresentation of his services than the automotive salesman who promises a first rate vehicle knowing
that inevitably one must come off the assembly line with
defects.
ADAPSO recognizes the danger that some members of a
lay public may misunderstand such statements, and proposes to conduct a far ranging remedial public information
program. As a first step, it calls upon all responsible EDP
service industry members to distribute this position paper
to customers, in order promptly and effectively to inform
those key members of the public most likely to be reliant
0
on the results of the industry's output.
ANNUAL COMPUTER PROGRAMMING
CONTEST FOR ELEMENTARY AND
SECONDARY SCHOOL STUDENTS
Hyman Speck
Loop College
64 E. Lake St.
Chicago, III. 60601
A contest designed to stimulate inventive interest among
young students in the computer programming field is being
sponsored for the seventh year by the Association for
Educational Data Systems (AEDS).
Prizes of $25 in U.S. Savings Bonds will be awarded to
winners in seven judging categories: Business, Biological
Science, Computer Science, Games, Human ities, Mathematics, and Physical Science. The grand prize win ner wi II be
selected from these seven winners. He will receive an
additional $100 in bonds, plus an all-expense paid trip to
the AEDS convention in Miami Beach, Florida.
'Project Genes, 'Electoral College Calculations, Comparative Computer Linguistics, Five Card Draw Poker, and
Model Rocketry Altitude and Flight Path Pred ictions are
some of the outstanding projects submitted last year. First
prize winner last year was John O'Donnell of Tates Creek
High School in Lexington, Kentucky, who developed a
computer language and compiler called SIGMA, which he
designed to run on teleprocessing terminals.
Projects may be submitted by an individual or a team of
two or more pupils. Deadline for entries is April 1, 1970.
Persons wishing more details and application blanks are
0
invited to write me.
COMPUTERS and AUTOMATION for December, 1969
A TIME-SHARING COMPUTER TERMINAL VS. THE SMALL COMPUTER
Lewis C. Clapp
DIAL-DATA Inc.
429 Watertown St.
Newton, Mass. 02158
Those of us who have been fortunate enough to
work with small computers know how exciting it is
to have your own little machine available for problem-solving and numerical computation. As the
memory hums and console lights flicker, there .is a
certain fascination at having complete control over
the logical mechanism under your fingertips. For
quite some time the novelty of the exposure itself
is enough to motivate us to spend many long nights
experimenting to find the fastest way to multiply
two vectors or how best to pack a large array of
data into the tiny memory. But ultimately the euphoria wears off and we learn that there is often a
great deal of tedious effoLt required to solve sci-entific problems with the aid of a small computer.
One alternative to the small computer is having
access to a time-shared computer terminal. Although
the small computer is a very useful and important
resource in many applications such as communications,
process control and data preprocessing; it is not
as effective a tool for interactive problem-solving
.as the time-shared computer.
What the small computer has to offer is raw computing power -the ability to manipulate and process
bits of information through a machine. Now it is
true that when computer time-sharing first became
available it too was looked upon simply as a source
of raw computer power. But in the three or four
years which have passed, the time-sharing vendors
have become more sophisticated. They have begun to
realize that their job is not simply to provide computer power over phone lines or to make" computer
juice" coming out of the walls like electric current; but rather their job is to help people process
information and problems. They realized that to do
this job well they had to provide sophisticated
tools which would make it easier for people, who
were not computer experts, to use the computer ·to
solve their own problems. In providing these tools,
the time-sharing services have done admirably well.
So well in fact, that the small computer probably
will never be able to catch up.
The initial step in time-sharing was to provide
programming languages which could be learned rather
quickly by anyone with a minimum of intelligence and
a minor degree of patience. These languages such as
BASIC and CAL could truly be mastered in an hour or
two and then could be used to solve very sophisticated engineering and scientific problems. Indeed,
this use of time-sharing computers was so popular
that the small computer manufacturers were themselves compelled to copy the idea and produce versions of some of these languages on their own machines. However, the small computer by its very
nature is limited in size and flexibility. Because
of these limitations it is very difficult to reproduce all of the features of a truly sophisticated
interactive time-sharing language on a small computer. For example, the time-sharing machine has
power-in-reserve which can be called upon when it is
really required. The small computer on the other
hand always seems to have the habit of running out
of steam just at the critical moment.
Having developed simplified programming languages
which an engineer could conveniently learn, the next
Excerpts from a talk presented at the "IEEE-NEREM
Seminar on PERSONAL DESK TOP COMPUTERS"; Boston~
Mass. November 4~ 1969
COMPUTERS and AUTOMATION for December, 1969
big step that the time-sharers took was to develop
specialized packages in their own subject area of
concentration. One ·company, for instance, has developed·a whole se~ies of tools to be used by the
printing and typese;ting prQiession. Another company has concentrated on developing aids for small
manufacturing concerns who use numerically controlled
milling machines to produce parts. However, at
DIAL-DATA we concentrated a great deal of effort Qn
developing tools for automated electronic engineering and design.
Now let me repeat, that to use one of these packages, requires no knowledge of programming skill
whatsoever. For example, an engineer who has just
drawn a circuit diagram on the back of an envelope,
wants to test his circuit and verify that it will
perform as desired. In the old days he would have
built the circuit with a soldering iron and a box of
components. Several hours later, if he was lucky
enough to have all the parts on hand, he ~ould be
ready to test it; Today, this time-consuming process is eliminat~d or at least delayed until the
computer has checked the circuit and verified its
operational characteristics. Our engineer simply
dials the computer, calls the Circuit Analysis Program (CODED) and,describes hjs circuits to the computer in a simplified language.
He can then ask the computer to calculate the
voltages at each. of the nodes. If he really~ants
to get sophisticated, he may ask the computer to
perform a Worst Case Ana1ysis. That is, by varying
the values of all the components in the circuit
from their high to low tolerance values, the computer determines which values will have the most deleterious effetton the performance of the circuit
anp how bad that performance will actually be. The
computer performs the analysis in a matter of seconds and depending upon the results~ our engineer
is either out on the golf links or back at the drawing board. Contrast this with the old-fashioned
way of designing circuits, which essentially meant
standing over a breadboard and designing by intuition, which at best allowed the development of one
circuit every few hours. The modern circuit designer uses the computer to generate dozens of alternative circuits in a single afternoon. He may then
select the most desirable circuit from the rest before he leaves his office to try for his hole in
one.
In developing circuit analysis programs we have
found that a form of Parkinson's Law also applies
to computers. Namely, the programs, memory, and
speed requirements always rise to fill the capacity
of the machine. There is always one more case or
one new feature that needs to be added before the
program is acceptable. Therefore, I conclude that
any serious work on Computer Aided Circuit Design
will not be done with the limited memory of a small
computer.
.Other program tools have been developed to aid
the electronics engineer. Since ~e don't have time
to describe all of them, I shall merely give some
indication of the range of applications which have
been investigated to date. Again, I will give the
DIAL-DATA name for these design tools, although
other time-sharing companies may have some of the
same programs with different names and perhaps even
different capabilities.
DIGILOG - Digital circuit design simulator.
15
AS WE GO TO PRESS
THE USE OF COMPUTERS FOR BALLOT COUNTING
SHOULD BE DISCONTINUED UNTIL ADEQUATE SAFEGUARDS AGAINST TAMPERING HAVE BEEN DEVELOPED, according to testimony presented to the California State Assembly Committee on Elections and Constitutional Amendments by William C. Clauer, president of
Intellectron, Inc., a computer software company. Mr.
Clauer presented to the Committee a research report,
"Vulnerabilities of the Present Computer Vote Counting
Systems to Computer Fraud", published by his company.
The report concluded that present computerized votecounting systems are highly vulnerable to program modification techniques which may not be detected under present security procedures.
To demonstrate the possibility of tampering, methods
were developed by Intellectron which would permit an
altered vote model counting system to pass currently
used tests for "logic and accuracy". One method automatically inserts a count bias after the completion of
normal testing procedures. Another permits a computer
operator to introduce a pre-selected count bias by simply pressing a series of regular switches on the computer console in a manner that would not be regarded as unusual or abnormal. Mr. Clauer stated that neither of
these methods would necessitate any wholesale or high
level conspiracy to defraud.
As a result of his company's research, Mr. Clauer
recommended that "full and adequate safeguards to protect the integrity of the ballot counting by computers be
researched, developed, and made a legal requirement. "
CONGRESSMAN CORNELIUS E. GALLAGHER RECENTLY ASKED THE U. S. HOUSE OF REPRESENTATIVES
TO ESTABLISH A SELECT CQMMITTEE ON TECHNOLOGY, HUMAN VALUES, AND DEMOCRATIC INSTITUTIONS. The purpose of the Committee would be to clearly
ANALOG
FILTER
FILSYN
MICAP
- Simulates Analog Processes.
- Designs ladder type filter networks.
- Synthesizes filters given external
design parameters.
- Microwave Design Package.
In addition to the above, there are programs for
minimizing logical equations and for fabricating
electronic components once the design has been established. Most of these programs are available today only through a time-shared computer terminal.
Although less extensive versions of these programs
may some day be_available on a small computer, it
is unlikely that they will have the versatility and
sophistication of the same packages provided through
the time-shared computer terminal.
It takes a great deal of time and talents of
many people to develop good reliable programs to
aid in electronics design. Very few organizations
will find it economic to recreate these programs
~hemselves for their small computers even if they
had the resources. Among their alternatives are to
secure the service from a computer time-sharing firm
or to prod the computer manufacturers into delivering these programs with their machines.
The manufacturer, of course, is interested in
selling his computer and he may make some effort to
fulfill such requests. But manufacturers' software
has often left something to be desired since they
cannot be expected to be expert in every scientific
discipline. And, even if the manufacturer is real16
find out where we are heading technologically, and to be
sufficiently informed to meaningfully ask the question:
"Is that where we want to go?"
Gallagher cited potential dangers in highway use surveys, data banks for migrant children and psychiatric
records, and the dependence on the computer to create
technical means of dictating human decisions as evidence
of the need for such a Committee. He pointed to air and
water pollution as examples of bad - and unanticipated
- effects of technology, and emphasized the need to evaluate a new technology before such results occur.
The Harvard Program on Technology and Society, The
Computer Science and Engineering Board of the National
Academy of Sciences, the American Academy of Arts and
Sciences Committee on the Year 2000, and the Special
Program on Security and Privacy of the Business Equipment Manufacturers Association were commended for
their research into the effects of technology on society.
The Congressman recommended that one crucial task for
the proposed Select Committee would be to work with
these groups and to act as a clearing house for the public
for data developed by these groups and elsewhere.
THE EFFECTS OF THE CARTERFONE DECISION AND THE FUTURE OF DATA COMMUNICATIONS were the main topics of discussion at the Fourth Annual
Digitronics Users Association Conference held in New
York in mid-November. (Digitronics Corp., Albertson,
L. I., N. Y., is a pioneer company in the design and production of data accumulation and transmission equipment.) The Carterfone decision, handed down by the
Federal Communications Commission (FCC) 16 months
ago, ruled that equipment other than that manufactured
by the telephone companies could be bought by users and
(Please turn to page 62)
ly able to deliver the program he will probably
follow IBM's lead and charge you a use fee for the
program. Finally, if he should provide you with a
program today, what happens tomorrow or next week,
when the state of the art advances and better programs are developed?
The time-shared computer center manager, on the
other hand, does not have a built-in sale once he
signs a contract with a client. He must earn every
single dollar by providing his users with new capabilities and continuously improving software. He
must learn what tools his clients need to do their
jobs better, and then work his programmers feverishly to develop the programs.
In conclusion, I think the most interesting
argument in favor of the time-sharing terminal for
interactive problem-solving is provided by the
small computer itself. Many small computer users
now use time-sharing service centers who provide
programs which simulate the small computers on the
time-sharing terminal. These small computer users
develop the programs under time-sharing and then,
after the programs have been written and checked
out, run the production jobs on the small machine.
They have found that the job of program creation
can be done more conveniently using all the resources of a good time-sharing system. Having all
those lights, registers and keys of a small computer under one's immediate gaze may not have been
so exciting after alII
COMPUTERS and AUTOMATION for December, 1969
Who's Who in Computers and Data Processing
Who's Who in Computers and Data Processing will
be published jointly (as an annual publication) by
The New York Times Book and Educational Division
and Computers and Automation. The fifth edition
is scheduled to be published in three volumes in
hard cover in early 1970, and will include upwards
of 8000 capsule biographies. The three volumes
are:
Vol. 1
Vol. 2
Vol. 3
Systems Analysts and Programmers
Data Processing Managers and Directors
Other Computer Professionals
Who's Who in Computers and Data Processing is
to be typeset by computer. As a result, it should
be possible to include new entries (and to mOdify
previous entries) CONTINUOUSLY -- especially since
Who's Who will be published periodically.
Consequently, if you have not yet sent us your
up-to-date filled-in Who's Who entry form, PLEASE
SEND IT TO US QUICKLY -- the chance is good that
your entry can be promptly included. Use the entry
form below, or a copy of it.
SEND US YOUR ENTRY TODAY!
Following are sample capsule biographies which
will be published in the first annual edition of
Who's Who in Computers and Data Processing.
Special Abbreviations
_M_a_i~n_I_n_t_e~r~e~s~t~A~b~b~re~v~l~'a~t~i~o~n~s
b: born
A Applications
B Business
ed: education
C Construction
ent: entered computer
D Design
field
L Logic
m-i: main interests
Mg Management
t: title
Ma Mathematics
org: organization
P Programming
pb-h: publications, honors, memberships
Sa Sales
and other distinc- Sy Systems
tions
h: home address
PAGEN, Dr. John / director - CAl project / b: 1926
/ ed: BS; MEd; EdD / ent: 1967 / m-i: A P Sy; computer assisted instruction / t: director - INDICOM
/ org: Waterford Township School District, 3101 W
Walton, Pontiac, MI 48055 / pb-h: AERA; Phi Delta
Kappa; MASA; AASA; reports on CAl / h: 463 Berrypatch, Pontiac, MI 48054
PALM, John N. / EDP management / b: 1938 / ed: BA,
math / ent: 1957, part time; 1960, full time /
m-i: P Sy; management of systems, programming,
operations, etc. as applied in solving retail problems / t: vice president, information systems /
org: Target Stores, Inc., 8700 W 36 St, Minneapolis, MN 55426 / pb-h: CDP, SPA / h: Route 1, Box
27, Wayzata, MN 55391
PALMER, Dennis W. / EDP mgr / b: 1937 / ed: 2 yrs
college / ent: 1959 / m-i: Mg P Sy / t: EDP mgr /
org: Protected Home Mutual Life Ins Co, 30 EState
St, Sharon, PA 16146 / pb-h: DPMA, SPA, CDP / h:
Rt 3, Box 700, Corland, OH 44410
PALMER, Fred E. / systems & programming / b: 1935 /
ed: 3 years college / ent: 1960 / m-i: A B P Sy /
t: manager of programming / org: Western Farmers
Association, 201 Elliott Ave W, Seattle, WA 98119
/ pb-h: CDP, DPMA / h: 19611 62nd NE, Seattle, WA
98155
PAN, George S. / senior technical management / b:
1939 / ed: BSEE, Illinois, MSEE, Syracuse / ent:
1960 / m-i: A Mg Ma P Sy; simulation / t: director,
management sciences division / org: Interactive
Sciences Corp., 170 Forbes Rd, Braintree, MA 02184
/ pb-h: "Weighted File System Design Method", 1965
IBM National Systems Symposium, "Generalized File
Structure and Optimum Design Considerations", 5th
Nat'l Computer Conference of Canada / h: 5146 N
11th Ave, Phoenix, AZ 85013
COMPUTERS and AUTOMATION for December, 1969
WHO'S WHO ENTRY FORM
(may be copied on any piece of paper)
1.
2.
3.
4.
5.
6.
7.
Name? (Please print).~~____________________
Home Address (with Zip)? ________________
Organization? ____~~~------------------It s Add re s s (w it h Zi P )'~ _______________
Your Ti tle?
Your Main I-n~t-e-re-s-t~?~-------------------)
Applications
( ) Mathematics
)
Business
( ) Programming
)
Construction
( ) Sales
)
Design
( ) Systems
)
Logic
( ) Other
Management
()
(Please specify)
Year of Birth?~---~----------------
~: ~~~~a~!~~r:~dc~~~:~:~?~F7i-e~ld~?~------------10. Your Present Occupation?
---------------11. Publications, Honors, Memberships, and other
Distinctions?
(attach paper if needed)
12. Do you have access to a computer? ( )Yes ( )No
a. If yes, what kind of computer?
Manufacturer?
.
Mode 1?
-------------------b. Where is it installed:
Manufacturer?
Address?
------------------------c. Is your access: Batch? ( ) Time-Shared? (
Other? ( ) Please explain:
d. Any remarks?
-----------13. In which volume or volumes of Who's Who do you
think you should be included?
Vol 1. Systems Analysts and Programmers
Vol 2. Data Processing Managers and Directors
Vol 3. Other Computer Professionals
14. Associates or friends who should be sent Who's
Who entry forms?
Name and Address
(attach paper if needed)
When completed, please send to:
Who's Who Editor, Computers and Automation,
815 Washington St., Newtonville, Mass. 02160
1i'
You're looking
at output
from the first
great product
byPath.
A new non-impact communications and
computer printer that's so unique it makes
other high-speed printers old fashioned.
Our first product, the Path 1200.
Which works on a new but relatively simple electronic~optical
technique. And does away with
the things that cause trouble in
other printers. Like ion guns,
ink deflectors, cathode ray
tubes and special paper in nonimpact printers. And the hundreds of mechanical parts in
impact printers.
New Reliability
The Path 1200 is simple, lightweight and compact. Over 80% of
its parts are electronic. And
the few mechanical parts are
simple rotational drive mechanisms.
Speed and Quality
It's fast. Producing single copy
output at 1200 lines per minute.
The text you're reading now is
our output.
Enough said?
Quietness.
The Path 1200 is virtually noiseless.
In fact, we've built a
pleasant "hum" into it just so
you can tell when it's turned on.
New Economy
Because of its unique advanced
design, the Path 1200 costs much
less to buy than other printers.
And much less to run. On the average, about 75% less to buy.
And, while other printers cost
$30 to $75 a line, ours costs $6.
So why put up with old fashioned,
noisy, unreliable printers when
you can own the Path 1200 for
less? For more information write:
Path, 20 Beckley Ave., Stamford,
Conn. 06901. Or phone:(203) 3484245. Or you can see the Path
1200 at the FJCC - Booth 8601/2.
We find new ways to apply whats already known.
Designate No. 14 on Reader Service Card
SELEC'T A
MINICOMPUTE~R
Walter R. Anderson and Edward H. Sonn
IRA Systems
332 Second Ave.
Waltham, Mass.
"There are no comparable figures of merit for computer system architeoture, nor are there comparable figures of merit for software. How,
then, should a prospective customer evaluate minicomputers?"
There has been much discussion over minicomputers
revolutionizing the market for computers, from the viewpoint of both replacement of "hard-wired" logic systems
and opening up increased use of computers by new users
who could not economically afford to purchase or rent
computers in the past. The purpose of this article is to
present the thought processes which the authors believe to
be necessary in the selection of, minicomputers.
What is a "Mini"?
The definition of a "mini" has not been clearly established, but has generally been rather arbitrarily defined as
"a computer which sells for less than $20,000 in its basic
configuration". The value per dollar received should be the
key for selecting a "mini" for a given application.
As President of I RA Systems, Mr. Anderson directs the
management of the various design groups within the company. These groups are engaged in the design of computercentered systems for instrumentation, control, data processing, and display applications. He holds a B.S.E.E. and an A.B.
degree, and has completed an advanced course in logic
design at the Massachusetts Institute of Technology. Mr.
Anderson is presently the Editor of the Production Technology Newsletter. and is a member of IEEE and the EIA #TR-31
Committee on Numerical Control.
20
The "computer horsepower" of a "mini" varies from
manufacturer to manufacturer, so no agreed "figure of
merit" has been established to adequately define just what
a mini is. Generally speaking, however, a "mini" can at least
perform most simple arithmetic functions and has at least
some associated peripherals.
Most "mini" computers are priced low by virtue of the
fact that either large printed circuit boards are used for
low-cost production of the machine or little or no input/
output capability is provided. Typically, many mini's do
not have substantial software supplied with or even developed for the machine, hence, development costs are
lower, and these savings are passed on to the customer.
Minicomputers are characterized by providing many
customer-selected options; many times this may cause the
cost of the "mini" to be adjusted upwards subs.tantially.
Service, maintenance, training, and installation are
generally not included in the price of a typical minicomputer.
Edward H. Sonn is the Manager of Computer Applications
for I RA Systems. He holds the S.B. (E.E.l degree from the
Massachusetts Institute of Technology and the M.S. (E.E.)
degree from Columbia University. He directs the hardware
and software development of computer-based systems for
IRA.
COMPUTERS and AUTOMATION for December, 1969
Computer Horsepower
Computer horsepower (CHP) may be defined as:
(amount of work per instruction)
X
(number of instructions per second)
where the amount of work done per instruction is determined by:
instruction set and fixed paging. A bench-mark subroutine
program would be useful in th is evaluation.
The computer memory must be large enough to hold the
program and the data. The data storage requirements are
usually independent of the computer. The program storage
requirements are determined by the algorithm and the
particular computer selected. If you are estimating requirements, be sure to use a generous safety factor.
1. The size and organization of the instruction set,
2. The flexibility of addressing and indexing modes,
and
3. The input/output structure of the machine.
and the number of instructions per second is determined
by:
1. Memory cycle time of the machine,
2. Execution time for instructions, and
3. The extent to which the processor must share
memory with devices having direct memory
access.
Interfacing
The most exciting feature of the minicomputer is that it
can be used to control a wide variety of devices; this may
be the reason you desire to purchase one. The prospective
purchaser should determine what interfaces are available
and what compatible packages are available for housing the
logic required in special-purpose interfaces. If you must
design your own interface, be sure to check on the
availability of a well-documented interface manual.
Evaluation of Computers
Input/Output Features
One of the "safest" ways to justify the selection of a
"mini" for use in your product line or for use in your
automated process is to point to the numerous surveys in
the field by "experts" and pick the processor as a result of
the surveys.
What the surveys often do not tell you is probably more
important than what they do tell you - strange as that may
seem. No figure-of-merit criterion has Qeen developed for
the computers or stored-program controllers mentioned in
most surveys. No mention is made of financial stability of
the company or the probability that the company offering
the "mini" will exist during the period vital to the product
life.
Surveys usually list items which may be compared
quantitatively, such as memory cycle time. They also
comment on the availability of assemblers and compilers
and various hardware options. The figures which are given
are not comparable from machine to machine because of
differences in architecture. There are no figures of merit for
computer system architecture, nor are there figures of merit
for software.
How, then, should the prospective customer evaluate
competitive minicomputers?
Computers that do things with the outside world usually
require a great deal of software to deal with input/output.
In most systems the computational function is the most
straightforward part of the entire program. The input/output and interrupt service modes of a 'Computer should be
examined closely to determine the effect on software.
Some computers require that interrupts be sorted out by
software or that all input/output pass through the accumulator, creating additional overhead, unless expensive direct
memory access options are purchased. Computers of more
advanced design sort interrupts by transferring control to a
unique memory location for each interrupt. Such computers also feature input/output directly to memory, either
under program control or with direct memory control.
Read-Only Memory
The sophisticated user having a special-purpose application with large quantity sales potential may be interested in
exploring the use of a special-purpose instruction set. This
is practical in a computer equipped with a read-only
memory (ROM). ROM offers advantages in both effective
speed and core memory.
Measuring Speed
Speed is often used as a yardstick of computer performance. The best criterion is effective speed, which combines instruction power and time for execution. Instruction
power is influenced by: number and types of instructions;
addressing modes; indirect address capability; indexing;
interrupt servicing; and input/output facilities. The best
way to evaluate the effective speed of a computer is
through the use of a bench-mark subroutine which is
representative of the most stringent requirements of the
proposed application. The bench-mark subroutine should
be programmed by someone familiar with the computer
being evaluated.
The Importance of Peripherals
In many minicomputer systems the central processor is
only a small part of the entire system. The selection of a
computer should therefore be based also on the availability
of the required peripherals with driver software fully
integrated into the computer's operating system software.
The user will find that some of the independent
peripheral manufacturers have already developed interfaces
for their products to some of the more popular computers.
One should determine the extent of software support for
these independent-supplied periphemls.
Another Measure of Efficiency
Core memory economy is another useful yardstick with
which to evaluate the power of a computer. A computer
having a good selection of instructions combined with a
page-free addressing scheme requires fewer instructions to
accomplish a given task; thus it executes instructions faster
and with less core than a computer having a limited
COMPUTERS and AUTOMA nON for December, 1969
Construction and Maintainability
Today, all computer manufacturers use integrated circuits supplied by the semiconductor iridustry. The difference between computers is mainly packaging. Large circuit
boards can make the cost of spares very high; also they may
21
make the problem of locating a defective component more
difficult than need be.
If you intend to do your own maintenance, investigate
the documentation and training which is available to you.
Find out if any special test equipment is needed and if it is
available at a reasonable cost.
If you must rely on the manufacturer for maintenance,
be sure to find out what response time to expect from his
field office.
Software Evaluation
If the user intends to do a reasonable amount of
programming, it is very important that he evaluate the
available software very carefully. Some common software
packages are listed here with some points to check:
user of all syntax errors. An operation mode in which only
error lines are listed saves much time on basic machines.
What types of expressions are allowed? - Expressions
for expressing data values and addresses are very useful if
the values change from assembly to assembly. Basic assemblers usually limit expressions to the use of the addition
and subtraction operators; more advanced assemblers permit multiplication, division, logical and relational (greater
than, equal, etc.) operators.
Are conditional assembly pseudo-ops available? - The
capability of determining which lines will be assembled by
evaluating an expression saves much programming time. It
i~ possible to write a large prototype program which
handles a large class of problems and then customize it by
setting control variables. The result is a program which is
only as large as is necessary to do the immediate task.
Assembler
Is it possible to produce relocatable code? - Relocatable
code can reduce assembly time because subroutines need
not be assembled every time they are used, nor is it
necessary to always locate routines at fixed origins.
What features are available and what are the configuration requirements for each? - A macro assembler requiring
8K of core is useless if you purchase a 4K machine.
Is the assembler device independent? - If you intend to
purchase high-performance peripherals, determine whether
they can be used with the standard software.
Is an assembler available which may be run on a
large-scale machine? - The availability of an assembler for a
large-scale machine can make life a lot more pleasant, if a
large amount of programmi ng must be done without h ighperformance peripherals.
Is there a macro capability? - Macro's are very useful if
you intend to develop an assembler tailored to a specific
appl,ication j such as electronic testing.
How many passes are required for assembly? - The
important point to consider here is the number of times
that the source tape must be passed in order to produce a
listing and an object tape. Some systems require three
passes with basic equipment. One-pass assemblers, while
gaining speed in assembly, leave much to be resolved by the
loader and, therefore, use many indirect addresses.
What pseudo-ops are available? - Pseudo-ops are directives to the assembler. They increase the usefulness and
flexibility of the assembler.
Are literals permitted? - Literals provide a convenient
method for defining constants and make a program easier
to follow.
What forms of number conversion are available? - Most
assemblers allow for the conversion of decimal integers. The
conversion of mixed decimal and floating point numbers is
usually restricted to machines having more than 4096
, words of core. If the basic assembler does not have these
conversions, check on the availability of a stand-alone
conversion program for this purpose.
Is the object code compatible with Fortran? - Compatible object code permits intermi xing Fortran and assembly
language programs, which communicate through the use of
the COMMON pseudo-op.
How large a program can be assembled? - The limit on
how large a program the assembler can handle is determined
by the room available for the symbol table.
Is a concordance provided? - A concordance is a list of
the symbols used in a program with the address assigned
and the identification of the program I ines that refer to the
symbol. This feature is usually not available in 4K assemblers.
What syntax error detection is available? - It is important that the assembler be able to detect and inform the
22
Fortran Compiler
What type of Fortran is it? - There are two accepted
versions which are identified as BASI C Fortran (USA
Standard 3.10-1966) and STANDARD Fortran (USA
Standard 3.9-1966) which is also called Fortran IV. There is
also Fortran II, but it is not completely interchangeable
with Fortran I V.
How efficient is the output code? - On a small computer one is usually interested in the running time and core
required for the program rather than the compiling time.
Efficient code generation can decrease running time by
almost 50%.
Is it possible to get an assembly listing of the code
generated? - The usefulness of the compiler for real-time
use is enhanced if the user can obtain a listing of the code.
I s the object code compatible with assembly
language? - Intermixing of Fortran and Assembly language
routines is very useful.
Subroutine Library
•
•
•
•
•
What functions are supplied?
What is the execution ti me?
What are the core requirements?
Have the routines been verified?
Are the algorithms described?
Utility Routines
Are the following available?
Debug?
Editor?
Media Copy?
File Maintenance?
Maintenance Programs
•
What programs are available and how well
documented are they?
• What devices must be operational for the execution
of each?
Conclusions
There is no single "right" decision in selecting a computer for a system or systems. Any number of machines can
be used to fulfill a given need. In one-of-a-kind systems the
optimum choice is usually that which can be integrated into
a working system with the least expenditure of time and
effort. The time and effort required depend greatly on the
software and hardware support from the computer vendor.
o
COMPUTERS and AUTOMA nON for December, 1969
CalComp puts the sun out
in 3:1 minutes flat.
It's not doomsday, exactly, but
What's more, CalComp microit's almost as weird.
film drawings can be viewed sepaWhat you're looking at is a rately, like ink and paper drawings.
drawing of what a radio telescope
Or in sequence for time-lapse
sees when it looks at the sun.
motion picture studies.
A computer-controlled CalCalComp is the leader in comComp 835 plotter drew it on micro- puter graphics.
film for The Aerospace Corporation
With sales, service and softin Los Angeles.
ware in 34 cities. And
The people at Aerospace need the greatest variety of
a large quantity of sun complexion plotters in the world.
maps like this one for their study of
Including compact
the sun's radio burst mechanism.
drum plotters. Giant
And by using our 835 plotter,' flatbed plotters. And in
they get them fast.
our 800 series, plotters
In fact, the 835 can draw at a that draw not only on
rate of 100,000 plot increments a microfilm, but on microsecond. Because it draws with film and paper.
modulated light. On microfilm.
So if you have a computer that
(With a choice of five fully auto- can't draw, call your nearest Calmatic microfilm camera options.)
Comp man today.
Designate No. 35 on Reader Service Card
He can help you with just about
any graphic problem under the sun.
Including the sun.
&.88(;) • •
TEACH YOUR COMPUTER TO DRAW.
California Computer Products, Inc.
Oept.A-12. 305 N. Muller Avenue
Anaheim, California 92803
ON THE MOVE
Allen Z. Kluchman
Data General Corp.
Route 9
Southboro, Mass. 01772
They have been nicknamed minicomputers and it's no
secret that they are creating the fastest growi ng market for
computer mainframes. Dozens of new machines have been
announced in the past year and in the next 12 months each
of the five largest producers will deliver more than 1,000
units. One of these large producers, Data General Corporation, is less than two years old.
The machines that are the object of all the interest are
nothing more than small-scale general-purpose computers.
They perform the same kinds of arithmetic and logic, work
with the same kinds of computer peripherals, and use the
same kinds of computer languages as traditional computers.
The minicomputers unique position comes from the fact
that they are the least expensive form of general purpose
computer, and that they are growing to be the most
pervasive.
In the Beginning
The first extensive use of very small-scale general-purpose computers grew out of the demand from instrumentation engineers who were building military aerospace
systems or working in research laboratories. They encouraged the development of the small computers which would
be used as a more economical alternative to special purpose
automatic control systems. These small computer systems
depended on the use of emerging data from complex
instrumentation, but were programmed primarily for control functions. These instrumentation markets for small
computers grew quite naturally to include on-line manufacturing control, such as process control in the chemical field
and the numerical control of machine tools.
These control fields are still a major part of the minicomputer market, and this market is growing rapidly.
Whereas the first small computers were custom tailored to
each application, there is a growing number of standard
scientific instrumentation and industrial control products in
which minicomputers are built in as automatic control
dev.ices. These minicomputer-based systems include such
diverse products as automated drafting machines, computer-based laboratory instruments and milling machines.
Data Processing Capabilities
Allen Z. Kluchman is the Director of Marketing for Data
General Corporation. His responsibilities include market planning and market services for the company. He was previously
the Director of Advertising and Sales Promotion at Digital
Equipment Corporation. He has been involved in the minicomputer field since its beginnings. He holds a B.A. degree
from the University of California at Los Angeles.
24
But something else is happening, too. The data processing capabilities of small-scale general purpose computers are
being discovered. These computers are being used in open
shop time sharing applications, in educational applications,
in dedicated calculation applications, and for computeraided design. In part, these new applications for minicomputers are developing simply because the minicomputers are
becoming a more visible alternative to other sorts of
computation.
COMPUTERS and AUTOMATION for December, 1969
"Computer power now conIes in a new standard size. In the next 12
lnonths each of the five largest producers of minicomputers will deliver
more than 1000 units.))
Data Communications
It was natural development to expand use of minicomputers beyond traditional control applications into the data
communications field. In these systems, minicomputers are
used as specialized control devices for message switching
and message concentration to improving the efficiency of
real time sharing networks. The next step was to extend the
use of minicomputers to become a free standing element
within large multi-computer networks. In these networks,
minicomputers are used as sophisticated terminals. Not
only do they collect and send data, but they also perform
small-scale data processing. The next step is evident. The
distance from small data processing terminal to small-scale
free-standing data processing is short. A variety of data
processing packages are being developed. The mi nicomputer
is becoming truly general purpose.
A New Style Computer Network
An example of the natural extension of minicomputers
from control to data handling applications can be seen in a
new style computer network such as the transcontinental
time-sharing system being established by Computel Systems
Ltd. of Canada. Computel has two large scale UNIVAC
1108 computers and an IBM 360/65. To these they have
added a group of minicomputers manufactured by Data
General Corporation.
Computel sells data processing capabilities as an extension of the existing in-house data processing facil ities of
their customers. Computel uses three minicomputers in
front of each of their large computers. The minicomputers
will be used to collect data sent in to Computel by
customers allover Canada via telephone lines. The small
computers will not only save money on valuable central
processor and memory time on the large computers, but
they will save money on the phone bill as well. Whereas
large computers used alone would require one phone line
per device per customer, each small computer handles up to
six high-capacity customer devices on a single line. Computel will also make minicomputers available to its customers to be used at the terminal end of the network for
peripheral processing, and to save on data communications
costs.
Major minicomputer manufacturers can afford to support new markets with application products. For example,
each of the major minicomputer manufacturers has developed time sharing systems built around their computers.
These time sharing packages extend the basic price/performance features of minicomputers into broadening
markets.
Minicomputer Middlemen
But perhaps the most important reason for the growth
of minicomputers into new applications areas has been the
development of a group of minicomputer middlemen whose
business is to take the minicomputers that others manufacture and apply them to new areas. Aggressive new compan ies have started whose entire business plan is based on
the minicomputers' improvement in price/performance over
the way a job has been done before.
A few examples will illustrate the contribution that
these middlemen are making in the application of minicomputers to new areas.
A System for Schools
A complete sixteen bit word x 4096 memory on a single 15-inch
square circuit board.
COMPUTERS and AUTOMATION for December, 1969
Educational Data Systems (E OS), Newport Beach, California, has designed computing systems specifically for the
secondary school curriculum. They not only provide the
necessary hardware and software, but they also provide an
extensive library of instructions materials. Furthermore, the
company assists schools to plan and implement a complete
educational computing program including instructional
a?plications for teacher training and administrative applications. The EDS series of systems is designed around Data
General Corporation's Nova computer, and provides time
sharing to schools at a fraction of commercial time sharing
25
,
A minicomputer in a typical rack-mounted configuration in a control application.
service. Whereas the cost from commercial sources would
range from $5 to $10 per terminal hour, the cost of the
minicomputer time shared system is about $1 per termi nal
hour, including all terminal and communications costs. The
EDS systems use the BASIC language developed at Dartmouth as its primary software package. But the company
also has a time sharing calculator package designed to
provide students on this machine easier access to a computer.
Warehousing
Another example of the use of minicomputers in noncontrol applications is the system developed by Industrial
Computer Laboratories, Inc., a Division of Information
Technology and Systems Inc., Salt Lake City, Utah. This
system, used for warehousing applications, consists of a
series of free-standing minicomputer systems which are
used as terminals to large-scale computers. Each free-standing remote minicomputer system is tailored to a particular
warehouse situation, and each is designed to be used by
unsophisticated warehouse personnel. The small computer
is programmed to operate in a question and answer mode.
The computer presents the series of questions and warehouse personnel check off the correct answers. The on-site
minicomputers avoid using expensive data communications
facilities for routine man/machine interaction. They also
perform certain data processing, such as inventory control,
payroll, etc., on-site at the warehouse.
When the on-site computation is done at the warehouse,
the minicomputer terminals condense the data and send
26
relevant information over telephone lines to the corporation's central facility and large-scale computer for further
data processing and for management information.
Small control terminals are a network such as the one
I nternational Computer Laboratories has designed for a
major retail operation to allow the retailer to take advantage of a party line telephone cost and reduce the number
of telephone lines required. In this instance, it was possible
to cut the telephone bill from a potential $10,000 per
month to $1200 per month. Just as important, the retail
warehouse network was designed around existing manufacturer production. Thus every computer in the network acts
as a translator between the individual warehouse department and the machine's central processor.
In each ·of these examples, the potential of the smal~
scale general-purpose computer has been applied to an area
in which the middleman has extensive experience. In each
case a complete packaged solution was developed to extend
the use of minicomputers to areas in which they were
underemployed.
For the Future
What does the expanding use of minicomputer mean to
the computer user and to the world at large? Basically, it
means that computer power now comes in a new standard
size. Because there are many, many uses for a computer
that can justify no size of computer other than a mini,
minis will extend the use of computers. This extensive use
will make computers more accessible-and better understood-tools. Computers will become more common.
D
COMPUTERS and AUTOMATION for December, 1969
..
We may have helped, because we've been recording computer outputs on film since 1964. State of the art today is our FR-80 Computer
Output Microfilm Recorder - a data processing oriented system
that produces higher quality images than any C.O.M. system
in the world.
What does "data processing oriented" mean? What does "quality"
mean?
Data processing oriented in the FR-80 means you plug it in, load
your mag tape, and record. No special programming. No elaborate
procedures. FR-80s are usually recording customer data less than 4
hours after uncrating. FR-80 is already programmed, and it accepts
tape formatted for any output device, from any host computer.
Quality in the FR-80 means reduction up to 60x on 16mm film with
resolution that yields book-quality characters on printing plates.
Business graphs and drawings are equally sharp. And forms are
perfectly registered with their contents. It also means around the
clock performance without adjustments.
But the best thing about the FR-80 is its flexibility and responsiveness to varying requirements without mechanical or software
changes. Just about everything is controlled at the keyboard.
Turn-around time is measured in moments.
FR-80 sounds too good to be true. But remember, we're delivering
the results of 5 years of hardware and software development. Let
us send you the complete story. Or call us: 213-478-2571.
Information International, 12435 West Olympic Boulevard, Los
Angeles, California 90064, (213) 478-2571; 89 Brighton Avenue,
Boston, Massachusetts 02134, (617) 787-4700; 7880 Coolridge
Drive, Camp Springs, Maryland 20031, (301) 449-4248.
INFORMATION INTERNATIONAL
Designate No. 16 on Reader Service Card
C.O.M. has come a long way
THE IMPACT OF MINICOMPUTERS ON INDUSTRY
Dr. Karl Hinrichs
Lockheed Electronics Co.
6201 E. Randolph St.
Los Angeles, Calif. 90022
The potential for improved products at lower unit cost
by automation and computerization is very well known,
but the implementation is still in its infancy. The recent
and continuing development of minicomputers (high speed
and competent machines selling for less than $20K) will
provide a very rapid acceleration in industrial automation.
Cost Breakthrough
The inertia that has kept us from full automation in the
past is rapidly vanishing because of the dramatic economies
provided by minicomputer systems. You can rent a competent minicomputer on a 40-hour week basis for approximately $1.80 an hour. This is truly a breakthrough which
was difficult to predict a few years ago.
In terms of purchase price, a 16-bit competent computer
with Teletype, high speed paper tape punch and reader, and
four hardware interrupts was priced at well over $100K five
or six years ago. This same power-except several times
faster-is available today at one-tenth the price.
The rapidly expanding market for minicomputers is
producing a snowball effect in industry: the more production, the lower the cost and price; the more applications, the larger the program libraries available for new uses;
the wider the variety of customers, the broader the variety
of peripherals available.
The large, centralized computer (" max icomputer") will
not vanish from the typical industrial plant, but wi II be
augmented by many dedicated minicomputer installations.
Some of these minicomputers will be connected to the
maxicomputer, forming a hierarchial memory and computer system.
Tolerances in industry continue to decrease. Step-andrepeat operations are becoming much more common. Digital control of mach inery is an accelerated trend. I n all these
applications, computer automation offers distinct advantages over manual techniques. Computers are wonderful for
routine functions, since they do not get bored with the
most tedious of tasks or strained by high precision work.
A Three-Way Choice for Automation
A three-way choice is available for automation. A dedicated minicomputer system may be purchased or leased to
solve specific functional requirements. A computer terminal
may be obtained for interaction with a maxicomputer at
MANAGEMENT
!o
Po
0
....._ _ _ _. . .
00
(j
o
n
o
1/0
PERIPHERALS
0
00
000
0
Dr. Hinrichs is the Director of Engineering for the Data
Products Division of Lockheed Electronics Company. He is
responsible for all of the engineering activities of the
Division, including research and development, new product
planning, all product design and engineering, customer liaison, and technical contacts. He received his B.S. in Electrical
Engineering from Swarthmore College in 1945; his Master if
Science from Harvard in 1947; and his Ph.D. in Electrical
Engineering from the University of California at Berkeley in
1955. Dr. Hinrichs has published five technical papers and
holds 14 patents. He is a member of IEEE, ISA, and several
other technical organizations.
MARKETING
CONTRACTS
o
o
o
o
o
IpURCHASINGI
0..
OPERATIONS . , 0 0 c»
MANUFACTURING
MAXI
COMPUTER
DATA
BANK
.
0
0
0
0
ADMIN I STRATI ON
ACCOUNTING
OOOO~L._P_E_R_S_O_N_N_EL.......
ENG INEERING
P RODUCTIP ROJECT
INDUSTRIAL WALKWAYS
TO CENTRAL COMPUTER
flo
Figure 1.
28
COMPUTERS and AUTOMATION for December, 1969
"How do you decide when to use a minicomputer? If a process is relatively isolated frorn the data sources used by the remainder of the
plant, if it is a job that is prone to change, if the solution is needed
rapidly, if it involves high input/output rates, if it involves manmachine interaction-then you should use a minicomputer."
some remote site. A centralized maxicomputer can often be
employed for automation, with inputs provided by specialpurpose remote devices or batch tapes and cards.
Today's typical industrial plant has a maxicomputer
which serves all segments of the plant as illustrated in
Figure 1. The Table 1 lists associated applications. The' first
item, batch processing, is one where minicomputers usually
cannot compete with centralized maxicomputers. Despite
the high initial cost of the centralized maxicomputer, its
greater flexibility, memory storage, span of peripherals, and
program library offer a more economic solution to batch
processing' than that which can be obtained with today's
minicomputers. Accounting ledgers,. payroll, timekeePing,
charge allocation, P/L analyses and reports, order entry,
marketing records, and personnel· records are all typical
industrial fUl'lctions economically performed by the centralized computer. A very large and valuable library of software
programs has· been developed for these functions, usually
with: specific adaptations for individual. plants .. Most of
these essential records and reports need large amounts of
memory. verY' extensive program I ibrar.ies, and a wide span
of peripherals. Except for new and small industries, minicompu.ters will not compete in the near future.
I nventory Counting
Inventory counting is a little different from batch
processing. Historically, it has been a batch record-keeping
function; but we are more and more concerned with
virtually real-time control of inventory and purchasing. We
would like to see a system in which Purchasing, Stock
COMPUTER-AIDED INDUSTRIAL TASKS
1.
Batch Processing and Records
2.
Inventory Accounting
J~
Production Plan and Control
4.
Process Control
5.
Product Testing
6.
Release and Engineering Control
7.
Cost Estimating
8.
Drafting and Printed Circuit Layout
9.
Engineering Design
Room, and Receiving I nspection are inexpensively interconnected with inputs from Engineering (Bill of Materials) and
Manufacturing (kit scheduling). The minicomputer (particularly one of the larger and more powerful types) competes well in many cases with the centralized facility, since
it does not require a large peripheral investment, can be
easily relocated, and permits ready entry of data and
program.
Production planning and control is also an area where
minicomputers often provide economies compared with
maxicomputers. Production planning and control, work-inprocess reports, exception reporting, station reporting,. and
other nearly-real-time manufacturing information services
can often be efficiently provided by ded icated mi n icomputer systems.
Process' Control
Process control is the classical area for minicomputer
application; and the current industry obtained its impetus
from this field .. Special-purpose computers have been used
in processes for over 20 years. The early systems for data
acquisition and control used both digital and analog computers. The economics of scale are swinging the processcontrol automation towards purely digital computers. Our
large, complex and intertwined processes in many industrial
plants wou Id appear to recommend a maxicomputer installation to simultaneously assess and compute all parts of the
control process. This grandiose concept is analogous to the
universal plant Data Base. Realistically, however, there are
many unknowns in the interaction between processes in
total plant control; and a hierarchy of control is employed
with many minicomputers and special-purpose controllers.
A maxicomputer is sometimes used for centralized surveillance. For functions which are extremely well known,
the minicomputer does not compete with the specialized
controller. Most processes using direct digital control, however,. involve functions not presently well defined or optimized. A minicomputer can be used for optimization study
as well as fixed-program control. A minicomputer controller is also an excellent record keeper Clnd performance
analY/er. If the process is sllsceptible to irnprovement by
the incorporation of Clddi tional loops, additional sensors,
new outputs, or other modifications, a minicomputer may
be the most economical answer to process automation.
Product Testing
Table l.
COMPUTERS and' AUTOMATION for December., 1969
Product testing is also a classical area for the use of
minicomputers. In the past, economics have dictated the
use of special-purpose machines. Now the lower cost of
min icomputers permits a vast increase in flexibi lity at low
29
cost. This market is increasing tremendously in response to
the testing requirements for more sophisticated products
with tighter tolerances and more stringent requirements for
proof of reliability and performance. One of these
products, of course, is the minicomputer itself. The proof
of performance of a minicomputer requires a high degree of
automation for economical testing. In contrast with the
applications for which the centralized computer is optimal,
here the programming is minimal, few general-purpose
peripherals are used, and the I/O requirements are high.
Although the product testing routine does not require
much flexibility, a tester must be readily modified on a
day-to-day or month-to-month basis. Therefore, the general-purpose minicomputer usually out-performs the specialpurpose or harGl-wired test-device controller.
Engineering release and control has received a lot of
attention but little actual use of computers, except in the
very largest industrial installations. The availabil ity of inexpensive minicomputers and the rapidly growing libraries of
software will readily invade the engineering control area.
The requirement for a very large data bank (perhaps with
extensive parts specifications, vendor records, and simi lar
files) can often justify the use of Teletype stations, CRT
displays, and a centralized data bank.
Cost Estimating
Engineering and manufacturing jointly share costestimating functions in a typical industry. It is surprising to
find a small extent of computer invasion in this function
today. Perhaps company management is lax in forcing a
marriage between the estimating staff (who do not know
computer programming) and the centralized programming
staff (who are often unaware of the amount of estimating
under way). The situation should change radically when it
is realized how efficient minicomputers with CRT and TTY
can be, in assisting the cost estimating process. Time-sharing
systems are also logical contenders for estimating service,
although many companies will not permit their company's
price files to be located outside company premises, for
security reasons.
Engineering Design
Engineering design is the classical area for time-sharing
service use. For very large plants, their central ized maxicomputer can be used with remote terminals. We are all
familiar with the advantages of computer-aided circuit
design (ECAP and similar programs). There are many other
important engineering functions which can be improved
and cost-reduced by utilizing computer assistance. The
minicomputer has definite advantages for use with graphic
terminals in design functions requiring intensive manmachine interaction.
Choosing the "Right" Computer
How do you decide when to use a minicomputer, a
maxicomputer, a time-sharing service, or no computer at
all? The return-on-investment factors are usually straightforward. If the job is a complex one involving large banks
of data, if it involves interaction with the data from many
physically separated plant functions, if it can be done on a
batch (non-real-time) basis, if it does not require manmachine interaction, if it can be interrupted, if it utilizes
extensive programs and employs a lot of peripherals - then
the centralized maxicomputer batch-processing system is
the obvious solutions and can be justified by a standard
cost comparison.
However, if the process is relatively isolated from the
data sources used by the remainder of the plant, if it is a
30
job that is prone to change, if the solution is needed
rapidly, if it involves high I/O rates, if it involves manmachine interaction, then we would want to use a dedicated minicomputer. A graphic CRT design terminal is an
excellent example.
There are many industrial functions which defy computerization. Creative and intelligent human beings are also
required, since you cannot buy hardware and build software to solve an undefined problem. Some urgent requirements of industry change faster than our ability to program
machines. If you have a product which changes completely
in six months, computerization is ill-advised. There are
many classical examples of large investments in automation
for processes which human ingenuity has obsoleted.
Examples of industrial functions which minicomputers
should dominate by the end of the 1970's are: drafting,
electrical and mechanical design, printed circuit layout,
numerical control tape production, template production,
data acquisition, component and system testing, and process control.
The I mpact of Minicomputers
The impact of minicomputer automation wi II be truly
remarkable in the next decade. We are apt to ignore the
magnitude of change because we are watching it every day,
and progress often seems slow. Drawing back for a longerrange look, however, we can readily detect some very
sign ificant impacts on industry.
The first effect, common to all improvements and
efficiency, is less raw-labor content in our products.
Economics cannot be denied, and we will obtain a larger
production of lower-cost products with a more skilled labor
force. The proliferation of minicomputers and minicomputer systems will result in a very sharp increase in
trained personnel: a new class of technicians or "miniprogrammers". One of today's critical industrial shortages is
that of skilled programmers. The intimate working experience of the hourly working force with small dedicated
computer systems will enrich our technical labor base.
Although much of the training and resultant skills will be
very specific, it is a valuable background for more general
programming education and in any event wi II free our
professional programming staff from much of today's
necessary routine.
The product cost reductions predicted will be particularly noticeable for complex products such as electronic
instruments. Minicomputers, therefore, will continue to
decrease the ratio of price to performance, both from
increased volume and increased manufacturing automation.
The lower prices will produce an avalanche effect not only
in industrial applications, but in computer use in virtually
all spheres of human activity.
It will be a big boost to small business. It will be possible
for small businesses to compete with large companies in
complex electronics because a minicomputer system wi II
provide them with almost all of the functions, if not the
efficiency per second, as the large companies obtain from
their combination of minicomputer systems and centralized
maxicomputer. Software libraries are the hold-up today,
but tomorrow that will not be true. Minicomputers will be
equipped with business systems and batch processors
w~ich, although perhaps not as cost-effective per dollar,
will nevertheless provide small businesses with the ability to
have a full set of features that the large company now
possesses.
The wide application of minicomputers and computer
systems will generate valuable additions to the United
States' technological base in new products, new programs,
technological training, and new computer system applications.
0
COMPUTERS and AUTOMA nON for December, 1969
f
>
,
Sangamo keyboard-to-tape Data Stations are the
easiest to learn and operate. Thanks to alpha-numeric display, there are no Octal or EBCDIC codes
to learn. Electronics are 400% faster than similar
equipment. Doubles as a verifier; magnetic core
memory allows fast, easy correction of operatorsensed errors. Human engineered for greater oper-
ator comfort and efficiency; tape threading takes
mere seconds. There's more desk-top work space,
more leg room. Memory display is in line with keyboard-no head turning. Operates more quietly,
and operators appreciate the personal belongings
drawer. Now being delivered in two series: 05-7000
and 05-9000. Demonstration? Call or write:
Information Systems Division, SANGAMO ELECTRIC CO., Springfield, Illinois 62708
Designate No. 25 on Reader Service Card
AUTOMATED EXPERIMENT CONTROL AND DATA
ACQUISITION - A MINI-COMPUTER APPLICATION
M. A. Shah and W. H. Stieger
Chase Brass and Copper Co.
University Circle Research Center
11000 Cedar Ave.
Cleveland, Ohio 44106
Dr. W. Kipiniak
Computer Inquiry Systems, Inc.
Fort Lee, Virginia
hi~hly
((Although programming and system implementation require a
skilled staff) once a control system using mini-computers is runnzng)
it is so reliable and simple to use that no regular computer staff is
required.))
I n recent years, scientists have used small computers as a
versatile research tool for automatic control of experiments
and data acquisition. At the Casting Laboratory of the
Chase Brass and Copper Company, a mini-computer-the
Hewlett Packard 2116B-is working in a system which, in a
fully automatic fashion, performs thermodynamic and
kinetic measurements in metallurgical systems over a wide
temperature range.
I n these experiments, control as well as data acquisition
are well-defined procedures to be repeated a large number
of times. The use of a computer results in increased
research productivity and reproducibility of measurements.
The very high computational speed of today's computers
makes possible the automatic and nearly immediate display
of results of such experiments. The time of the experimenter is released for experimental planning and analysis of
results. The operating costs are well below those associated
with continuous human monitoring of such experiments.
h
\1
!.i
,
main memory space and computation time. While some
manufacturers of small computers offer basic FORTRAN
and ALGOL compilers, it is often necessary to use an
assembly language to fit all the programming into the
computer memory.
LIQUID METAL PROCESSING-THERMODYNAMrC STUDIES
Automatic Data Acquisition and Experiment Control
ANALOG RECORDING
IXPfRIMfNT
I--'~-"""
I
Why a Mini-Computer?
"The works in a drawer", as a recent television advertisement tells us, indicates the reduced size and modularity of integrated circuits and the cleverness of modern
circuit designers. Their advances have measurably improved
the versatility and reduced the cost of computing equipment. Furthermore, current designs incorporate universal
input/output interface cards which allow the connection of
external instruments and other devices to the computer.
This makes implementation of real-time, on-line applications significantly easier.
Application software design and programming for data
acquisition with a small computer presents a challenging
problem in data manipulation. It is usually necessary to
reduce the output records to minimum size to conserve
Mr. Shah is a leading computer analyst with the Information and Communications Services Division of Chase Brass
and Copper Co.
Mr. Stieger is the Acting Manager of the Information and
Communications Services Division of that same company.
Dr. Kipiniak is currently the president of Computer
Inquiry Systems, Inc. He was formerly associated with Digital
Applications, Inc., the company in charge of dev~loping the
system described in this article.
32
f:
[~~
r
- - . -~-._- . ____"J
~'Y,,"'M'~~,,"aa<
__ ,u.~
1
Source Exif Data:
File Type : PDF
File Type Extension : pdf
MIME Type : application/pdf
PDF Version : 1.3
Linearized : No
XMP Toolkit : Adobe XMP Core 4.2.1-c043 52.372728, 2009/01/18-15:56:37
Producer : Adobe Acrobat 9.1 Paper Capture Plug-in
Modify Date : 2009:03:20 22:27:17-07:00
Create Date : 2009:03:20 22:27:17-07:00
Metadata Date : 2009:03:20 22:27:17-07:00
Format : application/pdf
Document ID : uuid:6d003729-d4c7-4787-b4b5-5c8d5a369027
Instance ID : uuid:7dc9e62c-5b7b-4f0c-9860-9d127b450547
Page Layout : SinglePage
Page Mode : UseNone
Page Count : 60