Computer Centre Bulletin, Volume 2 Number 10, 6th October 1969

User Manual: Computer Centre Bulletin, Volume 2 Number 10, 6th October 1969

Open the PDF directly: View PDF PDF.
Page Count: 16

DownloadComputer Centre Bulletin, Volume 2 Number 10, 6th October 1969
Open PDF In BrowserView PDF
UNIVERSITY OF

QUEENSLAND

COMPUTER CENTRE

COMPUTER
CENTRE

BULLETIN
_

- . ... uu~==~=========================
Volume 2, Number 10.

Editor:

6th October, 1969.

H. L Smythe.

";'.\" .. :"

THIS EDITION
This issue of the Bulletin contains two major articles that should be
of interest to the reader, one discussing, in general terms, the position of
the law and computers, and the other describing the addition of hardware to
the PDP 10 to improve further the efficiency of multi-programming.

The

Bulletin continues to inform readers of FORTRAN errors on the PDP 10, and a
Letter to the Editor is also published.

LETTERS TO THE EDITOR
Recent events have shown that man is capable of travelling to the moon
and back in less than two weeks.

In the same amount of time, it is possible for

any of us, using more conventional means of transport, to travel around the world,
even stopping off for occasional sight seeing.

In far less than two weeks, as recent history shows us, a war can commence
and be won - or lost, according to one's particular outlook.

Yet recently, some

data I submitted to the Computer Centre were still awaiting preparation after
two weeks.

The only apparent movement was the transposition of my name from the

end of a list to a position third from the top.

The data were eventually prepared

two weeks and five days after their submission and subsequently analysed by the
GE 225 in four minutes.
It seems to me that this situation is ludicrous and unnecessary.

If the

university is able to spend a large. sum on a modern computer, surely it should
be possible to provide those ancillary services necessary to ensure its
efficient use.

If one has to wait for over two weeks to have one's data

prepared, efficient use of the computer is not being made.
Surely it is possible to have faster data preparation than thts.

Many of

us have either seen or heard of centres where a delay of more than a day for data
preparation is the exception rather than the rule.

144

I have no immediate ambition to travel to the mooh, to have a
round-the-world trip, or participate in a war, however little time is required.
But together, I am sure, with many others at the University of Queensland, I
would like to receive my prepared data cards before I have had time to forget

MaZcoZm A. CoZston (Department of
Kenmore~ Q. 4069.

what they are all about.

5 GZencarron

Street~

Education)~

[Ed. Note:- Mr. Colston's lett~r raises a query about the type and quality of
service that the Computer Centre provides for keY-punching.
The Centre is essentially a scientific data-processing centre, not a
The data preparation service we offer is only a token
service. In'fact, departments with significant punching reciuestshave installed
keypunches for their own data preparation,

data-preparation centre.

Our records show that the average turnaround time for punching is
approximately three days. There are, however, periods when the delay builds
up because of sickness or changes of staff. This is unfortunate, but,to
guarantee a maximum.turnaround time of only one or two days, key-punching staff
would have to be increased considerably in order to cope with the peak demands
rather than the average. This would mean a significant increase in charges for
punching, and would, no doubt, inspire numerous Letters to the Editor!
Incidentally, Mr. Colston's implication that the University purchased the
computer and finances the ancillary services, is incorrect. The comp~ter was
purchased with funds from a variety of sources, the University contributing about
one-third. In particular,the University does not contribute to the Centre's
data-preparation section. In fact, the cost of providing the key-punch service is
barely covered by the charges ~hat are made.]
..

FORTRAN IV ERRORS
1.

It has been discovered that certain types of integer expressions do not

compile correctly on the PDP 10.
For example:
K

= L*(N-l)+(M*(N-ll)/2

At first sight, this would seem to be a fundamental

ta~ling

ot the

comp~ler.

However, a number of circumstances must exist before the error occurs.
(a)

There must be some common expression in the statement.
example, this is (N-l).

In the above

1'.......

.....

6,..

~

11

'BETTER GET THE A.S.1.0. ONTO THIS'

(b)

This common expression must be further operated on by other var;taole$,
different for each occurrence of the common expTession, e.g.
L*(N-l) and M*(N-l}

(c)

The last of these composite expressions :must oe further operated on by
an integer divide, e.g. (M*(N-lll/2

Incorrect code is generated by the compiler.
conditions does not

exist~

If anyone of the aoove

the correct code wtll oe produced,

It is

emphasized that the problem occurs only with integer expressions.

146

2.

Variable dimensioning of arrays is permissible within subroutines and

is a convenient means of creating a generalized routine for, say, matrix
manipulation.
Where the type of an array is not implicitly defined, the appropriate
type statement must appear before the dimensioning statement for correct
results, although documentation would lead one to believe that the order of
specification statements is not significant.
Thus:

SUBROUTINE SR(A,Nl,N2)
COMPLEX A
DIMENSION A(Nl,N2)

or

SUBROUTINE SR(A,Nl,N2)
COMPLEX A(Nl,N2)

are correct, but

SUBROUTINE SR(,A,Nl,N2)
DIMENSION A(Nl,N2)
COMPLEX A

will lead to incorrect results.

3.

GE 225 FORTRAN IV ERROR
The form of expression that may be used as a subscript, is limited to a

few simple expression types and does not include the use of a subscripted integer
variable.

Thus, an expression of the form:

141

DIMENSION A(12),II(15),JJ(15),X(15)
A(II(J))

= A(II(J))

+ X(JJ(J))

will not produce correct code although no compilation error is generated.
While the compiler

may, in fact, produce the correct code in some similar

cases, it cannot be relied upon to do this, and thus should not be used.

Any suspected errors should be reported to the Administrative Officer
(Mr. John Jauncey, extension 8471).

PDP 10 MEMORY PROTECTION, AND RELOCATION

E.
In a previous article,

Mu~tiprooessing

J.

SokoU

Conoepts by 1. Oliver

(Vol. 2, No.3, 3rd March, 1969), reference was made to the concept of

multiprogramming and the necessity for special hardware to provide this
method of operation.

This feature is available on the PDP 10 and allows a

preset number of programs to be executed, even though they may not all
exist in core at the same time.

This is achieved by swapping programs between

core memory and the fixed head disk (which has the characteristics of a drum)
as dictated by the core space requirements of the program to be executed.
Thus, in the process of execution, a program may be swapped in and out of core
many times.

Under these conditions , it is desirable that a program be capable

of execution regardless of where it is stored in memory.
provided by hardware relooation.

This capability is

Since a number of programs, including the

executive for controlling the operation of the whole system, may exist in
memory at the one time, provision must be made to ensure that these programs
do not destroy one another, and this is provided by hardware proteotion.
Before describing the operation of the relocation and protection hardware,
it is necessary to consider the events leading to the execution of a program.
As a result of compilation or assembly of a source program, a relocatable

148

object program is produced, where the instruction addresses are relative to
zero.

By means of a relocatable or linking loader, a number of relocatable
\

programs and subprograms may be combined, or linked into a composite program
ready for execution.

The linking of program A, consisting of subprograms AI,

A2 and A3 of length k, m and n, respectively, is illustrated in Figure 1.

0

0

Al

Al
k-l

-k

-------

-

-- ------

k+m

Program

A2

0

A

A3

A2
m-I

k+m+n

o
n-I
Before linking

After linking
Figure 1.

LinkJ.ng of. Program A

The addresses of programs AI' A2 and A3 were each originally relative to
zero.

After operation of the relocatable loader, addresses of programs A2 and A3

have undergone software relocation by an amount

k, and k + m respectively,

and now all addresses in the complete program A are relative to zero.

So far then,

the individual subprograms and .components of a complete program have been linked
together, relocation of addresses has taken place, and the program is ready for
execution.

However, the program will not be stored from physical location zero

since the resident executive occupies the first 16K of memory.
determine what physical space the program will occupy.

The executive will

To ensure that the program

will be executed correctly in the physical addressing space, relocation of
addresses must be performed.

This is achieved by hardware reZocation during

execution of the program.
To simplify the design and implementation of the protection and
relocation hardware, core memory is divided in blocks of lK words.

To ensure

that each program is loaded at the start of a block, each program is allocated
an integral number of lK blocks.

The program A illustrated in Figure 1 occupies

k + m + n locations, and may require PI + 1 blocks of core memory, designated

o

to f l '

Clearly, a portion of the last block, PI' may not be utilized.

When a program is placed in core memory for execution, it must be stored
beginning at a lK boundary.

As shown in Figure 2, the program A occupies PI + 1

blocks, starting at block RI'

However, the addresses in program A are still

Block No.
Q

Executive
(Pl+l)K
words

A

Other User
Programs

P

A

Other User
Programs

Non~

1/

777777 __e~/x_i.s_t~~_nt~~.
logical addressing space

Figure 2.

physical addressing space

Logical and Physical Addressing

150

relative to zero prior to execution.

The valuesof protection, P l , and relocation
Rl , are placed in the protection and relocation registers respectively by the
executive. When execution begins, a memory reference is first checked to ensure that
it is legal.

The number of the block of memory being addressed is compared with

the contents of the protection register, and if the block number exceeds Pl'
execution of program A is interrupted and control is transferred to another program.
In this case, the executive would indicate that an illegal memory reference had
occurred.

If the protection check is satisfactory., then relocation is performed

to transform the relative program address to an absolute physical memory address.
This is achieved by adding the contents of the relocation register, R1 , to the
relative program address.

This is not done on all addresses.

Index registers

and accumulators reference one of the first sixteen locations in memory, and since
these are integrated circuit registers with a very low access time, relocation is
not performed to allow all programs to use these high performance registers.
Program A may be swapped out of core and later swapped back in'to continue
execution.

If this occurs, it is not necessary for A to occupy the same physical

area that it occupied previously, nor is it likely or desirable.

Before execution

of A is continued, the executive reloads the relocation register with the
appropriate relocation constant.

base address.

This constant is sometimes referred to as the

A program may also dynamically alter its memory requirements, in

which case the executive will modify the contents of the protection register to allow
more or less memory to be used by a program.
In a static situation, during which swapping is not in operation, a number
of programs may exist in core memory.
$aheduZel'~

A program in the executive, known as the

determines which program should be run next.

When this is determined,

the executive will set the appropriate values in the protection and relocation
register, and pass control to the selected program.

After a short period of time,

control may be passed to another program which requires use of the processor, but
with values for pro.tection and relocation which apply to this particular program.
So far, the operation of a simple form of relocation and protection which
was incorporated in the PDP 10 has been described.

Recently; additional hardware

has been installed in the PDP 10 to improve further the efficiency of
multiprogramming.

This additional hardware includes a second set of proteation and

relocation registers, and the advantages of this feature are now discussed.

151

It is possible that a number of jobs may require a FORTRAN compilation.
In the original PDP 10 system, a copy of FORTRAN compiler was loaded for each
job which required it, requiring 11K of memory for each copy of the compiler.
With more than two such jobs, swapping was necessary to service other programs
requesting access to the processor.

As the number of programs requiring

execution increases. then the processor is involved in a larger amount of
swapping, and so processor efficiency decreases.

It also results in a poor

response time of terminals in a time-sharing system.
A program may be designed so that it is capable of being re-entered by
a number of different programs.

With this capability. a

be shared by any task in the system which requires it.

re~entrant

program may

For frequently-used

programs such as the FORTRAN compiler, this capability is very desirable, since
only one copy of the compiler is required in core.
In general, a re-entrant program consists of a pure part and an impure
part.

The pure part is that part of a program which is never modified during

execution, and would normally include most of the instructions and constants,
while the impure part may be modified during execution. and would include areas
for buffers, tables and results.
In the case of the PDP 10 FORTRAN compiler, the pure part occupies9K,
while the minimum size of the impure part is 2K.

With 32K of memory available,

a maximum of 11 FORTRAN compilations could be performed without the need for
swapping.

If swapping is necessary. only the impure part is swapped out onto

disk, with a resultant saving in the swapping time, and more economic utilization
of disk storage.
For flexibility, it is desirable that the two parts of are-entrant
program occupy non-contiguous areas of memory,

and~

to ensure execution of either

part and access from one part to the other, two sets of protection and relocation
registers are used.

To ensure that the pure part is not altered, the hardware

can be enabled to prevent information being written into this part.
In this environment, as control is passed from one job to another, the
executive now must load both

set~

of

152

reg~sters

with the appropriate values.

Figure 3 illustrates the transformation from logical addresses to physical
addresses with the two sets of registers.

PI and RI are the values of protection

and relocation for the impure part, while P 2 and R2 are the values of protection
and relocation for the pure part.

o
PI

0

Impure A

Executive

f

RI

f

Other Users

1

-t

R2

II
roo-

- I~

Impure A
x
Other Users
x

Pure A

1

Pure A

1//

Other Users

Illegal

I___

777777 ...I__

-L

logical addressing space

Figure 3.

7777 77

l/NIon-existent/
/ / / I

l?li.y~;tcal

I

I

I

I

address:;[.ng space

Relation between Logical and Physical addressing.space with
dual protection· and relocation

153

There is no doubt that the re-entrant hardware will improve the
efficiency of the multiprogramming capability of the PDP 10.

The executive

and system programs such as the FORTRAN compiler have been modified to make
use of the re-entrant hardware,
difference in batch jobs.

Externally, there will be no noticeable

When time sharing is provided,however, it should

mean that a larger number of users will be serviced without any reduction in
response time.

THE lAW AND COMPUTERS
This artiole is adapted from the Inaugural Leoture of
Douglas J. Whalan, LL.B. (N.Z.), LL.M. (N.Z.), Ph.D. (Otago), Professor of Law
in the University of Queensland. The University of Queensland Press is to
publish the leoture in its Inaugural Leoture Series.
Professor Whalan was born in New Zealand in 1929, qualifying as a
solioitor (1950) and barrister (1951) of tne Supreme Coux>t of New ZeaZand.
In 1958, he was appointed Senior Leoturer in Law at the University of
Auokland, and gained the first Dootorate of Philosophy in Law to be awarded
by a New Zealand University. While on study .leave in London in 1963-1964,
Professor Whalan furthered his researoh into the registration of titles to
land. From 1964-1967 he was suooessively Visiting Fellow, Fellow and Senior
Fellow in the Researoh Sohool of Sooial Soienoes at the Australian National
University, and, in 1967, he was appointed Professor of Law at the University
of Queensland.
Professor Whalan is a member of a Federal Attorney-General's Working
Group on Refo~ in the Australian Capital Territory. His publioations inolude
a book on the law of trusts and trustees, and numerous artioles on the fields
of wills, equity, registration of titles t9 land, and oomputers and the law.

In his Inaugural Lecture, Professor Whalan pointed out that the issue
of the law and computers was of vital importance to the development of law
in society.

He strongly deprecated the fact that, in its attitude towards

computers, Australian society was, like Gaul, divided into three parts - one
part fears, one sneers, and the other cheers.
the law barely hears I

Regrettably, said Professor Whalan,

This is one of the principal themes of the lecture:

either

lawyers quickly wake up or they will be run over by the rapidly-revolving wheels
of the computer revolution.

Professor Whalan also discussed the legal problems

arising from the application of computer technology to fields such as tort,
evidence, individual freedom and privacy.

154

The conclusion he reaChes is that

radical legislative changes are urgently needed to meet the demands of the
computer age.
adapt;

The lawyer must think for the future, be flexible and ready to

above all, he must recognize that scientific devel.opments can be used to

assist him, not to enslave him.
Professor Whalan described briefly .four main ways of using the computer
in legal practice and analysis.

Firstly, the computer is an invaluable tool which

can rapidly and efficiently perform the more mundane functions of costing,
accounting, and general clerical and office management work.

However, the problem

of confidentiality arises as material will be processed outside the regulating
bounds of the law office by an independent person who is not bound by legal ethics.
Secondly, the computer provides convenient facilities for storage as a memory
bank that never forgets.

This power to retrieve information can be used to

produce legal documents composed of stored data, and, at a more advanced level, it
may be used to assist lawyers in analysing and comparing matter.

Thirdly,

statistical material can be processed with different variables used, and predictions
may be made to assess future trends.

England, Professor Whalan pointed out, was

using data collected on accident cases and awards in this way.

Fourthly, and

perhaps the most complex of all, court judgments and statutes could be stored,
indexed and analysed to determine prevailing judicial attitudes and plan future
legislation.

As this process depends very much on the particular interpretation

that the person doing the processing gives to certain phrases, the number of
combinations and resulting complexity are greatly increased.
Professor Whalan said that he was not being unrealistic in his desire to
use computer techniques to revolutionize law and to rationalize law reform in
Australasia.

His conclusions, in fact, were derived from successful methods that

were already operating in several U.S. jurisdicttons, and would soon be instituted
in England.

While he admitted that the great cost of the computer was an inhibiting

factor, he said that this was no suitable reason for postponing feasibility studies.
Professor Whalan then went on to discuss SOme of the areas in which problems
were likely to arise with the application of computer
law.

technology to most fields of

One major problem area was the quest;tcm 0f the valj:d;t.tY' of evidence.

As most

data are supplied to a computer through human agency, the aearsay rule is held to
apply unless all the data are sworn to in the Court by the particular individual

155

who initially supplied the data.
urgent legislation.

This problem, said Professor Whalan, requires

Professor Whalan also mentioned the difficulties in

probing and investigating evidence as he pointed out that identical output
can result from the same program and data run on similar computers.

With

interesting far-sightedness, Professor Whalan suggested the use of the
simulation processes to aid in the obtaining of evidence by setting up the
conditions of the situation.
Professor Whalan then discussed the issue of to~t or civil iiability,
an increasingly significant question as computers become generally accepted
in all kinds of enterprise.

In this context, Professor Whalan pointed out

that it could become negligence not to use the superior facilities of a
computer!

Will it, he asked, be negligence to disregard the use of simulation

models to test stresses and strains when building an aircraft or a motor car or
a bridge?

Will it be negligence for the doctor to ignore the availability of

computer diagnostic programs?

Problems are also raised as to precisely where

liability lies should a mistake occur.

These are only a few of the issues that

will be reality for Australian law in the very near future.
While Professor Whalan indicated that there was cause for concern with
numerous other fields such as insurance, copyright, patents, property and
labour relations, he felt that the issues of
in particular required serious thought.

p~ivaay

and individual

f~eedom

He warned that, if no legislative action

is taken, individual freedom will be blighted and our personal rights of privacy
destroyed by indiscriminate access to computerized information.

With an apt

misquotation of Sir Winston Churchill, Professor Whalan said that never in the
history of human endeavour has there been the opportunity for so much to be
made available to so many by so few.

He pointed out that not since the Domesday

Book of William the Conqueror was such a vast amount of personal information
available about the individual, ranging from his health record to his business
and financial assets.

This was accessible to both governmental and private

authorities, to be used and misused.

The computer simply collates quickly

and efficiently all the scattered records existing in various places about each
person.

With grim accuracy, it can add and delete information, recall selected

facts instantly, and, ominously, forget nothing.

Professor Whalan pointed out

that legislation must be enacted concerning data banks:

156

it should not be

piecemeal, but rather an overall legislative provision dealing with the problem
as a whole.
Professor Whalan also suggested that information supplied unknowingly
through the use of secret or bugging devices be dealt with severely.
information supplied knowingly had considerable complications.

However,

Should, for

example, information given willingly for one purpose be used for another purpose?
Should governmental and private agencies have access to the same memory dumps?
How can a person know whether the information supplied about him is accurate or
not?

If inaccurate, can the person sue for damages, and whom lor what} does he

sue?

Broadly, said Professor Whalan, rules must be established to restrict access

to records, be they accurate or inaccurate.

PrOfessor Whalan then put forth the

interesting concept of habeas notae ("that you may have the writings") to control
information banks.

Thus, if a person knows or suspects that a record is maintained

about him in a data bank or computer complex, he will be legally empowered to ask
for a copy of its contents.

Professor Whalan said that the idea of habeas notae

could be extended to include records other than those which are maintained by the
computer.
Towards the end of his lecture, Professor Whalan pointed out that today's
legislation explosion must be tamed and rechannelled with the aid of computer
technology.

There was a great need to stem this legislative and interpretative

explosion, the 39 miles of law books in the American Library of Congress being a
typical example.

Professor Whalan urged lawyers to take advantage of the computer

age, to use the computer to lessen burdens of technical routine, and make the
lawyer freer to think and create.

He did not envisage the computer replacing the

judge and jury. but said it was a useful tool to handle the ever-growing quantity
of legal material.

Law must be stable, he said, but it must not stand still.

By

harnessing computer facilities, the lawyer will be relieved of laborious detail;
he will be free to analyse, interpret and plan for the future.

157



Source Exif Data:
File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.6
Linearized                      : Yes
Has XFA                         : No
XMP Toolkit                     : Adobe XMP Core 4.2.1-c041 52.342996, 2008/05/07-21:37:19
Modify Date                     : 2017:08:06 06:04:06-07:00
Create Date                     : 2010:06:20 16:04:28+10:00
Metadata Date                   : 2017:08:06 06:04:06-07:00
Format                          : application/pdf
Document ID                     : uuid:6a4a2bf1-419f-4d47-8c86-934e430ffa48
Instance ID                     : uuid:aa4c5544-dafa-ec44-a33e-0ee8f2766a0b
Producer                        : Adobe Acrobat 9.0 Paper Capture Plug-in
Page Layout                     : SinglePage
Page Count                      : 16
EXIF Metadata provided by EXIF.tools

Navigation menu