IBM 709, 7090 General Information Manual

~~

General

Information

Manual

iii

709·7090

Data

Processing System

MINOR

REVISION

(August, 1960)

This

edition,

Form

D22-6508-2, is a

minor

revISIon of

the

pre-

ceding

edition

but

does

not

obsolete

Form

D22-6508-1.

The

principal

change

in

this

edition

is

the

addition

of

a section

on

1401

operation

on

page 31.

INTRODUCTION

Binary

Notation

Octal

Notation

Magnetic

Cores

CENTRAL

PROCESSING

UNIT

Stored

Program

. Assembly Programs . Computer

Operations

Information

Paths

Indexing

and

Indirect

Addressing

Sample

Problems

. Operator's

Consoles.

.

INPUT-OUTPUT

COMPONENTS

•

Magnetic

Tape

Storage.

Auxiliary

Equipment

.

Data

Synchronizer

IBM

755

Tape

Control. Multiplexor Data

Channel

. . External

Signal

Direct

Data

Feature

.

Magnetic

Drum

Storage

Punched

Cards

Card

Reader

Card

Punch

.. Printer Cathode

Ray

Tube

Equipment

SHARE

Organization Programming

System

FORTRAN

AUTOMATIC

CODING

SYSTEM.

APPENDIX

Contents

5

6

7

10

13

15

16

18

20 22

24 24

30

31

35 35

36

37 37 37

39 40

41 41

42

44

46

47

IBM

7090

Data

Processing System

IBM

709·7090

Data Processing System

Data

processing

sy~tems

are

finding new

application

in

virtually

every

phase

of

science, business,

and

in-

dustry.

Rapidly

expanding

scientific investigations in-

volve

many

complex

calculations.

The

vast

amount

of

data

constantly

being

used

in

aircraft

industries,

government

agencies,

and

business establishments

of

all kinds

demand

machines to

compute,

select,

and

correlate

data

at

electronic speeds.

Figure I shows

the

kind

of

problem

that

computers

are solving for scientists today.

Many

similar

prob-

lems

are

encountered

and

solved

in

the

scientific field;

however,

the

computer

is

adaptable

to business

appli-

cations as well.

The

computer

can

be,

and

is,

used

for payroll,

material

inventory,

or

any

of a great

num-

ber

of

other

business

problems

requiring

prompt,

accurate results.

In

the

sample

problem,

the

computer

calculates

the

density

and

velocity

of

air

at

each

of

the

mesh-points

shown.

From

these

data,

engineers

can

evaluate

the

lifting

power

and

drag

of

the

wing

section. By solv-

ing a number

of

these problems,

an

optimum

new

wing design is developed.

The

mathematical

formulas used are:

4.

p=

f (p)

Figure 1. Aircraft

Application

Equations

1, 2 and 3 are

replaced

by difference

equations

over a network

as

shown

in

the

figure.

This

difference system

and

equation 4 form

a set

of

simul-

taneous

non-linear

algebraic equations.

These

are

solved

on

the

computer

by

repeating

calculations,

called

an

iteration

method.

In

this

one

typical prob-

lem,

there

are

800 equations,

100

iterations, 80,000

operations

per

iteration,

and

8,000,000

operations

per

solution.

Figure 2 shows a comparison,

in

time, between

man-

ual

methods

of

solution

and

solution

by computers.

The

times

are

approximate.

This

comparison

shows

that

computers

of

today now have

the

ability

to solve

problems

that

cannot

be solved

in

a lifetime

of

man-

ual

labor.

One

of

the

first things to be considered

in

a com-

puter

system is

an

understanding

of

the

functions

of

its various components. A simplified

computer

sys-

tem consists

of

three

main

units

as

shown

in

Figure

3.

In

a typical

computer

system, several types

of

input-

output

devices

are

needed.

These

may be

units

such

~

r\

\ Pencil and paper {

15

years

Desk

calculator

80 weeks

IBM

701

2 minutes

IBM

704

30 seconds

IBM

709

25 seconds

IBM

7090

5 seconds

Figure

2.

Time

Comparison

of

Computation

Methods (Approxi-

mate)

Introduction

5

Figure

3.

Main

Computer

as

punched magnetic punched

An

ing

information

unit,

A storage

"remember"

card

or

paper

card

recorders,

input

component

usually called

unit

all called orders) to then

used

to tell

data

is

to be executed.

The

central

tion

to accomplish all

tions

and

adapt

to

processing

a section to modify

the

processing tablished procedure. A of

the

central

puter

in example determine result

of this test,

processing

its

operation

of

one

if a number

Components

readers, a device to

tape, a

telephone

and

is

any

printing

device

into a computing

the

must

input

the

computer.

the

computer

central

also be

data

unit

processing

provided

and

These how

has

arithmetic

or

to

variations

control

by

making

of

these decisions

section is also a

unit

and

logical decisions.

is positive

the

computer

can

read

or

write

on

or

teletype line,

equipment.

capable

and

data

of

feed-

handling

unit.

to store

or

instructions (also

instructions are

the

processing

an

arithmetic

and

logical opera-

of

sec-

change instructions

within

the

es-

part

directs

the

com-

An

would

be a test to

or

negative. As a

take

a different processing course for each alternative. Simple tests may be

The

output

recording

ing

unit

combined

the

has

to

components

resultant

operated

make

data

upon

complex

are

any devices

after

it.

logical decisions.

capable

the

central

process-

of

The

decimal system, by most serves well for

people

early

counting computers designed neers

and

businessmen, be designed to use

system

binary puters shorthand Octal

but

bols 0

starts

with

ever, there

of

numbers?

The

reason circuits; therefore, is

binary

method

notation

has

no

relation

The

binary,

and

1 to

in

the

same

a 0

for

zero

are

is

is used

or

no therefore necessary to take binary tem. to position. A a 2 an

order

time a is

Octal

the

system

This

left

that step is to place a I and

start

binary

in

the

decimal system.

analogous

every

as

follows:

Binar)

8 4 2 1

0

1

0 100 1 1 0

0

1

o 0 - 8 1 0 o 1 - 9

ten

manner

time

is

reached.

Decimal

=

0 1

2 1 1 - 3

-

4

- 6

Notation

with

its

in

their

purposes.

to

assist

that

current

the

in

nature.

of

writing when

to

the

The

discussing

internal

"base two" system, uses

represent

manner

and

more

is

taken

10

all

as

then

a I for one.

symbols to be used.

the

same step

at

again

with

is

equivalent

Counting

with

a carry to

a two is reached

Counting

Binary

84

1 0 1

11

ten

digits, is

training.

Why

This then

mathematicians,

the

digital

mathematics

computers

of

octal system

long

binary

numbers.

the

computer,

computer

the

two sym-

quantities.

in

the

ten

in

the

a 0

in

instead

in

the

2 1

=

-

1

=

Counting

decimal system

At

two, how-

at

two

the

decimal

in

the

this respect to

is

continued

the

binary

Decimal

1

5

7

learned

system

should

engi-

binary

use

com-

is

circuits.

It

is

in

the

sys-

original

in

higher

every

system

a

Binary

The scientific familiar it

is possible

ing

What

of convenience. Decimal

is

However, fingers familiar

rather

the

6

Notation

common

systems

most

than

decimal system.

IBM

decimal

world

that

its use is that,

are

numbering

familiar

had

our

instead

with a numbering

ten,

709-7090

is a

for

more

and

primeval

of

ten, we

and

notation

familiar

of

one.

hardly

some purposes,

convenient.

system

to

use

notation

is

understood

ancestors developed

would

system based

we

might

the

This

questioned. However,

is

probably

consequently

commercial

notation

other

entirely a

used

and

is so

number-

matter

because

by most people.

eight

be

more

on

eight,

question

It

has

already require mal

is

and string mitted hand

about

numbers

not a problem

writing, these

of

ones

from

method

fills this need. Because

it

binary,

another

is

4, 5,

numbers

by inspection.

8.

This

6,

and

means

7. relationship lent

to

one

bines

what

binary

with

been

pointed

three

times as

to express

to

the

computer

the

binary

and

zeros

one

individual

is necessary.

of

can

be

converted from

The

there

are

There

is

that

are

three

octal position.

has

been

shown

the

octal

numbers.

out

many

equivalent

numbers cannot

to The its simple

base

eight

no

8's

binary

The

concerning

that

binary

numbers

positions as deci-

number.

itself,

but

in

talking

are bulky. A

be effectively trans-

another.

octal

Some short-

number

system

relationship

one

system

of

the

octal system

symbols:

or

9's. positions following

The

0,

1,

important

are

equiva-

table

decimal

This

long

to to

2,

3,

com-

and

BINARY

OCTAL

DECIMAL

0

0 0

1 1 1

10

2

11

3

100

4 4

101

5 5

110

6 6

III

7 7

At

this

point

a carry to

the

position

of

the

number

is necessary, since all

eight

symbols

in

the

octal system have

been

used.

1000 1001

1010 1011

10

II

12 13

8 9

10

11

As

far

as

the

internal

circuitry

of

the

computer

is

con-

cerned,

it

only

understands

binary

ones

and

zeros.

The

octal system

is

used

to

provide a shorthand

method

of

reading

and

writing

binary

numbers

so

that

the

contents

of

a register,

when

shown

on

the

operator's

panel,

may

be

read

directly.

This

is

shown

in

Figure 4,

using

a 36-digit

binary

number.

Reg;

ster contents (b; nary)

Octal

value

Figure

4.

Binary

Representation

of Register

Contents

Magnetic

Cores

The

main

storage

medium

for

many

computers

is

the

magnetic

core.

Each

magnetic core is a

ring

or

doughnut

shaped

piece

of

ferromagnetic

material.

The

cores "remem-

ber"

information

indefinitely,

and

can

recall

it

in

a

few

millionths

of

a second.

When

a wire

is

inserted

through

the

hollow

center

of

a core (Figure

5),

an

electrical

current

passed

along

the

wire sets

up

a mag-

netic

field

around

the

wire.

This

field magnetizes

the

core.

When

the

current

is removed,

the

core

remains

magnetized.

1£

a

current

is

passed

along

the

wire

in

Core

Is

Set

Figure

5. Magnetic Core Action

the

opposite

direction,

the

magnetic

field set

up

around

the wire is reversed.

When

this occurs,

the

core

is

said to have

"flipped"

or

changed

its mag-

netic state. A

"sense" wire

is

inserted

through

the

core

and,

when

the

core flips, a small electrical voltage

is sent along

the

sense wire.

This

voltage

may

then

be amplified

and

used

in

the

computer

(Figure

6).

Computer

Figure 6.

Flipping

of a Magnetic Core

As

with

the

magnetic core, all

computer

elements

are

able

to

represent

two states.

These

two

distinct

states

provide

the basis by

which

these elements

hold

information.

For

this reason,

the

elements

are

called

bi-stable elements.

For

example,

one

state

may

be

interpreted

as

the

digit

0;

the

other,

as

1.

Similarly,

the

elements may be used to represent:

plus

or

minus,

on

or

off,

yes

or

no,

and

so on. Several of these ele-

ments are shown

in

Figure

7.

The

IBM

709

and

7090

Data

Processing Systems use

high-speed core storage units.

Each

unit

is

divided

into

distinct sections called locations.

Each

location

is

uniquely

identified by a

number

assigned to it.

This

identifying

number

is

called

an

address because

just

as

a street address denotes

the

precise

location

of

a

particular

building

on

that

street, this

number

de-

liD

..

State

Figure

7.

Bi·stable

Computer

Elements

Introduction

7

Number 1500

Figure

8.

Core Storage

Unit

notes the location of a

particular

item

of

data

inside

the core storage

unit

called a

"word"

(Figure

8).

Each word is

further

subdivided

into

elements

called

bits. Each

bit

has a value

of

either

a 0

or a 1.

Thus.

the bits

(36

in

each word)

are

the basic units

of

information

in

the computer. Figure 8 also shows

S6

"planes"

or

"stories"

in

the core storage "building."

The

operation

of locating a word

in

core storage may

be compared to the

operation

of

an

elevator

in

an

office

building.

The

elevator picks

up

passengers from

each floor;

in

core storage,

one

bit

of

information is

read

or

stored from each plane.

In

actual computer

operation, all

36

bits are read, stored,

or

operated

upon

simultaneously, one from each plane.

As

previously stated, one word contains

36

bits

of

information. Core storage contains two types of

words:

1.

A word

upon

which arithmetic

or

logical oper-

ations are to be performed

is

called a data-word.

2.

A word

interpreted

by the

computer

as a code

to

"order

or

instruct"

it

to perform a

particular

operation

is

called an instruction-word.

8

IBM

709-7090

Figure 9. Instructions

and

Data

in

Core Storage

Both types of words are combinations of

36

zeros

and

ones.

An

instruction is able to "direct"

the

computer

to perform some type

of

operation, e.g., read. write,

add, subtract,

or

test for zero.

If

a

data

word were

incorrectly used

as

an

instruction, the computer

might

perform

an

illegal

operation

depending

upon

the

data

bit

configuration.

Data

words form records. fields,

amounts, results,

and

so

on. Figure 9 is a schematic

of core storage

with

both

types

of

words contained

in

separate locations.

Core storage

units

are available to provide the com-

puters

with

a capacity

of

4096, 8192,

or

32,768 word

locations of

36

information

bits each. As a decimal

digit is expressed by three

of

these bits, a total of

II

significant decimal digits may be expressed by each

word.

In

BCD coding (alphamerical characters),

one

word contains six numerical

or

alphabetic characters.

Thus,

the 32,768 words of core storage may contain

the binary equivalent to

360,448 decimal digits of stor-

age

or

the BCD representation of 196,608 characters.

The

actual capacity of storage is thus directly related

to the type of coding used.

Figure

10

shows the three core storage units

that

are available for the 709

and

7090

Data

Processing

Systems.

The

IBM

737

Core Storage, used

with

the 709

system, has 4,096 words

of

storage.

The

IBM

738

and

the

IBM

7302 Core Storage have 32,768 words

of

stor-

age

and

are used

with

the

709

and

the 7090 systems,

respectively.

Figure

10.

IBM Core Storage

Units

Introduction

9

Central Processing

Unit

The

central processing

unit

can

be divided

into

two

basic parts,

information

processing elements

and

exe-

cution

controls.

The

information

processing elements

are normally referred to

as

the

arithmetic

section.

The

execution controls are called simply

the

control

section.

The

arithmetic

section

is

the

computer's problem-

solving

unit.

The

operations

of

addition, subtraction,

multiplication,

and

division, as well as shifting, trans-

ferring,

and

storing

of

results are executed

in

this

section. Figure

11

shows a few of

the

operations per-

formed

in

the

arithmetic

unit.

The

control section guides the

computer

through

the

operations necessary to complete

the

various in-

structions.

The

execution

of a group

of

instructions may be

compared to

compounding

a chemical formula.

The

arithmetic

section would be

the

laboratory where the

compounding

takes place;

it

would

call

on

core stor-

age for materials needed to

produce

the

result.

The

control section would follow directions given

in

the

formula.

It

would instruct

the

chemist as to

what

ingredients to mix, how

long

to

mix

them,

and

in

what

order

they should be combined.

Of

necessity, then, the control section contains a

device called

the

"instruction

sequencer."

This

de-

vice locates

the

proper

instruction to be executed;

then, while

the

instruction

is

being executed, this de-

vice sets

up

the

conditions for

the

At

different times

in

the program,

the

to

be

executed may

depend

on

the result of a test in-

struction, for example,

whether

an

error

indicator

is

on

or

off (Figure

12)

.

The

arithmetic

operations which computers are

capable

of

performing

treat

each

part

of

a word

in

014623

ADD

541034

555657

12345

Sub.

11111 01234

':"_

,,'_"'~"~

Unit

Control

Un

i t

Figure

II.

Arithmetic

Operations

10

IBM

709·7090

12345

SHIFT~

~

STORE

3416

Arithmetic Section

"test the indicator!!

error

indicator«.t-.

"if

the indicator

is

on obtain the next instruction-from location

1000.

If

the indicator

is

off obtain the next instruction

from

location

2000."

"the

indicator

is

on,

put 1000 into the instruction sequencer"

"get

the instruction located

at

address 1000 and

execute

it."

- - - -

COntral

se~iia;;

- - - - - - - - - - - - - - - - -

--

11000!;nstructian sequencer

Figure

12.

Conditional

Testing

and

Transferring

core storage as

either

a 1

or

a

O.

Thus, a computer

word

consists

of

36 zero

and

one

digits

and

is said to

be

expressed

in

the

"binary

system."

An

example

of

a

binary

word

as

it

appears

in

the

arithmetic

unit

is

001101110010011011001001111001001001 which is sim-

ply

a string

of

zeros

and

ones. By

doing

its arith-

metic

in

the

binary system,

the

computer

is able to

achieve

greater

speeds

and

efficiencies

than

would

otherwise be possible.

Normally,

in

solving a problem,

the

numerical

data

are

originally

entered

into

the

computer

in

decimal

or

IBM

card

coding.

The

computer

is

then

instructed

to translate

the

data

into

binary form, go

through

the

desired computations,

and

then

translate

the

results

back

into

decimal form (Figure 13). However,

data

may be

entered

into

the

computer

in

any form desired

by the programmer.

Instruction

0011010101111

ADD

11010001111,00

10000011101011

PRINT

Central

P(ocessing

Unit

Figure

13. Decimal

and

Binary

Data

Flow

Stored

Program

The

work accomplished by the

computer

in

solving

a

problem

or

processing

data

consists

of

executing

many

instructions

at

high

speed.

To

solve a problem,

it

must

first

be

reduced to equations

or

instructions

that

the

computer

is capable

of

performing.

The

en-

tire set

of

instructions used

in

solving a

problem

form

a program for

the

computer. Because these instruc-

tions are

held

in

the computer's storage

unit,

it

is

called a stored program system.

Normally, instructions

are

taken

from

sequentially

ascending locations

of

core storage. However,

the

exe-

cution

of

instructions does

not

necessarily have

to

occur

sequentially.

It

is possible,

when

control

or

transfer

instructions

are

given, for

the

computer

to

alter

the

process

of

sequential

execution

and

to

indi-

cate

some

particular

location

in

core storage contain-

ing

the

word

to

be

executed.

In

this

way,

the

execution

of

any

instruction

or

block

of

in-

structions

may

be

repeated

as

often

as desired.

For

some

control

instructions,

whether

the

struction

is

taken

in

sequence

or

from

some

other

specified location

may

depend

on

the

result

of

a test

(Figure

12).

In

this case,

the

control

operations

pro-

vide

the

program

with

decision-making abilities.

The

logical

path

followed by

the

program

(that

is,

the

precise sequence

of

instructions to

be

executed) may

be

controlled

by a series

of

tests

applied

at

various

points

in

the

program.

Doing

this gives a

stored

pro-

gram

the

ability

to

change

its course of execution.

These

conditional

operations

increase

immeasurably

the

scope

of

the

system's

application.

Most

computer

instructions

have

an

address

part

indicating

the

location

in

core storage

subjected

to

some

arithmetic

or

logical

operation.

This

address

part,

or

field, always occupies

bit

positions

21

through

35 (Figure 14)

in

an

instruction

word.

Address

part

I

s,

I

Figure

14. Address

Part

of

an

Instruction

The

15-bit address field is large

enough

to

hold

the

number

32,767 -

the

largest address

in

core storage.

This

number

expressed

in

the

binary

system is simply

15

consecutive ones (Figure 15) .

II

III

I I

11111

S,1

2021

35

Figure

15. Address

Part

Showing

Maximum

Address

In a data

processing system

with

8,192 words of

core storage,

the

largest address (8191)

is

contained

in

only

13

bit

positions

of

the

instruction

word.

The

contents

of

the

two left-most positions

of

the

address

field

are

ignored

during

the

execution

of

an

instruc-

tion

to

decode

an

address. Similarly, a

computer

with

4,096 words

of

core storage uses

only

bit

positions 24

through

35

as

its address field.

The

operation

part

of

an

instruction

normally

is

not

fixed

in

length

but

may

vary

depending

upon

the

instruction

itself.

Figure

16

shows

the

bit

pattern

for

an

instruction

specifying

the

addition

of

the

contents

S,1

35

Figure

16.

Add

Instruction

of

core storage

location

0001.

When

this

instruction

is

executed

by

the

computer,

the

contents

of

core loca-

tion

0001

are

added

to

the

contents

of

the

accumulator

register.

For

example, assume

that

the

accumulator

contains

the

number

+1.

If

the

number

in

location

0001

is

+2,

the

result

of

executing

the

ADD

instruction

(Fig-

ure

17) is a 1

bit

in

positions 34

and

35,

and

0 bits

in

all

other

positions

of

the

accumulator;

(II

is

the

binary

expression for

the

sum,

3).

The

first

position

of

the

word

is

used

to express

the

sign

of

the

amount

or

result.

Instruction

(Add)

location 1

(+2)

000000000000000000000000000000000010

, ,

AC

before

(+1)

a a a

000

a a a a 0 a

000

a a 0 a 0 0 a 0 0 0 a 0 0 a

00

a 0 a

01

,

AC

after

(+3)

1010000

a 0 0 0 0 0 a a 0 0 0 a a 0 a a a 0 a 0 0 a

00000

a a 1

Ii

s 1

J5

Figure

17.

Execution

of

an

Add

Instruction

A register is defined

as

a device

with

the

ability

to

accept

data,

hold

them,

and

transfer

the

data

to

an-

other

register

or

device.

The

registers

are

given dif-

ferent

names according to

their

functions.

Thus,

the

accumulator

register

"accumulates"

arithmetic

results.

The

multiplier-quotient

register

holds

either

the

mul-

tiplier

or

the

quotient

in

an

arithmetic

operation.

On

all

arithmetic

operations,

the

sign positions

of

the

registers

in

use

are

automatically

adjusted.

The

accumulator

register has 38 positions.

They

allow

the

register to

handle

a 35-bit

word

with

its

sign.

Two

extra

positions, P

and

Q,

are

provided

to

keep

track

of

overflow conditions.

If

two 35-bit

num-

bers

are

added

together,

it

is possible

that

the

result

would

be

larger

than

35 bits, as

shown

in

Figure

18.

The

accumulator

is

used to

hold

one

factor

during

arithmetic

operations.

The

other

factor usually is

Accumulator

register

Storage register

Combined in

the

adder

and

returned

to

the

accumulator

register.

S Q P I

35

! ! ! 1 000000000000000000000000000000000 1

1100000000000000000000000000000000111

Q P 1

35

I !

Ii

000000000000000000000000000000001001

!

iIi

000000000000000000000000000000001001

S Q P 1

35

}1igure 18.

Accumulator

Overflow

Condition

Central

Processing

Unit

11

Accumulator register before a

left

shift.

5 Q P 1 35

I ! ! :100000000000000000000000000000000001

5 Q P

1-----------35

Accumulator register I I I i I

after

a left shift

of

I

11

100000000000000000000000000000000000

one

place.

Figure

19.

Accumulator

Shifting

held

in

the

storage register.

The

two factors

are

com-

bined

in

another

register-type device called

the

adder.

The

adder

consists

of

35-bit positions

and

the

two

extra

ones, P

and

Q.

Thus,

any overflow

out

of

posi-

tion

1 will

be

placed

in

position

P. Likewise, any

overflow

out

of

position P will

be

placed

in

position

Q.

An

overflow

out

of

position Q will, however, be

lost.

Information

in

the

accumulator

may be shifted to

the

right

or

to the left.

1£

a left shift is involved,

an

overflow

condition

may

be

recorded

in

the

same man-

ner

as

in

an

arithmetic

operation

(Figure

19).

Not

only

is

the

entire

number

shifted

to

the

left,

but

zeros

are

inserted (position 35)

in

the

positions vacated.

Thus,

no

matter

what

the

content

of

the

accumulator,

if a left

shift

of

37

or

more

places is executed,

the

contents

of

the

accumulator

are

replaced by zeros.

Information

may also

be

shifted

into

the

accumu-

lator

from

the

MQ

register,

one

bit

at

a time,

or

from

the

accumulator

to the

MQ,

one

bit

at

a time.

The

effect is to create a register

that

is

74 positions

in

length,

38

positions for the

accumulator

and

36 for

the

MQ

register.

The

multiplier-quotient

(MQ)

register has

36

posi-

tions.

During a multiply

operation,

this register con-

tains

the

multiplier;

during

a divide,

it

receives

the

quotient.

The

register

can

store

and

shift a full

36-

bit

word.

In

addition

to shifting

right

or

left,

in

the

same

manner

as

the

accumulator,

it

can

also

"ring

shift."

That

is, bits shifted

out

of

the

sign position

enter

position

35

(Figure

20)

.

In

addition

to its

arithmetic

and

shifting functions,

the

MQ

register serves

as

a sending

and

receiving regis-

ter

for some

input-output

operations.

For

the

709

system, this means

that

information

coming from

the

magnetic

drum

or

going to

the

drum

and

CRT

equip-

ment

passes

through

the

MQ

register.

When a word

is

entered

into

or

taken

from a loca-

tion

in

core storage, a storage reference is said to

be

51

35

MQ

register before 111111000000000000111000111000111

000

a

ring-shift

of

three

places.

51

35

MQ

register

after

shift 11110000000000001110001110001110001111

Figure

20.

MQ

Register

"Ring

Shift"

12

IBM

709-7090

made.

This

operation

is non-destructive.

That

is,

the

contents

of

that

location

are

left

unaltered

after

the

operation.

However,

if

the

storage reference causes

new

information

to

be

entered

into

a location,

the

prior

contents

of

that

location

are

automatically re-

placed

by

the

new

information.

Only

one

such storage reference

can

be

made

during

any single

computer

cycle.

The

basic

computer

cycle

(or

internal

speed) for

the

709 is 12 microseconds

long

(12

millionths

of a second).

The

basic cycle for

the

7090 is 2.18 microseconds.

Normal

computer

operations

start

with

an

instruc-

tion

(I) cycle.

This

cycle does

the

following:

1.

It

obtains

the

instruction

to

be

executed from

the

core location designated

by

the

instruction

counter.

2.

It

locates

the

operand

(number

to

be

worked

with

or

upon),

if

any, as specified

by

the in-

struction's address

part.

An

example

would

be

the

instruction

ADD 0002 (add

the

contents

of

storage location 0002 to

the

contents

of

the

accumulator

register), shown

in

Figure

21. All

data

handling

concerned

with

storage

and

the

central

processing

unit

is accomplished

in

parallel.

This

means

that

all operations

are

concerned

with

a full

word

of

36 bits.

As

an

example, refer to Figure

17

where

an

ADD

operation

is performed.

Even

though

the

numbers

involved use

only

a few bits

of

the

entire

word,

the

remainder

of

the

word

is filled

with

zeros

and

all 36

positions

of

both

words

are

then

added

in

one

opera-

tion.

Thus

the

time

required

to

add

two

I-bit

num-

bers is

the

same

as

the

time

required

to

add

two

36-

bit

numbers.

This

principle

of

parallel

operation

is

one

of

the

basic differences between

the

709

and

7090 computers

when

compared

with

the

705 system series, which is

said

to

operate

serially (normally om;

position

or

character

at

a time) .

(ADD 0002) =

~

Instruction Register

5torage Register

~

Address Register

Adders

1-+004121503051

I (number in accumulator)

Accumulator

Register

Figure

21. Schematic, Decoding

an

Add

Instruction

The

instruction

ADD 0002

is

brought

from

core

storage

into

the

storage register.

At

this

point,

the

operation

part

is sent to

the

instruction register

to

determine

the

type

of

operation

to be executed.

At

the

same time,

the

address

portion

of

the

storage reg-

ister is

taken

to

the

address register so

that

the

proper

core storage

location

(operand) is

added

during

the

All

numbers

in

Figure

21

are

shown

in

octal

notation.

The

referred

to as

an

execution

cycle,

is used

when

a reference to core storage is

required

to

obtain

another

word, e.g., a

number

to be added.

Assume

that

the

address register has

the

address

of

the

core storage location to be added.

During

this second

(reference) cycle, the contents

of

that

storage location

are

brought

from core storage to

the

storage register,

and

then

to

the

adders.

At

the

same time,

the

pre-

vious

number

in

the

accumulator register

is

taken

to

the

adders. A

new

number

is

formed

in

the

adders

and

is

then

placed

back

in

the

accumulator

register

(Figure

22).

The

operation

is

now

finished

and

the

computer

is

ready

to execute

the

instruction.

The

type

of

operation

just

described is

called

fixed

point

arithmetic

or

integer

arithmetic.

The

computer

can

also

perform

operations

in

floating

point

arithme-

tic. A

complete

set

of

floating

point

instructions

is

provided

to

increase

the

range

of

numbers

used

and

to

reduce

programming

time.

In

integer

arithmetic,

the

size

of

the

numbers

used

is fixed by

the

design

of

the

computer

(36-bit

word

size) . By

using

floating

point

arithmetic, a larger

num-

ber

may

be

expressed

and

operated

upon,

because a

part

of

the

word

is used to express

the

exponent

(char-

acteristic)

and

another

part

of

the

word

is

used

for

the

fraction

(mantissa).

A

comparison

of

number

size shows

that

the

largest

number

the

computer

can

use

with

fixed

point

opera-

tion

is

I X

1011.

This

is

equal

to 100 billion.

With

floating

point

operation,

however, 3 X

10

35

is

the

larg-

Instruction Register

+132540020425 (number

from

location 0002)

+1

32540020425

Accumulator

Register

Storage Register

Address

Reg

i ster

(number

from

location 0002)

(previous

number

in accumulator)

(resulting number

in

accumulator)

Figure

22. Schematic,

Performing

an

Addition

I

Characteristic

Fraction

51

8 9

35

Figure

23.

Floating-Point

Word

Format

est

number.

This

number

is too

high

to be expressed

in

common

language

terms.

In

the

computer,

a floating-point

number

is stored

in a word

as shown

in

Figure 23.

The

fraction

is

con-

tained

in

bit

positions 9

through

35

with

the

sign of

the

fraction

contained

in

position S of

the

word.

The

characteristic

is

contained

in

bit

positions 1

through 8 and

is

formed

by

adding

+128

to

the

ex-

ponent.

For

example,

an

exponent

of

-32

would

be

represented

by a characteristic

of

128-32

or

96.

An

exponent

of

+ 1

00

would

be

represented

by a charac-

teristic

of

100+128

or

228.

Most

integer

arithmetic

instructions

have

a floating-

point

counterpart

instruction.

Thus,

it

is

possible to

ADD

or

to

perform

a floating-point ADD

(FAD).

To

summarize, a floating-point

binary

number

(X)

may be

represented

as

a signed

proper

fraction (B)

times some integral power (b) of

2.

Examples:

X

B

2

b

-.001

=-.100

X

2-"

.100

x

2"

l.l00

.110

X

2'

110.000

.110

X

2

3

Assembly

Programs

The

writing

of a complete

program

for

the

computer

in

its

machine

language

would

be

rather

awkward.

For

example, to

write

only

the

one

instruction

re-

quired

to

subtract

the

contents

of

core

location

0003

from

the

contents

of

the

accumulator

register

requires

the

recording

of

36 zeros

and

ones,

as

shown

in

Figure

24.

This

instruction

can

be

written

as

SUB

0003, a

more

convenient

form

of

expressing

the

instruction.

The

writ-

ing

of

instructions

in

this form is

an

improvement

over

the

system

of

writing

36

zeros

and

ones.

In

addi-

tion

to

reducing

the

amount

of

information

which

must

be

written,

it

has.

more

meaning

to

the

reader.

Namely,

"subtract

the

contents

of

location

0003."

However, to

determine

the

entire

machine

operation,

the

programmer

or

person

reading

the

program

must

Operation

Address

Figure

24.

Subtract

Instruction

Central

Processing

Unit

13

know

what

is stored

in

location 0003.

To

assist

the

programmer,

the

use

of

symbols

can

be

extended

to

in-

clude addresses as well as

the

operation

part

of

an

in-

struction.

Assuming

that

the

net

pay

for a payroll calculation

is

stored

in

location 0003,

the

instruction

might

now

be

written

as

"subtract

net."

This

clarifies to the pro-

grammer

the

operation

to be performed,

and

also

what

quantity

is involved.

Thus,

symbolic

programming

employs instruction-by-instruction coding

in

a lan-

guage

that

is a representation

of

the

basic language

of

the

computer

itself.

The

computer

can

execute instructions only

if

,they

are stored

in

machine language.

Thus,

the

instruction

"subtract

net"

must be converted to

the

basic

machine

language before its execution.

One

way

of

doing this

would be to

manually

convert all symbolic instruc-

tions to machine language before

entering

them

into

the

computer.

This

obviously would be a laborious

task. A more practical solution

is

to have

the

com-

puter

perform

the

conversion.

This

is accomplished

through

use

of

a symbolic assembly program.

An

assembly

program

performs

the

necessary trans-

lation

or

conversion from

the

symbolic language to

the machine language.

At

the

same time,

the

program

assigns absolute core storage locations to

the

imtruc-

Convert

the

Instructi on

"SUBTRACT NET"

Becomes

000100000010000000000000100000000

No

Yes

Print

and

punch

the

resulting

program

Figure 25. Flow

Chart

of

an

Assembly

Program

14

IBM

709-7090

tions

and

data

contained

in

the

symbolic program.

Figure

25

shows a flow

chart

of

the

process involved

when using

an

assembly program. Assume

that

the

instructions to

be

assembled are

on

punched

cards

and

that

the

assembly

program

itself is

on

a magnetic tape

(assembly tape) .

The

assembly

program

normally provides certain re-

strictions

upon

the

length

and

type

of

symbols

that

may

be used by

the

programmer.

It

assigns

an

abso-

lute

core storage location for

the

first symbolic instruc-

tion

and

increases this location by

one

for each sub-

sequent

instruction decoded.

In

this manner, the de-

coded

program

instructions are assigned sequential lo-

cations. Space

is

provided for

the

data

that

are a part

of

the

assembled program. Normally,

the

constants

and

other

data

are placed

at

the

end

of

the

symbolic

program

and

are assigned sequential locations

that

follow

the

instruction area.

During

assembly,

the

first reference to a symbolic

address assigns this address to

an

absolute storage lo-

cation. Any subsequent references to this symbolic address will also be assigned

the

same address.

The

man-

ner

in

which

the

assembly proceeds is described below.

Each

symbolic

instruction

is

read

from

the

card

reader

into

the central processing

unit.

The

"sub-

tract"

portion

of

the

first instruction is

matched

against

the

contents

of

the

assembly tape

and

is de-

coded

into

the

proper

bit

configuration for a

subtract

operation

000100000010.

The

"net"

portion

of

the

instruction

is assigned a location

in

core storage

that

is

not

being

used.

This

bit

configuration (assume

000 100000000)

is

then

inserted

into

the

address

portion

of

the

decoded

subtract

instruction.

The

full instruc-

tion

word

(000100000010000000000000000 I 00000000),

both

operation

and

address parts,

is

then

written

as

the

first instruction

on

the

program

tape.

The

original

"subtract

net"

is stored

so

that

it

may

appear

in

the

final

printing

of

the

program.

The

assembly

program

then

tests to

determine

if

more instructions

remain

to

be

decoded.

If

there

are more,

the

above

process

is

repeated

until

all symbolic instructions have

been

decoded

and

written

on

the

program

tape.

When

the

assembly process is complete,

the

resultant

as-

sembled

program

is

printed,

together

with

the

original

symbolic instructions.

The

print-out

takes

the

follow-

ing

form:

SYMBOLIC

INSTRUCTION

DECODED

MACHINE

EQUIVALENT

SUBTRACT

NET

000100000010000000000000100000000

When

the

printing

process is complete,

the

assembly

program

furnishes the

programmer

with a group

of

punched

cards

containing

the

machine

language

trans-

lation

so

that

if

the

program

is used again,

the

as-

sembly process

need

not

be

repeated.

Computer

Operations

The

format

of

the

instruction

word

is, for

the

most

part,

a precise one.

Although

slight

variations

exist,

in

general

the

format

is

as

shown

in

Figure

26.

Operation

part

Flag

Tag

Address part

I I I I

5,

1

11

12-13 18-20

21

35

Figure 26.

Instruction

I'onnat

The

operation

part

usually

is

contained

in

positions

S,

1-11

of

the

instruction

word.

In

the

following text

and

in

program

writing,

an

alphabetic

code

is

used

to

identify

the

instruction

operation

rather

than

the

full

name

of

the

instruction.

Thus,

the

code

CLA

signifies

CLear

and

Add,

or

SUB

means SUBtract.

If

the

numeri-

cal

machine

language

code is to be used,

it

is given

in

the

octal system.

For

example,

+0500

is

the

code

for

the

clear

and add

operation;

+0402

denotes subtrac-

tion.

All

abbreviations are simply

shorthand

methods

used to

reduce

the

manual

task

of

writing a program.

The

flag

part,

or

flag bits, is

contained

in

positions

12

and

13

of

the

word.

It

specifies

that

indirect

ad-

dressing is

to

take place.

This

operation

may

be

per-

formed

only

on

instructions

that

use

index

registers.

Both

indirect

addressing

and

the

use

of

index

registers

are

explained

later

in

the

text.

The

tag

bits

are

contained

in

positions 18, 19,

and

20.

They

specify

which

index

registers

are

to

be used.

Only

a few

instructions

may

not

use

indexing.

The

address

part

of

the

instruction

is

contained

in

positions

21

through

35

of

the

word.

It

tells

the

com-

puter

the

location

or

"operand"

that

is to be

used

with

the

instruction.

In

the

case

of

shifting

operations,

the

address

part

contains

the

number

of

places to

be

shifted.

For

other

instructions,

the

address

part

may

be a

part

of

the

operation

itself. Such a case

would

have

the

address

part

expressed as a four-digit num-

ber; read-1200 means

the

reading

operation

will

take

place

and

the

1200 will tell

which

input-output

unit

is

being

used.

In

this case,

the

tape

unit

numbered

0

on

data

channel A would

be used.

In

executing

the

instruction

ADD

1000,

the

computer

assumes

that

the

augend

is

in

the

accumulator

and

the

addend

(the

number

being

added) is specified

by

the

address

part

of

the

ADD

instruction.

The

sign

of

the

numbers

(S

position) is,

of

course, considered

dur-

ing

an

add

operation.

When

two

numbers

of

the

same

magnitude

are

being

added,

the

sign

of

the

result

is

taken

from

the

number

in

the

accumulator.

ACCUMULATOR

+6

-6

+

STORAGE

-6

+6

=

RESULT

IN

ACCUMULATOR

+0

-0

The

clear-and-add

(CLA)

instruction

is

similar

to

the

add

instruction

except

that

the

accumulator

is

cleared to zeros

and

the

contents

of

the

location

speci-

fied by

the

address

part

of

the

CLA

instruction

are

placed

in

the

accumulator.

Add

magnitude

(ADM)

is

another

similar

instruc-

tion.

Its

operation

is

the

same

as

ADD

except

that

the

sign

of

the

number

is

ignored

and

the

number

is

treated

as positive.

Other

arithmetic

instructions are

treated

in

a like

manner

since all

arithmetic

processes

are

accomplished

by

addition,

and

complementing a result

where

neces-

sary.

Subtraction

occurs

in

the

following

manner:

7 = 000111

Number

in

Accumulator

5 = 000101

Number

to

be

Subtracted

(Operand)

111000

Complement

of

Number

in

Accumulator

+ 000101

111101

Recomplement

the

Result

(No

High-Order

Carry)

000010

This

is

the

answer, 5

subtracted

from 7 = 2.

Multiplication

is accomplished by testing

the

low-

order

position

of

the

multiplier

and

adding

the

multi-

plicand

if this low-order position is a

1.

After

each

test,

the

answer is shifted

one

place

to

the

left.

This

is

repeated

until

there

are

no

more

numbers

in

the

mul-

tiplier.

5 = 000101

Multiplicand

X3=

000011

Multiplier

000101

000000

15

= 00001111

Product

Division is accomplished

in

a like

manner

but

shift-

ing

occurs

in

the

opposite

direction. By

shifting

a

binary

number

one

place

to

the

left,

the

result

is

the

same as

multiplying

by 2. A

number

shifted

one

place

to

the

right

has

been

divided

by

2.

A

group

of

word

transmission

instructions

is also

provided.

These

instructions

are

concerned

with

the

movement,

at

high

speed,

of

words

or

parts

of

words

from

one

location

or

register

to

another.

In

particu-

lar,

information

may

be

either

stored

or

taken

from

locations

in

core storage

and

various registers

in

the

central

processing

unit.

Since

the

word

transmission

instructions

are

concerned

with

the

movement

of

data,

they

are

used

frequently.

The

store

(STO)

instruction

stores

the

contents

of

the

accumulator

in

the

location specified by

the

ad-

Central

Processing

Unit

15

dress

part

of

the

store instruction.

Other

store instruc-

tions accomplish

the

storing

of

the address portion,

tag,

and

other

portions

of

both

the

accumulator

and

the

multiplier-quotient registers.

The

computer

also has certain load instructions

to

accomplish the same

end

as

the

store instructions, ex-

cept

in

reverse. A store instruction places the contents

of

a register

in

storage. Conversely a

load

instruction

takes the contents

of

a storage location

and

places

it

in

a register.

There

are also

load

instructions for

the

index

registers.

With

transmit, store,

or

load

type instructions,

the

contents

of

the storage location

or

register from which

the

data

are being moved

remain

unchanged.

Shift instructions are used to move the contents

of

the accumulator

and/or

the

multiplier-quotient regis-

ters

either

to the

right

or

the left

of

their original posi-

tions.

With

the exception

of

the

rotate-MQ-Ieft instruc-

tion, zeros are automatically inserted

in

the vacated

positions of the register.

Thus,

a shift larger

than

the

bit

capacity

of

the register causes the contents

of

the

register to be lost

and

replaced by zeros.

When

a shift

instruction is interpreted, the

amount

of

the shift is

determined by

bit

positions

28

through

35

of

the

shift

instruction.

An

example is given

in

Figure

27.

Instruction

Accumulator left shift

(ALS)

Operation part Address part

+0767

00000100 = 4

SQPl

35

Accumulator contents

,,0..Q01010oo100111oo101001110111ooog.Y),01

before shift

LOS;W

4'$

Accumulator contents 01000100111001010011101110000111010000 after shift

Figure

27.

Accumulator Shifting

Control instructions are defined

as

instructions gov-

erning

the flow

of

a program.

They

are the instruc-

tions

that

cause the

computer

to alter the

normal process of taking its instructions from sequential storage locations. Control

or

"transfer" instructions may

be divided

into

two types:

(1)

unconditional transfers

specifying the location of

the

executed,

and

(2)

conditional transfers performing a

test

of

some kind.

The

location

of

the

tion

then

depends

on

the

outcome

of

the test.

Unconditional

Conditional

Transfer

(TRA)

2000.

The

come from location

2000.

Transfer

on

If

the

contents

of

the

ac-

plus

(TPL)

2000.

cumulator

are plus.

the

be

taken from 2000.

If

not

plus.

the

instruction will be taken.

Certain

test instructions exist

and

are similar to

conditional transfer instructions

in

that

they cause

16

IBM

709-7090

some test to be performed. Unlike conditional transfers,

the

test instructions

do

not

specify a location for

the

is fixed relative to the position

of

the test instruction

in

the

program.

An

example is shown

in

the follow-

ing

program.

LOCATION

INSTRUCTION

100

Add

10l

P-BitTest

102 Shift

Right

103

Store

REMARKS

If

the P bit

of

the

accumulator

is a I.

the

(102)

is skipped

and

the

instruction located

after

that

(103)

is executed.

If

position P con-

tains a

O.

the

(102)

is executed.

Many

of

the indicators may be tested

in

this

manner

and

the

flow

of

the

program

adjusted to

fix

the cause

of

the

indication

or

to proceed

around

it.

Another

important

group

of

instructions

is

the

one

concerned

with

the indexing operations.

These

in-

structions are explained

under

the indexing section of

the manual.

Information

Paths

The

core storage

unit

is normally connected directly

to

the

central processing

unit.

It

is

also the site

of

the

stored program

that

controls the entire

computer

sys-

tem.

On

the

other

hand,

the auxiliary storage

and

in-

put-output

devices are normally disconnected (not

physically

but

logically) from the system.

They

be-

come connected only by execution of certain stored

program

instructions.

The

contents

of

these devices may control the computer only after being transmitted to core storage.

Thus,

data

flows

between

input-output

devices

and

core storage

through

the central processing

unit

in

the

704

system (Figure

28)

.

T

Figure 28. Information Flow.

IBM

704 System

Using Figure 28, assume

marked

are tape,

word. As these sembled sembled, this

unit

fied by Actually, two

from

"1"

arranged

it

takes six

in

and

from

the

instruction

any

input-output

is

reading

in

six-bit

of

groups

the

tape

word

there

instructions

first is a read-tape-I

eration

execution

tape

the instruction is should copy-1000 would single copy

age. more

a different address

loop"

plained

the

part

and

of

this

unit,

(2)

tape

moving

needed

the

operation

enter

(core storage) as

instruction

be

entered

instruction

If

two

or

copy instructions

INSTRUCTION

Read Copy Copy Copy 5000

This

type

of

or

a copy

under

The

information

709 system are

Dota

Synchronizer

Channel

II

A"

"tape instruction:

puts

in

the

is called a copy instruction.

part

into

more

part.

ADDRESS

Tape

I 1000 1001

instruction

routine. A true

the

heading

paths

Channel"

that

the

data.

Because

groups

these

are

control.

is

sent

to

the

that

device

across

groups

read

When

to

the core storage location specicaused are

required

into

instruction

1" as its address

it

into

read

proper

with

direction.

the

were executed,

magnetic

the

to

make

from

tape, they

a full

central

the

core storage.

with

"read"

(1) selects

status,

location

the

address

the

width

tape

to

and

that

the

tape

up

word

processing

to

read

as its op-

part.

the

(3) starts

The

part. data

core storage location 1000. A

moves

words

shown

Unit

must

Gets Puts Puts Puts

"Index

and

in

BII

one

full

are

to

be moved, two

be

furnished, each

REMARKS

tape

1 ready

that

word

second word

third

group

into

word

is called a "copy

"loop"

Registers."

the

components used

Figure

word

and

moving

location 1000

into

location

into

location 5000

is fully ex-

29.

Comparing

into

unit data

of

the

a 36-bit

are

as-

is

as-

be

read.

data The

The

proper

second

"Copy"

the

data

If

read

stor-

with

1001

Figures tween from central a put-output

an is to free ence between tral transmitted and panded

28

and

29,

the

two systems

or

going

to core storage

processing

data

synchronizer is used as

devices

With

the

709

computer,

input-output

to

transmit

receive

to

the

do

processing

operation

(read tape)

data;

other

the

unit

calculations.

704

while

is

then

free

to

under

the

note

that

is

unit.

Instead, a new

and

core storage.

then

the

and

the

must

wait

the

709 merely starts

do

other

section

nizer."

The

7090

a

or

information added

in

the chronizer's and

"data

the

section

shows

information

computer

paths

7090 system.

operation.

channel."

"IBM

system uses basically

as

does

the

They

Their

766

Data

flow

and

system.

in

There

tached have shown 80

punches,

Figure the and

The

may be a

to

the

same

in

Figure 30.

magnetic

and

30 shows

the

maximum

multiplexor.

complement

This

tape

units,

eight

printers. that

eight

the

7090 system has even less

the

input·output

main

difference

devices

in

feature

all

gression from a synchronous

asynchronous put·output is

defined

computer

controls

as

one

(709-7090),

are

concerned.

operation has finished, while asynchronous means occurrence

of

several operations.

the

major

difference be-

the

fact

that

data

coming

do

not

pass

through

device called

the

go-between for in-

the

stored

by

defining

and

what

central

Note

709. until

work.

"IBM

709.

They

are

operation

area

processing

the

In

the

all

This

766

Two

replace

called

is

program

which

in

storage

starts

device

unit

main

differ-

704,

the

cen-

data

have

been

the

operation

subject

Data

new

the

is

Synchro-

the

same units data

syn-

"multiplexor"

explained

Synchronizer." Figure 30

components

of

eight

Each

data

of

input·output

gives a

card

readers,

central

control

than

does

three

computer

happening

in

data

channels at-

channel

maximum

eight

processing

over

core storage

the

709 system.

systems is

(704) to

as

far

as

Here,

synchronous

after

the

7090

may

devices as

card

unit

the

pro-

the

another

simultaneous

the

is is

ex-

are

in

of

an in-

Figure

29.

Information

Flow,

IBM

709 System

Figure

30.

Information

~~~~J~-"""1Central

L

Flow,

IBM

7090 System

Central

Unit

Processing

Unit

17

Indexing

and

Indirect Addressing

The

copy

routine,

described previously, is acceptable

if a limited.

number

of

words are to

be

copied.

For

example, 1,000 copy instructions are needed to copy

1,000 words.

This

wastes storage space. However,

the

copy address

could

be altered by

the

program

each

time

it

is

used.

A

program

to copy 1,000 words from tape

number

I

and

place

them

into

consecutive storage locations

could be as

in

Figure 31.

In

this program,

thirteen

instructions

and

three constants are

required

to modify

the address

of

the

copy instruction

and

test for

the

end

of

the

copy loop.

The

program

is considerably

improved over the

routine

in

which 1,000 copy in-

structions are needed. However,

the

use

of

an

index

register makes

the

program

much

more

efficient.

The

computers

contain

three

index

registers

that

are

important

to

the

system's

operational

abilities.

These

registers are

termed

A, B,

and C or

1,2,

and

4.

The

latter

terminology is more convenient for

the

programmer

because

the

numbers used are

the

octal

representation

of

the

"addresses"

of

the

three

index

registers.

Inst

.

Loc.

Instruction

Remarks

50

Read

tape 1

Get

tape I ready

to

read

51

Copy

200

Put

the lst

word

into location

200

52

CLA62 Clear the accumulator

to

zeros and add

10

it the

contents

of

location

62

(0000----000)

53

ADD

64

Add

to

the accumulator contents the contents of lo-

cation 64

(0000----001).

These

instructions

make a counter which

is

increased

by

one

for

each

copy instruction

executed.

54

STO

62

Store the result

in

location

62.

55 CLA63

Clear and add into the accumulator, the contents

of

location

63

(0000----1000).

56

SUB

62

Subtract

from

the accumulator (1000) the contents

of location 62.

(This

is

the location that will in-

crease

with

each copy instruction

execution.)

57

TRZ

65

Transfer when the accumulator contents are

zero.

This

will occur

when

1000

copies have been

ex-

ecuted.

58 CLA51

Clear the accumulator and add the contents of lo-

cation

51

(copy 200).

59

ADD

64

Add

the contents

of

location 64

to

the contents of

the accumulator.

This

will increase the copy

address

from

200 to

201

the first time.

The

ad-

dress will then be increased

by

one each time the

instruction

is

executed.

60

STO

51

Store the result back

at

location

51

•

61

TR

51

Transfer back

to

location

51

so

that the next

word

will

wi

II

be read in.

(Remember

that the copy address

has

been increased

by

one.)

62

00---0000 Constants:

63

00---1000

64

00---0001

65

STOP

The

program will reach this instruction after

1000

words

have been copied and will stop.

Figure 31. Sample

Program

18

IBM

709-7090

operation

tag-bits

address

1-1

___

.....

1-=0_0"--'1

..... 1 __

,---_1

register

specification

octal

alphabetic

A

10 1 0 I

11

0 0 I

S,2

11

18

2021

35

Figure 32. Index-Register

Tag

Bits

2

4

B

c

The

addresses

of

index

registers are

stipulated

in

a

part

of

the

instruction

word

known as

the

tag field.

Such addresses are normally referred to as tags.

The

tags

perform

the

same function as

the

address

~f

the

instruction.

They

tell

the

computer

whether

an

mdex

register is to

be

used

and

which register is to

be

used.

The

tag field is located

in

positions 18,

19,

and

20

of

the

instruction

(Figure 32) . By

having

more

than

one

tag

bit

in

the

tag

field, two

or

more

index

registers

may be used

in

a single instruction.

Thus,

the

con-

tents

of

the registers would be combined

and

the

re-

sultant

sum

would

be

used.

The

index

registers are

15

positions

long-large

enough to

hold

the

largest possible storage address.

They

are used to modify

an

address by

adding

the

complement

of

their

contents to the address.

This

re-

duces

the

address by

the

contents

of

the

index

regis-

ter.

Many

instructions may specify

index

action, thus

making

them

useful for such functions as address

modification

and

counting.

As

an

example

of

the

arithmetic involved

when

in-

dex registers

are

used, assume

that

index

register I

contains

the

number 2 and

that

the

copy instruction,

with

an

address

of

200,

is

to be executed.

The

follow-

ing

occurs (Figure 33) .

When

the

copy

instruction

is decoded,

the

tag

bit

in

position

20

specifies

index

action; thus,

the

contents

of

index

register 1 are complemented

and

placed

in

the adders.

Note

that

the

contents

of

an

index

reg-

ister are always complemented when sent to

the

ad-

ders.

This

feature results

in

subtracting

the

contents

from

the

address.

The

address

portion

of

the

copy in·

struction is also placed

in

the

adders;

after

adding

the

two numbers,

the

result (called

the

effective address)

is

used

in

execution

of

the

copy instruction

instead

of

operation

tag

address

1000111000000100000010011000000010000000 I

address

part

of

instruction·-·0000000100oo000

index register contents - - -111111111111101

_111111111111101

carry

is

added into ( 000000001111101

units postion - - - - )0 1

Effective address -

- - - 000000001111110 =

176

Figure

33.

Index-Register Arithmetic, Subtracting

/0001110000001000000100110000000100000001

Address

part

000000010000000

index

register contents 000000000000010

--000000000000010

Effective address

OOOOOqoHl,OOqolO

= 202

Figure

34.

Index·Register

Arithmetic,

Adding

the

actual address.

In

this case,

the

effective address

is 176.

If

the

programmer

wishes to increase

the

effec-

tive address,

the

number

to

be

placed

in

the

index

register is inserted

in

complement

form.

Thus,

when

the

address

and

index

register contents

are

combined,

the

result

is

an

additive process.

Using

the same facts

(as

in

Figure

33)

with

the

index

contents

in

comple-

ment

form,

the

effective address

is

now

202

instead

of

176 (Figure

34).

With

an

instruction available to reduce

the

contents

of

the

index

register each time

it

is used,

the

effec-

tive address is easily modified.

The

instruction

is

called

"transfer

on

index"

(TIx)

and

has

the

format

shown

in

Figure 35.

The

operation

of

the

instruc-

tion

is

as follows:

If

the

number

in

the

specified

index

register

is

greater

than

the

decrement portion,

the

in-

dex register is

reduced

by

the

amount

of

the

decre-

ment

and

the

computer

takes its

from

the

location

specified

by

the

instruction

address.

If

the

index

register contents

are

equal

to

or

less

than

the

decrement,

the

register is

not

changed

and

the

in

sequence

is

executed.

IOperation

I Decrement I Tag

Address

S,1

23

17

18

2021

35

Figure

35.

TIX

Instruction

Format

To

apply

this instruction, assume the same condi-

tions set

forth

for

Figure

31. Follow

the

copy instruc-

tion

with

the

transfer-on-index

instruction

with

a de-

crement

portion

of

I.

The

result

would

be

as

follows:

True

Index

Register

Contents

Decrement

Contents

of

TIX

Instruction

True

Result

after

Decrement

Is

Subtracted

000000000000010 000000000000001

000000000000001

This

use

of

index

registers for address modification

can

now be

applied

to a

program

to copy 1,000 words

from a

tape

unit

into

1,000 storage locations.

The

program

can

be

written

as

follows:

LOCATION

INSTRUCTION

DECREMENT

TAG ADDRESS

50

Read

tape

51

Load

index

register

100

52

Copy

1200

53

Transfer

on

index

(0001)

52

54

Compute

100

Constant

(1000)

Location 50.

This

instruction

selects

tape

unit

1 for

reading.

Location 51.

The

load

index

instruction

places

the

contents

of

location 100

into

index

register

I.

Loca-

tion

100 contains

the

quantity

1000.

Location 52. Because

the

instruction

is

tagged, the

address

of

the copy

instruction

is

treated

as

explained

in

Figure 33.

The

first time

cpy

is executed, the con-

tents

of

the

index

register

are

subtracted, resulting

in

an

effective address

of

200.

The

index

register is

then

reduced

by

the

decrement

of

the

transfer-on-index in-

struction

so

that

successively

higher

locations

of

stor-

age are used each time

through

the

loop.

Location 53.

The

TIX

instruction,

having

a decre-

ment

of

I,

reduces

the

index

register each time

it

is

executed for a total

of

1,000 times.

After

1,000 words

have

been

copied,

the

decrement

of

the

TIX

will

equal

the

index

register.

No

transfer will

occur

and

the

instruction

(54) will be executed.

Location 100.

The

constant

of

1,000 is

needed

to

load

the

index

register

when

the

instruction

at

loca-

tion

51

is executed.

By

comparing

this

routine,

properly

called a copy

loop,

with

the

routine

explained

before, the

number

of

instructions

needed

to copy 1,000 words

from

tape

has

been

reduced

from 1,001 to

four

instructions

and

one

constant

(through

the

computer's

ability to

count,

modify addresses,

and

test for the

end

of

loop

with

one

instruction) .

By using

the

many

indexing

operations

available

in

the

computer,

an

index

register

may

be

added

to, sub-

tracted from,

and

tested

in

various combinations

so

that

many

desired effects

may

be

accomplished.

1£

more

than

one

index

register is specified

by

the

tag field,

the

procedure

is

the

same except

that

the

contents

of

two

or

more

index

registers

are

combined

and

the

result

is

then

subtracted

from

the

address field

of

the

instruction

being

executed.

The

variable address concept is

expanded

and

in-

cludes a feature called indirect addressing.

Just

as

index

registers

are

addressed by tags,

indirect

address-

ing

is specified

by

the

presence

of

l's

in

positions

12

and

13

of

the

instruction.

This

portion

is

called a

flag.

For

example, assume

that

the

instruction

shown

(Figure 36) is to

be

executed.

Instead

of

subtracting

the

contents

of

location 2054 from

the

contents

of

the

accumulator

register,

the

computer,

because

of

the

operation flag

tag

address

location

2054

I subtract

III 1

000

120541

I 1

1500

I

S

21

35

Figure

36.

Instruction

Format

Central

Processing

Unit

19

flag, examines the location specified by the address

(2054)

and

uses the address

part

of

that

location

(1500) to

obtain

an

effective address.

The

subtraction

is

then

executed

as

if

its address

part

had

contained

1500 instead of 2054.

Thus,

the

contents

of

location

1500 are subtracted from

the

contents

of

the

accumu-

lator

register.

The

indirect addressing feature may be combined

with

indexing to

obtain

even greater flexibility

in

op-

eration. Using the same example

and

adding

an

index

register tag,

the

operation

would be

as

shown

in

Figure 37.

location 2000

I I 3750 I

21

35

index reg i ster 1

[OO5fJ

Figure

37.

Indirect Addressing

and

Indexing

Format

The

contents of

index

register 1 are subtracted

from the address

portion

of the subtract instruction

giving

an

effective address of 2000. Location 2000 is

then

examined

and

its address

portion

(3750) be-

comes a second effective address.

Thus

the contents

of

location 3750 instead

of

2054 are subtracted from

the

contents

of

the accumulator.

Sample Problems

The

following figures show a few examples

of

some

of

the

problems

that

computers are solving today. Al-

though

the examples shown are

in

scientific fields,

the

computers are

adaptable

to business applications.

They

are used for payroll, material inventory, control

of

crude

oil

processing,

and

many

other

business ap-

plications where efficient

operation

depends

upon

prompt,

high-speed results.

The

first application shown is

in

the field

of

scien-

tific research (Figure 38) .

Scatterer I Scatterer

II

-"

-----.

Figure

38.

Neutron

Scattering

20

IBM

709-7090

I

~

System

Used

~709system

rfl

~7090

system

Time

Taken

75 seconds

15

seconds

Figure 39.

Time

Comparisons. Neutron Scattering Problem

Neutron

Scattering

Given:

l.

A spherically symmetric system consisting

of

two

scattering materials.

2.

An

initial

neutron

distribution

in

space

and

energy.

3.

Scattering laws for the materials

and

dimensions

for the system.

Find:

I.

The

neutron

leakage

rate

from the system,

caused by scattering.

2.

The

asymptotic

neutron

distribution

in

space

and

energy.

The

initial

collision distributions are combined

ac-

cording to

the

scattering laws

and

are inserted

into

matrix

equations.

These

are

multiplied

by

the

mat-

rices

and

a new set

of

distributions is obtained.

The

procedure is repeated

on

an

iterative basis

until

the

asymptotic

or

limiting

phase is reached.

Time

comparisons for

the

solving of

25

of

these

iterations

appear

in

Figure

39

and

are approximate.

Flight Trajectory

The

second application

is

in

the field

of

national

de-

fense (Figure

40).

The

path

of

an

object which is

launched

from some arbitrary

point

is calculated. Its

trajectory is

determined

by its flight characteristics,

Initial

--l

Velocity I

Components

j

I

Initial

j-

Initial

AI

Figure 40. Flight

Trajectory

Ground

System Used

Time

Taken

00

---

709 system

100 seconds

ri-l-- 7090 system

20 seconds

Figure 41.

Time

Comparison for Each

Trajectory

the

initial

components

of

its velocity,

an

initial

alti-

tude,

and

the

initial

angle.

The

position

of

the object is identified by its

co-

ordinates with respect to fixed axes passing

through

the

launching

point.

Factors

in

this

problem

are:

1,100 time steps

per

trajectory.

1,000 operations

per

time step.

1,100,000 operations

per

trajectory.

The

time comparisons for each computer are ap-

proximate

and

are shown

in

Figure 41.

Electron Density

The

third

application is

in

the field

of

fundamental

research (Figure

42).

The

electrons

surrounding

two nitrogen atoms which form a nitrogen molecule) are treated

as

a degenerate Fermi-Dirac gas. Such treat-

ment

results

in a partial

differential

equation

called

the Thomas-Fermi-Dirac

or

"statistical field" equa-

tion.

The

computer calculates solutions to this

equation

and

thereby determines

the

density of electrons

in

the

molecule

and

the energy

of

the molecule.

The

Thomas-Fermi-Dirac

equation

is replaced by a differ-

ence system over a network

of

points.

The

network

consists

of

the intersections

of

60

quasi-Iemniscates

Figure 42. Electron Density

[

"1"

P~;I

.....

,..".,

r

:W

800 years

Desk

co I cu I

ator

4000 weeks

IBM 701 system

2 hours

IBM

709 system

25 minutes

:Ii:

IBM 7090 system

5 minutes

Figure 43.

Time

Comparison of

Computer

Systems. Electron

Density Problem

with

15

curves

at

right

angles to them.

The

following

procedures are used to solve the problem:

900 simultaneous non-linear equations.

80

iterations

per

solution.

900,000 operations

per

iteration.

72,000,000 operations

per

solution.

This

application graphically points

out

one of the

problems

that

could

not

be solved

without

the use of

high-speed computers. Again,

the

time comparisons

are approximate. Note the

great

range

of

times.

The

span

of

800 years

required

with

a pencil

and

paper

is

reduced to

five

minutes by the

IBM

7090

Data

Process-

ing

System (Figure 43) .

Average execution speeds for

both

fixed

and

floating-point arithmetic operations (including the time needed to take a word from

or

put

a word

into

core

storage) is shown

in

Figure

44.

The

use

of

transistors

in

the

7090 system provides

a saving

in

electrical power

and

air

conditioning re-

quirements.

The

reduction

of

electrical power may

be as great as seventy percent

of

the

709 requirements.

The

nodular

unit

design

of

the 7090 makes

it

pos-

sible to place units side by side

and

thus minimize

space requirements.

As

much

as

fifty

percent

of

the

709 system space requirements may be saved.

Central

Processing

Unit

21

FIXED

POINT OPERATIONS

Operation

709

Ti

me

and

Op/See

7090

Time

and Op/See

Add

or

Subtract

24

usee·

41,660

4.4

usee

227,270

Logical Operations

24

usee

41,660

4.4

usee

227,270

Multiply

190

usee

5,260

25.3

usee

39,640

Divide 240 usee

4,160

30.5

usee

32,780

FLOATING POINT OPERATIONS

Operation 709

Ti

me

and

Op/See

7090

Time

and

Op/See

Add

or

Subtract

84 usee

11

,900

14.0

usee

71,420

Multiply 170 usee

5,880

24.0

usee

41,660

Divide

216 usee

4,620

28.3

usee

35,330

Operations-per-second numbers are approximate.

*usec = microseconds

Figure 44. Average Execution Speeds

Figure

45.

IBM

709 Console

22

IBM

709-7090

Operator's

Consoles

The

709

operator's

console

is

a physical

part

of

the

central

processing

unit

(Figure

45).

It

consists

of

neon

lights, incandescent lamps, keys,

and

switches.

The

contents

of

the

main

central

processing

unit

reg-

isters are displayed

in

neon

lights. A 1 is

represented

by

the

light's

being

on; a 0,

by

the

light's

being

off.

The

36

entry

keys

are

available to

manually

insert

a

word

or

words

into

the

central

processing

unit

and

core storage.

The

IBM

7151

Console

Control

(Figure 46A) is a

separate

unit

providing

centralized

control

of

the

7090

system.

It

contains indicators, switches, keys,

and

reg-

ister displays for

the

operator's

use.

Channel

indica-

Figure

46A.

IBM

7151

Console

Control

(7090)

tors for

data

channel

operation

are

provided

and

the

register displays have

been

grouped

for

convenience.

Data

in

any

storage

location

can

be displayed

and/or

changed

by

manual

insertion

through

use

of

the

entry

keys

and

switches.

A

maintenance

feature,

marginal

checking device,

is also a

part

of

the

console.

This

feature

includes

the

capability

of

varying voltages

and

frequencies

during

programmed

diagnostic testing to detect

potential

difficulties before

actual

failure

of

components

occurs.

The

IBM

7617

Data

Channel

Console

(Figure

46B)

provides

for

greater

input-output

flexibility

and

effi-

ciency

between

the

IBM

7090

Data

Processing System

and

its

operator.

One

data

channel

console is used

with

each

IBM

7607 I

or

II

Data

Channel.

The

con-

sole

may

be

located

up

to 50 feet away

from

the

data

channel,

permitting

the

computer

system console

and

data

channel

consoles to be

grouped

for

greater

op-

erating

efficiency.

Several

operations

may

be

performed

from

the

data

channel

console

when

the

console is

in a manual

mode.

These

include:

1.

Data

transmission

operations

2.

Non-data

tape

operations

3.

Loading

operations

4.

Storing

operations

Figure

46B.

IBM

7617

Data

Channel

Console

Central

Processing

Unit

23

Input-Output

Components

Magnetic

Tape

Storage

The

principal

input-output

medium

for

the

computer

is magnetic tape.

The

tape

is used,

in

addition

to in-

put-output

functions, for storage

of

intermediate re-

sults.

It

is also used as

permanent

storage for large

files

of

data. Magnetic tapes may

be

re-used

many

times, because

old

data

are automatically erased as

new

data

are recorded.

The

tape is a plastic

material

that

is coated

on

one

side

with

a metallic oxide.

It

is one-half

inch

wide

and

is packaged

on

reels

with

tape lengths as great as

2,400 feet.

The

tape can

be

easily magnetized

and

yet retains

the

magnetized "spots"

when

put

in

static

storage for indefinite periods

of

time.

The

basic electronic

principle

of

magnetic tape re-

cording, as used

in

data

processing systems, is similar

to

that

of a home

tape recorder.

Instead

of

recording

music

or

voice, business

and

scientific

data

are re-

corded

in

the

form

of

magnetized spots. A schematic

of

the

recording

head

together with a section

of

mag-

netic

tape

is shown

in

Figure 47.

When

electrical

current

flows

through

the record-

ing

head

coil,

the

magnetic oxide particles

on

the

tape

are

oriented

in

one

direction.

If

the

current

in

the

coil reverses its direction,

the

particles

on

the

tape will

be

oriented

in

the opposite direction.

This

is called

reversing

the

polarity

of

the

magnetic flux.

Each

such

change

of

polarity

is

given the value

of

1.

If

no

change occurs,

the

value

is

given as

O.

Now, if

the

tape is moved past

the

recording

head

and

the

current

in

the coil is

alternated

at

time inter-

vals,

writing

on

magnetic tape is accomplished.

To

read

the

tape, the process

is

reversed.

As

the

magnetized "spots" pass

the

recording head, small

voltages

appear

in

the recording coil.

These

voltages

are amplified (as

in

core storage)

and

are

then

used

by

the

computer.

The

advantages

of

tape storage are shown

in

Fig-

ure

48.

R/W

HEAD

Figure 47. Magnetic

Tape

Recording

24

IBM

709-7090

Master Records

Low

Cost

•••

Compact

•••

One foot of

tape

wi

th its

compact starage has a

re-

cording density of 200 or

556 and

m~y

contain 14,400

or

40,030

binary

digits.

Fast

••• Reading and writing speed of

90,000

ta

375,000

binary

digits per second.

Variable Record Size

•••

From

one to several thousand alphamerical characters per record: Recording Density 556

=

4,620 records of 3000 characters to 37,380

six-character records

on each

reel of

tape.

Updated Records

Working Records

Permanent

••• Reels may be used many times

for

many

different jobs.

One

reel of tape has

a

capacity

equal

to

several thousand

IBM

cards.

Figure

48.

Magnetic

Tape

Advantages

Check

Bits-..(

1

r

60

121

1.0

241

301

Data

bits-

i

0

1 0

0 1

1

0

1 1

1

0

1 0 1

1 1 1

1

0

51

,,0

170

231

2.0

3.1

Figure

49.

Binary

Word

Recorded

on

Tape

During

actual

tape

unit

operation,

seven recording

heads instead

of

one

are

placed

vertically across the

tape.

Thus,

recording occurs

in

seven columns

or

"tracks."

With

the

7'09

or

7'090

systems, two different

"modes"

of

recording

are

used,

binary

and

binary

coded decimal

(BCD).

If

data

are

recorded

on

magnetic

tape

just

as

they

are

found

in

core storage (except for

the

check

bits),

the

tape

is

said to

be a binary

tape

or

to

have

been

recorded

in

the binary mode.

The

seven recording

tracks

are

used as follows:

1.

One

track, C, is used for check-bit recording.

2.

Six tracks, B, A,

8,

4, 2,

and

1,

are

used to

record six bits

of

the

36-bit word.

Thus, a binary

word

requires six recordings

of

six bits

each, as is shown

in

Figure

49.

The

check track contains a 1

bit

whenever the ver-

tical (across

the

tape)

sum

of

the

1's

in

the

B, A,

8,

4,

2,

and

1 tracks

is

even.

This

check

bit

is

computed

by

the

tape

control

in

the

7'09

and

the

data

channel

in

the

7'09'0.

Alphabetic

and

decimal

information

may

be re-

corded

on,

or

read

from, a magnetic

tape

without

de-

pending

upon

the

computer

through

use

of

auxiliary

equipment.

Magnetic tapes

prepared

or

used

on

this

equipment

have a special coding system

known

as

binary-coded-

decimal

(BCD).

As

the

six-bit BCD characters

are

read

from

the

tape, some

of

the

characters

are

altered.

This

alteration

is

performed

so

that

the

digits

'0

through

9

and

the

characters A

through Z are

represented

in

core

storage by six-bit

binary

numbers.

The

alteration

of

characters is shown

in

the following table:

Core Storage

K M 8 9 7 3

Figure

50.

BCD

Word

Recorded

on

Tape

CLASS

IN

CORE

STORAGE

ON

TAPE

B

A

B A

Numerical

0 0

A

to

I 0 I

1

J

to

R

I

0

S

to

Z

0

The

digits 1

through 9 are

represented

by

the

six-

bit

binary

numbers

'0'0'0'0'01

through

'0'01'0'01

(their

ex-

act values as

binary

integers).

Thus,

the B and

A

(zone parts)

of

these digits

would

be

'0'0.

The

num-

ber

zero is represented

on

tape

by

the

bit

configura-

tion

'0'01'010.

This

representation

would

be automati-

cally

altered

to

'0'00'0'00

when

reading

in

the

BCD mode.

During

writing

in

the BCD mode,

the

alteration

procedure

is reversed

so

that

the

BCD characters

in

stor-

age

are

transformed to

the

BCD

tape

format.

The

zone

portions

(B

and

A bits) are

altered

and

the

number

zero

('00'0'0'0'0)

is replaced by

'0'0101'0.

In

addition

to

alphabetic

and

numerical

characters,

the

BCD

format

provides for

punctuation

marks

and

other

special symbols.

Included

is

the

BCD character

BLANK

which

suppresses

printing

or

punching

in

any

desired

position

during

auxiliary operations.

Figure

5'0

shows

the

characters K M 8 9 7 3 as they

would

appear

on

tape

and

in

magnetic core storage.

Each

word

in

core storage may

contain

six BCD char-

acters.

Note

that

the check

bit

occurs

when

the

sum

of

the

l's

in

both

the

numerical

and

zone tracks is

odd.

Tapes

prepared

in

the

BCD

mode

are

compatible

with

those

prepared

on

the

IBM

65'0,

702,

7'05,

and

7'07'0

Data

Processing Systems.

Figure

51

represents a

section

of

magnetic

tape

with

all

of

the

possible char-

acters recorded

on

it

in

the

BCD mode.

0123456789

A8CDEFGHIJKlMNOPQRSTUVWXYZ

&.a-S·/,'IU@

Check

{

C

II

I

II II

I

II

III

I I

II

I I I

II

Zone

{

8

111111111111111111

111111

A

111111111

11111111

III

N_,i,o'

{

8

II

II II

1111

4

1111

I

III

1II1

1111

I

I I I

2

I

II

I I

I I I I I I I

I

II

I I I I I I I

I

II

I

Figure

51.

Tape

Character

Coding.

Input-Output

Components

25

Tape Mark

Check Character

Figure 52. Record

and

File Gaps

on

Magnetic

Tape

As

information

is

recorded

on

tape,

the

number

of

1 's

in

each track across

the

tape

is

automatically

counted

and

a check

bit

(1)

is

placed

in

the

seventh,

or

C, track.

When

the tape is read,

the

check

bit

is

recomputed

and

compared

against

the

recorded check

bit. Any discrepancy will signal

an

error

in

tape

op-

eration.

In

addition

to this vertical checking of each column,

a horizontal

count

of

each track

containing

l's

is taken

and

recorded

at

the

end

of

the

record

as a separate

column

called

"longitudinal

redundancy

check char-

acter"

(LRCR).

A

word

or

group

of words, recorded consecutively

on

tape, is referred to

as

a record.

The

number

of

words

making

up

a record is unrestricted

within

the

storage limitations

of

the

system itself.

If

more

than

one

record is placed

on

the

same tape, a %

inch

sec-

tion

of

blank

tape

called a record gap

is

placed be-

tween them. Records

may

also

be

grouped

together

forming

files.

These

files

are

separated by

other

blank

sections

of

tape

called file gaps.

This

file gap consists

of

a section

of

blank

tape

and

includes a special

character

called a tape

mark.

This

character

and/or

gap

defines the difference be-

tween

the

records

and

files

on

the

tape

(Figure 52).

The

number

of

words

in

a record

and

the

number

of

records

in

a file are variable

and

are

determined

by

the

stored

program

and

the

limitations

of

the system.

When

data

are

to be recorded

on

tape, a full word

of

36

bits is taken from storage

and

is subdivided

into

Figure 53. File

Protect

Ring

26 IBM 709·7090

groups

of

six bits each.

Each

group,

with

its associated

check bit, is recorded

in

one

column

across

the

tape.

Tapes

are

written

or

read

in

the

forward direction

only. However, the same tape

may

be written, back-

spaced,

and

then

read

or

rewritten

if

desired. Back-

spacing is

halted

by the record

gap

or

file gap; thus,

one

backspace

instruction

results

in

moving the tape

in

a backward direction to the

or

file gap.

Because the writing

operation

automatically erases

any previous

information

recorded

on

the tape, a file

protection

device is provided to

prevent

accidental

erasure

of

information. A circular groove

is

molded

around

the

center

of

each

tape

reel to fit a demount-

able plastic ring.

Without

the

ring

in

place,

writing

is suspended

and

only

reading

may

occur.

The

reel

in

this

condition

is protected.

When

the

ring

is

in

place,

either

writing

or

reading

may

be

performed

(Figure

53).

Like movie film, the

tape

must

have a

short

length

of

blank

space

at

the

beginning

and

at

the

end

of

the

reel

that

can

be

threaded

through

the

feeding mechan-

ism

of

the

tape

unit.

Reflective spots

of

aluminum

foil,

placed

on

the

tape

by

the

operator

at

any de-

sired distance from

the

ends

of

the

tape,

are

photo-

electrically sensed to indicate

the

physical

end

of

tape

and

the

starting

point

for

recording

(Figure 54) . Dur-

ing

writing

operation,

the

reflective

spot

signals

that

the

end

of

the

reel has

been

sensed.

Figure

55

shows schematically

the

position

of

the

tape reels

in

relation

to

the

read-write head, feed roll-

ers,

and

the

vacuum

columns.

While

reading

or

writ-

ing,

tape

is

transported

from the file reel (left side)

past

the

recording

head

to

the

machine

reel (right

side) .

Since

it

is impossible to

start

and

stop

high-speed

motion

of

the

tape

without

some slack

in

the

tape, a

Figure

54. Reflective Spots

on

Tape

loop

of

tape

is

held

in

the

vacuum

columns

and

acts

as a buffer for this

motion.

As

tape

is

drawn

from

one

column,

it

is

replenished

from

the

reel above it.

As

it

is fed

into

the

opposite

column,

the

associated

reel takes

up

the

slack.

The

read-write

head

assembly consists

of

two parts,

the

lower

of

which

is

stationary,

and

the

upper

part

moves

up

and

down

under

control

of

the

load

and

un-

load

keys.

Tape

threading

must

be

done

with

the

tape

Left

Ptoloy

_L---r-T

.....

mbly

1

'~-U

Read-Write

Head

As.ernbly

Figure

55. Schematic,

Tape

Feed

Right

Prolay

Assembly

unit

in

an

unload

status.

(Changing

reels

or

thread-

ing

tape

takes

approximately

two minutes.)

Rewind

is

under

the

control

of a stored

program

in-

struction

or

may

be

initiated

by depressing

the

load-

rewind

key

on

the

unit.

If

more

than

one-half

inch

of

tape

has

been

wound

on

the

machine

reel,

the

unit

automatically

performs a high-speed

rewind

(500

inches

per

second).

The

head

is

raised,

tape

is

pulled

from

the

vacuum

columns

and

the

tape

is

moved

in

a reverse

direction

until

less

than

one-half

inch

of

tape

remains

on

the

machine

reel.

When

this occurs,

the

tape

is lowered

into

the

columns,

the

head

as-

sembly

is

lowered

and

the

backward

movement

is

reduced

to

the

normal

operating

speed

of

the

unit.

During

the

high-speed rewind,

tape

is

passed be-

tween a

light

source

and a photo

cell.

If

the

tape

breaks,

the

light

strikes

the

photo

cell causing

tape

motion

to immediately stop.

The

IBM

729 I

Magnetic

Tape

Unit

is

used

with

the

709 systems; 729

II

and

729

IV

units

are

used

with

the

7090 system (Figure 56) .

Each

unit

is

equipped

with

a selector

dial

to

be

set

from

0

through

9 by

the

operator.

The

selected

num-

ber

then

becomes

the

"address" of

the

unit.

Instruc-

tions from

the

stored

program

condition

or

alert

the

unit

for

use by specifying

this

address.

The

729 I reads

and

writes

at

a density

of

200 char-

acters

per

inch

of

tape.

This

means

that

a 100-char-

acter

record

would

occupy one-half

inch

of

tape.

Tape

is

moved

past

the

read-write

head

at

a speed

of

75

inches

per

second,

thereby

making

the

rate

of

read-

ing

or

writing

15,000 characters

per

second

or

67

microseconds

per

character.

Input-Output

Components

27

IBM

729 I

Figure 56.

IBM

Magnetic

Tape

Units

The

729

II

unit

reads

and

writes

at

one

of

two

~haracter

densities,

either

200

or

556 characters

per

mch

of

tape.

This

dual

density

feature

is

under

stored

program

control.

Tape

moves

past

the

read-write

head

at a speed

of

75

inches

per

second.

The

combination

of

character

density

and

tape

speed give

character

rates

of

15,000

or

41,667 characters

per

second depend-

ing

upon

the

recording

density used.

Since a

record

gap

is placed between each record

on

tape,

the

total

time

required

to

read a record

must

include

time

to space over

the

gap.

This

is

termed

"access

time"

to

the

data.

Average access

time

for

the

729 I

and

729

II

is 10.8 milliseconds.

Figure

57

is a

table

showing a

comparison

between

100 different size records

written

in

both

density modes

on

the

729

II

tape

unit.

Note

that

2,370 feet

of

the

available 2,400 feet

are

used, allowing 30 feet for

the

beginning-

and

end-of-tape reflective strips.

The

729

IV

tape

unit

also reads

and

writes

at

two

densities,

200

or

556,

and

is

under

program

control.

28 IBM 709·7090

IBM

729

II

and

IV

With

this model, however,

the

tape

moves

at

a speed

of

112.5 inches

per

second.

Thus,

character

density

and

tape

speed give

character

rates

of

22,500

and

62,500 characters

per

second.

For

the

729 IV,

with

its

faster

tape

speed,

the

average access

time

is

reduced

to 7.3 milliseconds.

Figure

58

shows a time compari-

son for 100 records

on

the

729

IV

tape

unit.

Figures

57

and

58 show

that

the

effective

character

rate

increases

in

efficiency

as

longer

records'

are

proc-

essed.

Rate

per

character

for

the

729

II

is 67

or

24

microseconds; for

the

729 IV,

it

is 44

or

16

micro-

seconds.

Figure

59 shows rates for 100 records.

The

729

Magnetic

Tape

Units

incorporate

two mag-

netic gaps for each

of

the seven

recording

tracks.

One

Characters Per Record

6

36 360 720

1800

3600

729

II -Low

Density

Time in Seconds

1.12

1.32

3.49

5.90

13.14

25.20

729

II

- High Density

Time

in

Seconds

1.09

1.17

1.94

2.81

5.40

9.72

Figure 57.

Time

Comparison for

the

IBM

729

II

..

•

Characters Per Record

6

36 360 720

1800

3600

729

IV

- low Density

Time

in

Seconds

.76 .89

2.31

3.90

8.65

16.57

729

IV -High

Density

Time

in

Seconds

.74 .78

1.30

1.88

3.61

6.49

Figure

58.

Time

Comparison

for

the

IBM

729

IV

Characters

729

I, 729

II -Low

Density

729

II -High

Density

Per Record

Characters per

Second

Characters per Second

6

536

550

36

2727

3077

360

10315

18556

720

12203

25623

1800

13699

33333

3600

14286

37037

729

IV -Low

Density

729

IV

- High Density

6

789

811

36

4045

4615

360

15584

27692

720

18461

38298

1800

20809

49861

3600

21726

55470

Figure

59. Effective

Character

Rates

for 100 Records

gap

is used

for

writing;

the

other,

for

reading.

The

two-gap

head

(Figure 60) offers increased checking

while

writing. A

tape

that

is

being

written

passes first

over

the

write

gap

(to

record

the

data)

and

then

over

the

read

gap

(to check

the

writing).

With

this type

of

operation,

written

data

are

automatically

read

and

727

729 I

729

II

Tape Speed (Inches per Second)

75

Record Density (per Inch) 200 200

2000r556

Character Rate per Second 15000 15000 15000

or

41667

Record

Gap

Size in Inches

3/4

Record

Gap

Time

in

Milliseconds

10.8 10.8

10.8

Maximum Number

of Tape Units

IBM

753

10

per Control

IBM

754

10

Unit:

IBM

755 8

IBM

760

2

IBM

767(1)

5

and/or

IBM

777

8

IBM

7607(2)

10

Where used:

Aux.

Card-to-

Tape

(3

x

x Aux. Tape-to-Card x x Aux. Tape-to-Printer

IBM

717 x x

IBM

72D-730 x x

IBM

650 x

IBM

704 x

IBM

705

lor

II

x

IBM

705

III

x

IBM

709

x

IBM

7070

x

IBM

7090

x

Figure

60.

Two-Gap

Recording

Head

Dual

Channels

Good

Signal

Unwanted

- - - Noise -

- - - - - - -

--

Figure

61.

Error

Detection

checked.

Each

vertical

column

of

six bits

and

the

check

bit

for

that

column

are

analyzed.

If

any

dis-

crepancy

occurred

during

the

write

operation,

it

is de-

tected

at

the

read

gap

and a tape

check

indicator

is

turned

on.

Dual-level sensing is used

with

the

two-gap

head,

to increase

error

or

weak signal

detection

at'

the

time

of

writing. Signals received by

the

checking circuitry

associated

with

the

read-write

head

are

interpreted

at

two different energy levels

(EI

and

E2

in

Figure

61).

This

interpretation

or

analysis

determines

that:

1.

The

data

signal

strength

is

at

a level

that

pro-

729

III

112.5

556

62500

3/4

7.3

5

or

x

729

IV

112.5

2000r556

22500

or

62500

3/4

7.3

10

x

vides

good

signals

when

the

tape

is

read

at a later

time.

2.

No

unwanted

signal (strong

noise) is

present

on

the

re-

corded

tape.

3.

Compensating

errors

do

not

exist because

of

a bit-for-bit

match

that

occurs

at

both

levels.

By

testing

the

tape

check indi-

cator,

the

computer

can

institute

corrective

action

whenever

an

error

is sensed.

Figure

62 shows

the

compa-

tible

features

of

the

different

magnetic

tape

units

and

data

processing systems.

(l)

A maximum of ten 729 I tape units

may

be used if

no

729

III

tape

units

are

used; otherwise, a maximum of

five

of

each type

must

be

used.

(2)

Ten

intermixed

tape

units may be used on each

IBM

7607.

(3)

Dual

level sensing is not

active.

Figure

62.

Features

of

Magnetic

Tape

Units

Input-Output

Components

29

Auxiliary

Equipment

Auxiliary

equipment, a part

of

most

data

processing

systems,

is

described below.

1. Card-la-Tape (Figure 63)

The

IBM

714

Card

Reader,

IBM

727

or

729 I Magnetic

Tape

Unit,

and

the

IBM

759

Card

Reader

Control

per-

form

independent

card-to-tape operations.

Informa-

tion

can

be

read

from a punched

card

and

written

on

magnetic

tape

for

future

processing

in

the

com-

puter

at

a speed

of

250 cards

per

minute.

Auxiliary

card-to-tape

equipment

does

the

following:

a.

It

converts

punched

card

data

into

magnetic

tape records

without

using

the

computer.

b.

It

selects

or

rearranges

the

data

on

cards for re-

cording.

c.

It

emits

constant

information.

d.

It

checks each record

automatically

after

it

is

recorded.

e.

It

selects

data

from two cards,

if

desired, to

be

combined

into

one

record

on

tape.

2. Tape-la-Card (Figure 64)

The

IBM

722

Card

Punch,

IBM

727

or

729 I Magnetic

Tape

Unit,

and

the

IBM

758

Card

Punch

Control

per-

form

independent

tape-to-card operations.

Informa-

=

tion

can

be

read

directly

from

magnetic

tape

and

punched

into

cards.

The

tape

can

he

recorded

either

on

card-to-tape

equipment

or

on

one

of

the

IBM

data

processing systems.

Tape

to

card

equipment

per-

forms

the

following operations:

a.

It

converts

magnetic

tape

records to

punched

cards

independent

of

computer

operations.

h.

It

automatically

checks

punched

data

for accu-

racy.

c.

It

operates

the

card

punch

at

a speed

of

100

cards

per

minute.

3.

Tape-la-Printer (Figure 65)

The

IBM

717

Printer,

IBM

727

or

729 I

Magnetic

Tape

Unit

and

the

IBM

757

Printer

Control

perform

inde-

pendent

tape-to-printer operations. Information,

can

he

read

from

tape

and

printed

in

the

exact

form

in

which

it

appears

on

tape.

The

tape

can

be

recorded

either

on

card-to-tape

equipment

or

on

one

of

the

IBM

data

processing systems.

The

equipment

performs

the following:

a.

It

converts magnetic

tape

records to

printed

forms

without

depending

on

computer

opera-

tions.

b.

It

prints

150 lines

per

minute,

using

120 type

wheels

of

48 characters each.

Alphabetic,

nu-

merical,

and

II

special symbols

are

used.

c.

It

checks

printed

information

automatically.

IJlM

714

Card Reader

IBM 759 Card Reader Control

IBM

727

or

IBM

729-1

Magnetic Tape Unit

Figure 63. Card-to-Tape Auxiliary Units

IBM

727

or

729-1

Magnetic Tape Unit

Figure 64.

Tape-to-Card

Equipment

30 IBM 709-7090

=

IBM

758

Card

Punch

Control

IBM

722 Card

Punch

IBM

727

or

729-1

Magnetic Tape Unit

Figure

65.

Tape-to-717 Printer

IBM

757 Printer Control

4. Tape-to-High Speed Printer (Figure 66)

The

IBM

720

or

730

Printer

(Model

2),

IBM

727

or

729 I Magnetic

Tape

Unit,

and

the

IBM

760 Con-

trol

and

Storage may

perform

independent

tape-to-

printer

operations.

Information

can

be

read

from

magnetic

tape

and

printed

in

the

form

it

appeared

on

tape.

The

tape

can

be

recorded

either

on

card-to-tape

equipment

or

on

one

of

the

IBM

data

processing

sys-

tems.

Auxiliary

high-speed

printing

from

tape

does

the

following:

a.

It

converts

magnetic

tape

records to

printed

form

without

depending

upon

computer

operations.

b.

It

prints

500 lines

per

minute

with

the

720,

or

1,000 lines

per

minute

with

the

730 Printer_

c.

It

uses 120

printing

positions

of

47 characters

each. Alphabetic,

numerical,

and

special sym-

bols

can

be

printed.

d.

It

checks

information

automatically.

5.

Auxiliary

Operations with

IBM

1401

An

IBM

1401

Data

Processing System

can

be used for

all

auxiliary

operations.

This

system consists

of

a

1401

Processing

Unit

with

as

many

as

16,000 positions

of

core storage, a 1402

Card

Read

Punch,

1403

Printer,

and

six Magnetic

Tape

Units

(either

729

II

or

729

IBM

717 Printer

IV).

The

1401

is

particularly

adapted

to such conver-

sion

operations

as: (1) card-to-tape,

with

off-line

audit,

control,

and

edit

of

input

data

at

800 cards

per

minute;

(2) tape-to-card,

at

250 cards

per

minute;

(3)

tape-to-printer,

at

600 lines

per

minute.

Data

Synchronizer

In

most

computers,

internal

processing speeds are

much

faster

than

the

input

devices

that

read

data

or

the

output

devices

that

record

the

results.

For

ex-

ample,

the

internal

speed

of

the

709 is

approximately

34 times faster

than

that

of

the

729 I

tape

unit

and

833 times faster

than

that

of

the

card

reader_ How-

ever, a

word

is

not

always available

from

tape

every

34

machine

cycles. Because

magnetic

tape

is

a plastic

medium

being

read

by a

mechanical

device,

the

data

transmission

rate

varies. Actually,

input-output

de-

vices

are

asynchronous

with

the

computer;

that

is,

no

fixed

time

relation

exists

between

these devices

and

the

computer_

Some systems overcome this

problem

by suspending

all

computing

while

an

1-0

device is

being

used.

The

core storage waits

until

the

input

device has accumu-

lated a data

word; this

word

is

stored;

the

computer

then

waits

for

the

of

data_

The

stored

IBM

727

or

729-1

Magnetic Tape Unit

Figure 66. Tape-to-720

or

730

Printer

IBM

760 Control and Storage

IBM

720

or

IBM

730 Printer

Input-Output

Components

31

Figure

67.

Data

Transmission,

IBM

704 System

program

starts

an

input

operation

by defining

the

in-

put

device

and

the

receiving storage area. Processing

is delayed while any

data

are

being moved

in

or

out

of

storage.

One

advantage to this

method

is

the

ease

in

programming

effort.

The

great

disadvantage is

the

tremendous waste

of

possible

computing

time (Fig-

ure

67).

The

data

synchronizers

permit

the 709 central

processing

unit

to

perform a more

independent

role

in

controlling

the

input

and

output

than

that

of

the

704

central

processing

unit.

In

the

709 system,

the

central processing

unit

initiates

and

monitors

1-0

op-

erations

but

is

not

interrupted

with

the

detail

of

rout-

ing

the

data

(Figure 68) .

Each

data

synchronizer is composed

of

two com-

pletely separate

and

independent

input-output

chan-

nels.

These

two

data

channels

provide

for simultaneous

computing

and

transmission between

the

input-output

units

and

core storage. A

data

channel

may

be

thought

of

as a small

computer

having

the

responsibility for

controlling

the

quantity

and

destination

of

all

data

transmitted

between core storage

and

an

input-output

unit.

It

also performs

limited

counting

and

testing

operations exclusively concerned

with

the

transmis-

sion

of

data.

No

restrictions exist

on

the

type

of

transmission

being

performed

by a

data

channel.

All

channels

may

be

used for

input,

output,

or

for a

combination

of

input-output

operations

that

will

be

concurrent

with

calculation (Figure

69)

.

The

IBM

766

Data

Synchronizer, used

with

the

709

system, provides a powerful

link

between core storage

and

the

input-output

devices.

This

new concept of

input-output

control enables processing to

be

per-

formed simultaneously

with

reading

and

writing. One,

Input·

Device

709 System

Figure

68.

Data

Transmission,

IBM

709, 7090 Systems

!l2

IBM

709-7090

Core Storage

Printer

Central Processing Unit

Figure

69.

Simultaneous

Input,

Calculation,

and

Output

two,

or

three

data

synchronizers

can

be

used

in

the

system. Since each

data

synchronizer contains two

input-output

data

channels,

one

channel

can

be

read-

ing

from a

tape

while

the

other

channel

is

writing

on

another

tape,

printer,

or

card

punch.

At

the

same

time, processing may

be

executed

in

the

central

proc-

essing

unit.

The

IBM

766

Data

Synchronizer (Figure 70) pro-

vides for

improved

methods

of

input-output

control

listed below.

They

can

be summarized as

the

ability

to:

1.

Use

the

record

structure

of

other

IBM

input-

output

devices.

This

provides for

the

use of

other

IBM

computer

tapes

with

the

709 system

thus

making

it

compatible

with

other

systems.

2.

Use any previous

IBM

computer

coding

as

input

data.

3.

Transfer

input

data

to scattered core storage

locations

without

loss

of

programming

time.

4.

Ignore

unwanted

data

in

a record.

If

desired,

unwanted

data

may be skipped.

7090 System

,

Figure

70.

IBM

766

Data

Synchronizer

5. Use

minimum

time

in

transferring

data

between

input-output

units

and

storage.

One

core stor-

age cycle is used for each

word

processed.

6.

Synchronize

the

stored

program

with

the

input-

output

operations

when

desired.

Thus

synchro-

nization

is

optional

to

the

programmer.

To

these six refinements,

add

a seventh

and

totally

new

ability

to:

7.

Operate

six different

input-output

devices

and

compute

simultaneously.

The

709 system programs involving

input-output

and

processing

are

considered as two

separate

pro-

grams

although

all

instructions

are

stored

in

core

storage.

These

two

programs

are

called

the

process-

ing

program

and

the

data

channel

program.

The

processing

program

is responsible for all arith-

metic, logical,

and

some testing operations.

It

controls

all

operations

that

occur

within

the

central

process-

ing

unit,

plus

the

starting

of

input-output

operations.

The

data

channel

program

is responsible for

the

actual

movement

of

data,

counting

of

words,

and

some

automatic

testing.

The

two

programs

run

concurrently

and

the

central

processing

unit's

program

is

interrupted

only

when

a

data

channel

requires

data

from,

or

has

data

for, core

storage.

The

operation

of a data

channel

is

started

by

cen-

tral

processing

unit

instructions.

Once

started, how-

ever,

the

data

channel

operates

independently

of

the

main

program.

The

data

channel

is

said

to

contain

or

"stack" its

own

instructions.

For

example:

INSTRUCTION

1.

Rewind

tape

unit

1.

2. Backspace

tape

unit

2.

(These

instructions

are

executed

immediately

in

the

central

processing unit.)

3.

Write

tape

unit

3.

1.

The

rewind

instruction

starts

tape

unit

I re-

winding.

(This

may take as

long

as

I-Y2

minutes.)

2.

Two

machine

cycles

after

the

rewind

instruc-

tion

is

decoded

in

the

central

processing

unit,

the

is decoded.

Tape

unit 2 then

begins

its backspace

operation.

(The

time

to

complete

the

backspace

is

a variable

depending

on

the

record

size.)

3. Again,

after

two

machine

cycles,

tape

unit

3

starts

the

writing

operation.

To

summarize,

the

three

instructions

are

executed by

the

central

processing

unit

in

six

machine

cycles.

The

central

processing

unit

is

then

free

to

process

other

data

until

it

is asked for

a storage reference cycle by

the

data

channel.

Instructions

concerned

with a data

channel

are

called

"commands"

instead

of

instructions

in

order

to keep

their

understanding

apart

from

the

central

processing

unit

instructions.

Figure

7 I shows a simple

program

in

which

the

cen-

tral

processing

unit

is

processing

data

while

the

data

channel

is

reading

information

from

a tape.

The

data

channel

program,

made

up

of

commands,

and

the

instructions

for

the

central

processing

unit

are

both

stored

in

core storage. A single

command

may

trans-

mit

a large block

of

words

between

core storage

and

an

1-0

unit.

Normally,

many

instructions

in

the

main

program

may

be

executed

during

the

time

taken

to

execute a single

command

in

the

data

channel.

Main

Program.

The

first

main

program

instruction

selects

the

proper

tape

attached

to

data

channel

A.

The

second

instruction

would

then

load

the

first

command

(Ioep)

into

that

data

channel.

The

remain-

ing

instructions

would

then

start

processing some

other

central

processing

unit

program.

Main Program Instructions

RTDA

------

read tape decimal, channel A.

RCHA------

Reset and load channel A command.

central processing unit instruction!

II

II II

"

Data Channel Program Instructions

10CP

------

1-0

count and proceed.

10CPN-----

1-0

count and proceed, but

do

not transmit.

10RP-------

1-0

record control and proceed.

10CD------

1-0

count and disconnect.

Figure

71.

CPU

and

Data

Channel

Programs

Input-Output

Components

33

Word Count Address

17

3

Figure

72.

Data

Channel Registers

Figure

72

shows

the

format

of

data

channel

com-

mands.

The

operation

part

of

the

command

is con-

tained

in

positions

S,

1,2,

and

19.

The

address

part,

controlling

the

flow

of

data, is

contained

in

positions

21

through

35.

The

count

part

controls

the

number

of

words in-

volved

in

the

operation

and

is located

in

positions

3

through

17.

Position

18

designates

indirect

address-

ing

of

the

command,

in

the

7090 system only. Posi-

tion

20 is

not

used.

(Indirect

addressing

of

commands

is available for the

709

as

an

optional

feature.)

As

can

be

seen

in

Figure 72,

the

count

and

address

parts

are

put

into a word

count

register

and

an

ad-

dress register.

The

operation

parts

are

put

into

an

operation

register.

These

registers are,

of

course,

in

the

data

channel.

Assume

that

the

first

command

(Figure 71) has

an

address

part

of

3000

and a count

part

of

00006.

INPUT-OUTPUT,

COUNT

AND

PROCEED

(IOCP

00006

3000).

This

command

reads

the

first six words from

the

selected

tape

and

places

them

into

consecutive

core locations

starting

with

location 3000.

Each

time

a

word

is moved,

the

count

is

reduced

by one.

When

the

count

(00006) is

reduced

to zero, this command

is completed. Because

it

is a proceed (P) type com-

mand,

the

is

brought

into

the

data

channel

for execution.

The

second

command

is

INPUT-OUTPUT,

COUNT

AND

PRO~EED

IN

NON-TRANSMIT

MODE

(IOCPN

00005 0000).

ThIs

command

results

in

skipping

the

words

on

the

same tape. N designates

that

the

five words in-

volved will be

read

but

not

placed

in

core storage

(thus effectively skipping

them).

Again, because of

the

P, the

is

brought

into

the

data

channel.

The

third

command,

INPUT-OUTPUT

UNDER

RECORD

CONTROL

AND

PROCEED

(IORP

77777 3006), results

in

reading

the

remainder

of

the

words

in

that

record

into

core storage

starting

with

location 3006. Because

the

number

of

words

remaining

in

the

record is

not

known,

the

large

count

(77777)

is

used.

The

sensing

of

the

record

gap

on

the

tape

signals

the

end

of

this

command

and

the

is

brought

into

the

data

channel

(P).

!l4

IBM 709-7090

The

fourth

command,

INPUT-OUTPUT,

COUNT

AND

DISCONNECT

(IOCD

00000 0000), disconnects

or

dis-

engages

the

data

channel,

thus

signaling

the

central

processing

unit

that

the

data

channel

program

is fin-

ished.

A

maximum

of

three

data

synchronizers (six

data

channels) may be used

with

a 709 system.

The

data

channels

of

the

first

data

synchronizer

are

called A

and

B,

the

second two

are C and

D,

and

the

third

two

are E and

F.

Each

data

channel

may

have

as

many

as

eight

IBM

729 I Magnetic

Tape

Units

attached

to it.

Data

chan-

nels A, C,

and

E may also have

one

IBM

711

Card

Reader,

one

IBM

721

Card

Punch,

and

one

IBM

716

Printer

attached

to each channel.

While

all

data

channels

may

be

operated

concurrently

with

comput-

ing, only

one

input-output

unit

per

channel

may

be

in

operation

at

any given time.

A complete

709

Data

Processing System is shown

in

Figure 73.

The

additional

data

synchronizers

are

not

shown

but

would

be

attached

in

the

same

manner

as

the

first.

To

summarize

the

data

synchronizer,

the

word

"channel"

is associated

with

television.

Here,

a tele-

vision

channel

might

be

defined as a

band

of

fre-

quencies over

which

one

station

can

broadcast

sound

. '

pIcture,

and

pulses to synchronize a receiver

with

the

camera.

If

three

television stations

are

broadcasting

simultaneously they

must

use

three

separate channels.

A

data

channel

is a

group

of

components

that

can

transmit

data

between

one

input

or

output

device

and

core storage

and

can

synchronize

the

two.

If

three

1-0

devices

are

to

operate

simultaneously, they

must

use

three

separate channels.

A television receiver

can

receive

only

one

channel

at

a time; core storage

in a computer

can

service only

one

channel

at

a time. A television

transmitter

can

Data

Synchronizer

Unit

Figure 73.

IBM

709

System

broadcast

the

picture

from

only

one

camera

at

a time;

a

data

channel

must

use

only

one

1-0

device

at a time

although

it

is able to

control

several

of

them.

With

as

many

as 48

tape

units

and

three

sets

of

card

machines

controlled

by

the stored

program,

the

709 system

can

perform

many

jobs

"on

line"

that

were

formerly

performed

by auxiliary

equipment.

On

line

means

that

the

job

is

performed

using

the

entire

com-

puter

system

instead

of

using

the "off

line"

auxiliary

equipment.

IBM

755

Tape

Control

An

IBM

755

Tape

Control

(Figure

74)

is

attached

between every

eight

tape

units

and

the

appropriate

data

channel. Several reasons why this

is

done

are:

1.

Tape

signals

must

be

amplified

and

analyzed to

be

usable.

Strong

noise signals

must

be rejected,

and

weak correct signals

must

be accepted.

2.

Only

six

information

bits

can

be

written

on

tape

at

one

time.

The

tape

control separates

the

36-

bit

word

into

six 6-bit characters

and

also gen-

erates a seventh checking

bit

for each character.

The

characters

are

then

delivered to

the

tape

recording head,

one

at

a time.

The

process is

essentially reversed

during

tape

reading.

The

data

transfer is shown

in

Figure

74.

3.

A signal for

the

end

of

record,

end

of

file, weak

or

invalid character,

or

the physical

end

of

tape

is also passed

through

the

tape

control

to

the

data

channel.

Only

the

704

and

709 systems use

tape

controls.

The

IBM

753

Tape

Control

is used

with

the

704

and

the

IBM

755

Tape

Control

is

used

with

the

709 system.

The

problem

of

asynchronous balance is resolved

in

the

7090 system

through a combination

of

two new

units

that

perform

the

simultaneous reading, writing,

and

computing

functions.

These

units

are

the

IBM

7606

Multiplexor

and

the

IBM

7607

Data

Channel.

::,::

:

36

Bits

I

Figure 74.

Data

Transfer

between

Tape

Control,

Tape

Unit,

and

Data

Channel

Figure

75.

IBM

7606 Multiplexor

Multiplexor

The

IBM

7606

Multiplexor

(Figure

75)

accomplishes

all

of

the

data

switching

in

the

7090 system. All

input-

output

components

in

the

system

must

feed

their

data

through

the

multiplexor.

Likewise, any

data

coming

from core storage

must

go first to

the

multiplexor

and

then

to the component.

Data

flow

in

the

7090 system is shown

in

Figure

76.

The

data

flow

path

is from

an

input-output

device to

the

data

channel

to

the

multiplexor

and

then

to core

storage.

Figure 76.

Data

Transmission,

IBM

7090 System

Input-Output

Components

35

The

multiplexor

handles

as

many

as

eight

input-

output

data

channels.

The

data

word is

handled

in

the

normal

fashion

of

36

bits

in

parallel, coming from

or

going to core storage.

Data

Channel

The

IBM

7607

Data

Channel

(Figure 77) is

another

new

unit

used

with

the 7090 system.

Data

channels

replace the

data

synchronizers

and

tape controls used

with the 709 system.

Each

data

channel is basically a completely separate

and

independent

input-output

channel.

It

provides

for

the

transmission

of

data

between the

input-output

device

and

core storage, a transmission

that

is inde-

pendent

of

computing. As

in

the 709 system,

the

stored

program

starts

an

input

operation

by defining

which

input

device is to transmit

and

which core

storage area is to receive.

The

stored

program

can

then

proceed to execute instructions (compute)

not

related to the

input

transmission.

When

a full word

has been assembled

in

the

data

channel, this

word

is

put

into core storage.

An

interruption

of

the compute

program

could

occur only

during

the execution

of

a

convert or,

in

rare

cases, some floating

point

instruc-

tions.

This

interruption

could

be only for 2.18 micro-

seconds.

This

procedure is reversed

if

an

output

oper-

ation

is being performed.

Thus,

one

or

more

data

channels can be occupied

with

an

input

operation

while others are being used

STORAGE

Figure

78.

Data

Channel

and

CPU

Processing,

IBM

7090

36

IBM

709·7090

Figure 77.

IBM

7607

Data

Channel

for

output

purposes. All operations are simultaneous

with the processing

that

is occurring

in

the central

processing unit.

A schematic

of

two

data

channels

operating

asyn-

chronously

with

processing

in

the

central

processing

unit

is shown

in

Figure

78.

Each

data

channel

may

have

as

many

as

ten

inter-

mixed

IBM

729

II

and

729

IV

Magnetic

Tape

Units

in

addition

to a

card

reader,

card

punch,

and a printer.

External

Signal

A

standard

feature

of

the

computer

system

is

its abil-

ity

to accept a signal

from

an

external

source.

This

signal causes

the

computer

to execute a

trapping

operation

when

it

is

received.

The

instruction

being

executed

is

completed

and

the

location

of

the

in

sequence is

placed

in

the

address

part

of

core location 0003.

The

computer

then

takes its

from location 0004.

Direct

Data

Feature

As

an

optional

feature,

the

computer

may be

equipped

with

the

direct

data

feature.

This

allows

the

trans-

mission

of

data

between

the

computer

and

an

external

data

device.

With

this feature,

an

external

signal

initiates

the

trapping

operation

and

the

stored

pro-

gram

may

then

take

whatever

action

is

required

to

introduce

data

into

the

system

or

supply

data

from

the

system to

the

direct

data

device.

Thus

the

main

office

of a company

using

the

com-

puter

could

receive

data

from

branch

offices by

direct

wire.

This

direct

data

path,

as is shown

in

Figure

79,

could

be

an

IBM

65-66

Data

Transceiver

or

anyone

of

many

types

of

transmi

tting

devices.

Figure

79. Schematic,

External

Signal

and

Direct

Data

Features

Magnetic

Drum

Storage

The

magnetic

drum

is

another

storage device used

in

electronic computers.

It

uses

the

same

principle

as mag-

netic tape: A

magnetic

material

can

be

rapidly

mag-

netized

and

de-magnetized,

and

will, like

magnetic

tape,

remain

magnetized

until

deliberately erased.

Magnetic

drum

storage has

the

advantage

that,

once

information

is recorded,

it

is

retained

even

if

the

power

to

the

computer

is

interrupted.

An

important

additional

advantage

possessed by a

magnetic

drum

is

the

ability

to

read

or

alter

stored

information

selectively,

without

reading

or

rewriting

the

entire

contents

of

the

drum.

In

the

case

of

mag-

netic tape,

the

entire

file

must

be

searched

or

re-

written.

This

ability

to

read

or

write

selectively

is

a

valuable

asset

when

large tables

are

stored

and

search

operations

are to be

performed

or

when

information

stored

in

the

table

must

be

modified.

Therefore,

the

drum

is

used

primarily

for

the

storage

of

large blocks

of

such

as

subprograms,

rate

tables,

or

supplementary

data

needed

for

the

solution

of

a problem.

A comparison between

magnetic

tape

and

drum

storage shows

that

tape

is

a long,

thin,

storage device

using

one

recording

head

while

the

drum

is

a short,

wide, storage device using

many

recording

heads.

This

similarity

is

shown

in

Figure

80.

Figure

80. Magnetic

Tape

and

Drum

as Storage Devices

The

magnetic

material

used

on

the

drum

is a wire

made

of

an

alloy

containing

copper, nickel, cobalt,

and

iron.

This

wire is

wound

and

fused to

the

surface

of

the

drum

and

the

surface

is

then

ground

to

an

ex-

tremely fine tolerance,

providing a smooth

even mag-

netic surface.

The

writing

or

recording

process

is

similar to mag-

netic

tape

in

that a coated

surface passes

under

a re-

cording

head.

Information

is

written

on

this mag-

netic

surface by

an

electrical

pulse

that

causes

the

head

to become active.

Whenever

the

head

is

active,

the

area directly

under

it

becomes magnetized

as

shown

in

Figure

81.

This

area

is called a

bit

or

a one.

Thus,

information

is

stored

on

the

surface

of

the

drum

in a pattern

represented

by

combinations

of

magnetized areas.

Input-Output

Components

37

Figure 81.

Writing

on

a

Drum

Figure

82.

Reading

from a

Drum

To

read

the

information,

the

magnetized surface is

passed

under

the

recording head.

This

time,

the

mag-

netized area

on

the

drum

surface generates small volt-

age pulses

in

the

head

as shown

in

Figure

82, exactly

the

reverse

of

the

writing

process.

These

voltage

pulses

are

amplified

and

are

then

available for use

in

the computer.

The

36

recording heads spread across

the

surface

of

one

drum

accommodate

the

word

size

of

the

com-

puter

(Figure 83) .

Recordina---~-""~

Heeds

Drum

Figure

83. Magnetic

Drum

Recording

Heads

A magnetic

drum

unit

has a storage capacity

ot

8,192 words,

each

word

consisting

of

36

bits.

The

drum

unit

contains two distinct physical

drum~

each

with a capacity

of

4,096 words.

Each

of

the

physical

drums

contains two logical

drums

with

a capacity

of

2,048 words each. A logical

drum

is

selected by giving

the

appropriate

address.

The

2,048 locations

on

each logical

drum

can

be

individually addressed by integers

in

the

range

0000-

2047 (0000-3777

octal).

A record

or

block

of

words

Figure

84.

Drum

Sectors

and

Addresses

!l8

IBM

709-7090

Figure

85.

IBM

733

Magnetic

Drum

is

normally stored

in

sequentially

numbered

locations.

The

program

must

indicate the

drum

address where

the first

word

is

to

be

written

or

read.

The

number

of

copy instructions executed

then

determines

the

number

of

words

in

the

record.

Figure

84

illustrates

the

physical

arrangement

of

words

on

a logical

drum.

The

addresses

are

num-

bered

octally. Observe that,

when

reading

or

writing

a

continuous

record,

the

computer

refers to every

eighth

word

of a drum

for consecutive addresses.

Each

logical

drum

has 256 sectors (a

group

of

eight

loca-

tions)

and

therefore

must

make

eight

complete revo-

lutions

for all 2,048 words to be

read

or

written

as

a

continuous

record.

In

summary,

the

drum

is a high-volume, moderate-

speed, long-term

data

storage

medium

for use

with

both

the 704

and

the

709 systems.

It

is

particularly

useful for the storage

of

large blocks

of

information

such

as

subroutines,

rate

tables,

or

supplementary

data

needed for

the

solution

of

a problem.

The

data

stored

on

the

drum

are

not

destroyed

when

power

is re-

moved.

This

makes

the

drum

valuable

for

overnight

storage

of

intermediate

results

during

the

processing

of involved problems.

The

IBM

733

Magnetic

Drum

(Figure 85)

is

one

of

two

input-output

units

in

the

709 system

that

function

without

being

dependent

upon

the

data

synchronizer.

The

other

unit

is

the

IBM

740

CRT

Output

Recorder

and

Display

Unit.

The

7090 system does

not

use

the

magnetic

drum,

or

any

of

the

CRT

equipment.

Punched Cards

In

most applications,

magnetic

tape

is used

as

the

principal

input

medium.

It

may also

be

desirable to

use

IBM

cards as

an

input

medium

in

some

situations

where

the

volume

of

input

is sufficiently small to per-

mit

an

economical

operation.

In

either

case,

IBM

cards

are

used for

initially

recording

data

because of

their

great

flexibility.

Errors

are

easily

detected

and

cor-

rected,

input

data

may be readily

prepared

on

several

card

punch

locations

simultaneously

and

the

cards

may

be

collected

at a central

location

before

entry

into

the

computer.

Entering a program

on

cards may be

done

in

such

a way

that

instructions

are

punched,

one

to

a card,

in

the

most desirable

form

to

the

programmer.

The

computer

can

then

be

supplied

with a standard

pro-

gram

to

assemble

the

instructions

in

the

desired order.

Then,

if

errors are

detected

or

if

changes

must

be

made,

the

error

cards

are

removed,

the

correct ones

are

added,

and

the

computer

prepares

the

new pro-

gram.

Note

that

there

is

no

need

to

repunch

all cards

but

only

the

cards

in

error.

0

2345

67139A'BCD

F

IJjL

I:IOPQRS

'U

nil

I

y:+-

I

...

I

II

I

10

0

j

" "

'

..

1 1

1

22

2

333

33

4444

444

44

55555

5555

555

666666

66666

6666

7777777

777777

77777

88888888

8888888

88888881

888888

999999999

99999999

9999999

999999999999

12345611'WII1213"'516'7aI9m~u~N~ana~~3In~~»~v~~*~~U"u*~*

Figure

86.

Standard

IBM

Card

Code

Punched

card

input

and

output

may

represent

any

alphabetic

character, special symbol,

or

binary

punch-

ing

if

the

assembly

program

handling

it

is

able to rec-

ognize

the

code used.

The

standard

IBM

card

code is

shown

in

Figure 86.

Only

72

card

columns

of

the

IBM

card

can

be

read

into

or

punched

from

core storage

during

any

given

storage reference cycle.

Any

72

of

the

possible 80

card

columns

of

the

card

can

be

selected

through

use

of

a

control

panel

on

the

card

reader

and

the

proper

wir-

ing

on

that

panel.

Figure

87 shows

how

the

card

may

be

punched

to record

data

in

binary

form.

In

this

particular

example,

the

first 72

columns

of

the

card

are

used.

Each

row

is

split

into

half-rows

of

36

card

columns, each

corresponding

to

the

36-bit

binary

word.

Thus,

the

half-row identified

by

the

circled 9

is

named

the

5-row left. Similarly,

the

row

identified

by

the

circled 10 is

named

the

5-row

right.

I

~

,---------left

Hall-------

.....

+!

....

-------

Right

Hall

______

-!

•

..J!-4._~ni"ed

-i

o um"s

®

@

12-Raw

®

@

ll-Row

0000000000000000

O@OO

0 0 0 0 0 0 0 0

0000000

00000000000000000

®O

0 0

00000000000000

00000000

7374757&77781980

11111111111111111@11111111111111111 11111111111111111@11111111111111111

111

II 1 II

222222222

n 2 2 2 2 2

2@>2

2 2 2 2 2 2 2 2 2 2 2 2 2

222

2222222222222222

2@2

2 2 2 2 2 2

2222222222

22222222

33333333333333333@33333333333333333

3333333333333333

3@3

3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3

33333333

4444444444444444

4@4

4 4 4 4 4 4 4 4 4 4 4 4

4444

4444444444444444

4@4

4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

44444444

55555555555555555055555555555555555

5555555555555555

5@5

5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5

55555555

66666666666666666066666666666666666

66666666666666666@66

6 6 6 6 6

6666666666

66666666

7777777777777777

7@7

7 7 7 7 7 7 7 7

77777777

777777777

7 7 7 7 7 7 7

70777777

7 7 7 7 7 7

77

7 7 7

77777777

8888888888888888

8@888

8 8 8 8 8 8 8

8888888

88888888888888888088888888888888888

88888888

9 9

9.9

9 9 9 9 9 9 9 9 9 9 9 9 9 0 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9

999999999999999

9

9~9

9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9

99999999

123456789mnUDM~mnmw~n22nN~Hvna~M»nM~H

31383940414743444541474849505152535 5565758

59

60

6162

6364

65

6667 6169

70

7112

1374157617

78

7980

Figure

87. Sequence

of

Card

Reading

Input-Output

Components

39

Thus,

there

are

24 half-rows

in

the

card.

One

full

word

of

binary

data

(including

the

sign)

can

be

punched

in

any

half-row.

The

card

reader

regards

the

punched

hole

as a

binary

1.

An

unpunched

hole

indi-

cates a

binary

O.

For

example, a

punched

hole

in

the

8-row,

card

column

36, is

regarded

by

the

reader

as a

binary I in

the

least significant

position

of

the

word

in

the

8-row left.

The

left-most

position

of

each

half

row

is reserved for

the

sign

bit

of

that

word

(columns

1

and

37). A binary

1 represents a negative sign while

a

binary 0 (no

punch)

represents a positive sign.

When

reading

or

punching

cards, a record is de-

fined as

the

information

contained

in

one

card. A

file consists of

any

number

of

cards

and

takes

the

form

of

a deck.

Note

that

the

definitions

of

records

and

files differ

depending

on

the

particular

input-output

component

being

discussed.

Card Reader

The

IBM

711

Card

Reader

(Figure 88) is

used

by

the

computer

as

the

card

input

device.

This

reader

is

available

in

two models.

The

model

I reads cards

at

the

rate

of

150

per

minute

while

the

model

II

reads

250 cards

per

minute.

Data

punched

in

the

cards may be

numerical,

alpha-

betic, binary,

or

any special

character

code.

The

inter-

pretation

of

the

coding

and

reading

format

is con-

trolled

by

the

stored

program

and

the

control

panel

located

on

the

reader.

As a punched

card

proceeds

through

the

reader, a

"card

image"

is

built

up

in

core

storage (Figure

89).

As

additional

cards

are

read,

Figure

88.

IBM

711

Card

Reader

40

IBM

709·7090

,

'.

'..

~

1'1

lLI

'It'

Core

Storage

Figure

89.

Card

Image

in

Core Storage

they too become a

part

of

the

card

images

stored

in

core storage.

The

IBM

709

Data

Processing System

may

have

one

model

II

card

reader

attached

to

data

synchronizer

channel

A,

and

additional

readers may be

attached

to

channels

C

and

E.

If

a

reader

is

attached

to a

data

channel,

an

IBM

716

Printer

must

also

be

attached

to

the

same

channel

because

it

is

the

power

source for

the

reader.

The

IBM

7090

Data

Processing System

may

have

one

model

II

reader

attached

to each

data

channel,

making a total

of

eight

possible readers

on

the

system.

A

printer

is

required

for each reader.

Figure

90.

IBM

721

Card

Punch

Card Punch

The

IBM

721

punching

punches

Basic

of

the

a

card a card, card ure

the

format

The card a

card nel One punch.

The one a

total printer punch.

card

image

the

punch

91). card

of

IBM

punch

must

printer

IBM

card

of

or

cards

reader,

card

The

punch

the

709

attached

also have

7090

punch

eight

is

required

Card

Punch

recording

at a rate

punch

operations

except

in

core storage image is

to

be

recorded

stored

program

control

card

image.

Data

is

attached

an

(Figure 90) is

unit

for

of

100 cards

that

when

sent

as

and

the

coding

Processing System

to

data

channels A,

to a

IBM

716

will accommodate

Data

Processing System

attached

possible

to

card

each

punches

for each

IBM

cards.

per

minute. are analogous instead

from

of

data

are

core storage

punched

the

control

and

the

may

data

channel,

Printer

both

data

attached

the

channel

per

card

reader

used

as

The

to

those

building

read

from

to

holes (Fig-

panel

recording

have

C,

and

E.

that

chan-

to it.

reader

may

have

making

system.

and

card

the

721

up

the

on

one

If

and

One

Figure

9l.

Punching a Card

Card Punch

Image

I/L

1/;

...

I

,

"

I

A'

.Y

vVVVJ

'A'..vv'vvv

fVI/

v'v

Core Storage

Printer

The

IBM

printed

The

printer

(Figure

with a "0"

zone

b"O"

716

Printer

output

93).

hale 6

from

is

equipped

Each type wheel

zone

"11" hole

"7"

holes

(Figure 92) is

the

by

with a "12"

zone

"12" hole

~

used

to

prepare

computer

with

systems.

120

rotary

has

48 characters in-

type wheels

------------

----

"7"

---

hole

itself

~

Character Printed

Figure

92.

---

W

5

E

IBM

716

Printer

8 l R

o

Figure

93.

IBM

716

Printer

Type

Wheel

Zone Porti on

T 2 B

Input-Output

Components

41

I

Printed

I

Data

/

1

.11

I

Data to be

.

..

...

. ' .

.

Printe

y

Printer

IA

lfl/

Figure 94.

Printing

from Core Storage

eluding

numerals, alphabetic symbols,

and

special

characters.

Information

is

printed

at

the

rate

of

150

lines

per

minute

with

as

many

as 120 characters

per

line.

The

print

format

and

arrangement

of

the

data

to be

printed

are

controlled by

the

stored

program

and

a control

panel

located

on

the

printer.

Use

of

the

proper

stored

program

enables the

printing

of

any

desired

information

in

any form convenient to

the

programmer.

Figure 94 shows a

report

as

it

appears

in

core

storage

and

as

it

would

appear

being

printed

on

the

716

Printer.

The

choice

of

which type wheels to use

in

the

actual

printing

would

be

decided by

the

control

panel

wiring

on

the

printer.

The

IBM

709

Data

Processing System uses

one

printer

attached

to

data

channels A, C,

and

E.

The

IBM

7090

Data

Processing System may have a total

of

eight

printers,

one

attached to each

data

channel.

With

all systems,

the

716

Printer

supplies

the

power

for

both

the

card reader

and

the

card

punch

and,

therefore, each set

of

card

equipment

must

be attached

to

one

data

channel.

Cathode

Ray

Tube Equipment

Visual display

of

processed

data

is provided

by'

a

cathode ray

tube

display

unit.

Numbers

are con-

verted

into

alphamerical characters, curves,

or

any

type

of

visual representation.

With

this

unit,

the

actual

results

determined

by

the

mathematical for-

mula

used

in

Figure I could

be

used to display

the

cross section

of

the

wing design (Figure

95).

Associated

with

this display

unit

is a second cathode

ray

tube

with

a camera attached.

This

unit

provides

a

permanent

filmed recording

of

the

displayed

data

so

that

different designs can be

compared

and

evaluated.

When

a character is to be displayed, only

35

posi-

tions

of

the

word

containing

the

character are used.

The

sign position is considered positive.

The

remain-

ing

bits signify a display

or

no-display

of

a spot, de-

42

IBM

709·7090

Figure 95. Visual Display

of

Wing

Design

pending

on

whether

the

bit

is a I

or a O.

The

first

column

of

the

character is represented by

the

first

group

of

seven bits (following

the

sign

bit);

the

sec-

ond

column, by

the

second

group

of

7 bits,

and

so on.

The

first

bit

of

each

group

may

correspond to

the

lowest position

of

each column; succeeding bits will

move

up

the

column, one

bit

at

a time.

S I 7

14

21

28

35

Figure

96.

CRT Character

Plotting

An

example

of

CRT character

plotting

is shown

in

Figure

96.

The

word

responsible for

the

plot, as

it

appears

in

core storage, is shown above

the

plot.

The

program

consists

of

a write instruction addressing

the

CRT

and

a copy instruction specifying

the

storage

location

of

the

word

containing

the

character to

be

plotted.

The

IBM

740

Cathode

Ray

Tube

Recorder

(Figure

97) is basically a digital-to-analog converter.

It

is con-

nected to

both

the

704

and

709 systems.

The

recorder

contains a seven-inch cathode

ray

tube

as

the

actual

output

device. A camera,

mounted

over

the

face of

the

CRT, produces a filmed recording

of

the

plotted

data.

The

IBM

780

Cathode

Ray

Tube

Display (Figure

98)

is

connected to

the

CRT recorder

and

controlled by

it.

This

unit

contains a 21-inch CRT, similar to a

home

television receiver tube, to provide

an

immediate

dis-

play

of

the

plot

being

filmed by

the

740

unit.

The

CRT

display is,

in

effect, a "slave

unit"

to

the

CRT

re-

corder.

The

conversion from digital to analog data,

Figure

97.

IBM

740 CRT

Output

Recorder

the

display

of

that

data

on

the

faces

of

the

CRT'S,

and

the

operation

of

the

camera

are

all

accomplished

at

electronic speeds

under

control

of

the

stored

program.

The

7090 system does

not

use

either

the

CRT

re-

corder

or

the

CRT

display units.

Figure

99 shows comparative statistics

of

the

709

and

7090 systems.

Refer

to

the

Reference

Manuals

(IBM

709,

Form

A22-6536,

or

IBM

7090,

Form

A22-6528)

for

the

operational

details

concerning

the

systems.

Figure

98.

IBM

780 CRT Display

709

7090

Core Storage Size

4096, 8192 or

32,768

words

32,768

words

T ransi stari

zed

no

yes

Internal Speed (Basic Cycle) 12usec

2.18

usee

Simul

taneous Read-Write-Compute yes

yes

Tape Skip Abil ity yes yes

Automatic

Input-Output Priority yes yes

Data Channel Trapping

y~s

yes

Number of Instructions

208

220

Maximum Number of:

Tape Units 48

80

1-0

Data Channels

2,4

or 6

1 thru 8

Card Readers 3

8

Card

Punches 3

8

Printers 3

8

Magneti c

Drums

2

none

CRT

Recorder

and

Display 1

none

Figure

99. Comparison of

the

IBM

709

and

7090 Systems

Input-Output

Components

43

Share

Organization

The

present Share organization was conceived

when

three customers, involved

in

preparation

for a 704

Data

Processing System, began

informal

discussions

concerning

their

individual

plans

and

problems.

The

mutual

respect

of

the

participants

for

the

program-

ming

competence

of

each

other

was expressed; they

were willing

to

accept

the

ideas

of

others, even to

the

extent

of

obsoleting methods already

prepared

within

their

own

installations.

It

was agreed

that

a full-scale

attempt

should

be

made

to

bring

Share

into

being.

The

first meeting was

held

in

1955

with

seventeen

members present.

One

of

the

advantages

of

an

organization

of

this

type is

that

each

member

is closely

united

with the

development

of

computer

usage

throughout

the world.

A substantial

portion

of

the

major

users

of

high-

speed digital computers is represented

in

the

Share

membership.

It

has been

found

that

critical evalua-

tion

of

each other's ideas usually produces a distilla-

tion

of

thoughts, superior to any

individual

opinions.

Members realize substantial savings

in

programming

and

check-out

of

programs.

The

continual

inter-

change

of

ideas among members has demonstrated

that

a

high

degree

of

computing

sophistication is rapidly

built

up

in

a Share installation.

In

its

initial

phase, Share was concerned primarily

with

procedures

and

standards. Later,

the

prepara-

tion

and

distribution

of

programs was started.

No

slacking

of

activity is

anticipated

since,

as

the

most

essential programs are completed, emphasis shifts to

new areas

of

mathematical computation.

Programming

System

The

Share 709 system is used

with

the

7090

Data

Processing System as

the

main

programming

system.

It

enables

the

programmers to write, check-out,

and

alter

their

programs quickly

and

easily.

The

Share 709

System (sos) can

be

thought

of

as composed

of

four

distinct parts:

1.

The

Share compiler assembler

translator

(SCAT).

2.

The

program

testing

and

correcting system.

3.

The

input-output

system.

4.

The

MockDonald control system.

44

IBM 709-7090

The

SCAT

portion

of

the

sos consists

of

two parts,

the

compiler

and

the

modify-and-Ioad program.

The

compiler performs

about

the

same functions

for

the

709

and

7090 systems

that

the

Share assembly

program

(SAP)

performs for

the

704 system.

With

a

few

minor

exceptions, a Share assembly language pro-

gram

is acceptable as

input

for

the

compiler

and

re-

sults

in a program

listing

and

an

absolute

binary

pro-

gram

card

deck.

In

addition

to

the

results produced

by Share assembly program,

the

compiler

can

produce

a "squoze"

card

deck.

This

deck contains

the

sym-

bolic source

program

in

encoded

binary

form.

This

form may

be

converted

to

machine language

and

loaded

by

the

modify

and

load

program

almost as

rapidly

as

an

ordinary

binary

load

program

loads

an

absolute

binary

card

deck.

Two

main

reasons for the

intermediate

squoze

card

deck phase are:

1.

Modifications to

the

program

can

be

made

in

the

original SCAT language

and

then

added

to the

squoze deck for

loading

by

the

modify-and-Ioad

program.

2.

Enough

of

the

original symbolic

information

can

be

retained

during

the

program

execution to per-

mit

the

checking

and

correcting

program

to re-

turn

printed

output

in

the

original symbolic

language.

In

most instances, these two features

make

it

unnecessary to

"patch" a program

in

machine language.

Thus,

nearly all cross-ref-

erencing between symbolic

and

binary

may

be

avoided.

Another

powerful tool

in

the

compiler is

the

idea

of

"macro-operation."

The

compiler is

built

to recog-

nize a large fixed

number

of

macro-operations.

It

also

accepts

and

temporarily retains definitions

of

macro-

operations given by

the

programmer.

In

either

case,

it

generates

and

inserts

into

the

program

the

sequence

of

machine

words specified

by

anyone

of

these macro-

operations

in

a macro-instruction.

Among

the

sys-

tem (fixed) macros are all

of

the

pseudo-operations

making

up

the

checking

and

correcting program,

the

input-output,

and

the

MockDonald control system

languages.

A significant feature

of

the

SCAT system is

that

the

loading

process is also

an

assembly process.

The

squoze deck is

not

in

a form

that

can

be

inserted

into

the

computer

as is.

It

is

the

result

of

what

corresponds

to

the

first pass

of

the

704 Share assembly

program

sys-

tem.

With

SAP

a second pass is

needed

to decode

the

assembled deck

into

absolute binary.

In

the

SCAT

sys-

tem,

the

second pass is a function

of

the

modify-and-

load

program.

This

program

provides for

the

same

modification

of

the

original code

that

could

be

ob-

tained

by

changing

the

original

symbolic deck

and

reassembling.

Thus,

in

SCAT,

program

modifications

are

given to

the

loading

program

and,

for these

changes,

the

modify·and-Ioad

program

performs

both

the

first

and

second passes.

In

this sense,

the

loading

program

is a

full

assembly program.

When

program

modifications

are

made

with

the

loader,

it

produces

on

request

a new squoze deck

or

an

absolute

binary

deck,

as

well as

the

listing

of

the

modified program.

In

addition

to

the

above operations,

the

modify-and-

load

program

performs checking

operations

for

both

programmer

and

machine

errors.

The

input-output

system

permits

writing

of

1-0

pro-

grams designed for a

particular

customer's application.

Transmission

macro-instructions are used

and

are

executed by a

routine

called

the

dispatcher.

These

macros

provide

simultaneous

input-output

with

com-

puting.

Computing

is interlocked

with

data

transmis-

sion so

that

the

computer

will

not

attempt

to use

or

modify

data

to

be completed.

Transmission

orders

are

channel-stacked

when

required

and

subsequently

the

dispatching

of these orders

on

data

channels

is

au-

tomatic

when

the

channel

is

free.

The

programmer

may

interrogate

the

dispatcher

for

the

status of

any

transmission

at

any

point

in

the

program.

The

check-

ing

of

input-output

indicators

is

automatically

ac-

complished by

the

dispatcher.

The

MockDonald

control

system has

been

designed

to

enable

the

automatic

transition

from

one

problem

to

the

next,

to

maintain a machine

program

log, to

aid

in

the

parallel

operation

of

tapes

and

main

frame

processing

and,

in

general, to

perform

many

of

the

op-

erations

that

would

normally

be

handled

by a profes-

sional

machine

operator.

Share

45

Fortran

Automatic

Coding System

The

IBM·

Mathematical

FORmula TRANslating System,

709

FORTRAN,

is

an

automatic

coding

system

for

the

IBM

709

Data

Processing System.

More

precisely,

it

is

a

709

program

that

accepts a source program

written

in

the

Fortran

language, closely resembling

the

ordi-

nary

language

of

mathematics,

and

produces a ma-

chine

language

object program

ready

to

be

run

on

a

709. The

709

Fortran

therefore,

in

effect, transforms

the

IBM

709

into a machine

with

which

communication

can

be

made

in a language

more

concise

and

more

familiar

than

the

709

machine

language

itself.

The

result

is a

substantial

reduction

in

the

training

re-

quired

to

program,

as well as

in

the

time

consumed

in

writing

programs

and

eliminating

errors from them.

Among

the

features

which

characterize

the

709

Fortran

system are

the

following:

The

709

Fortran

is available

in

versions for all sizes

of

storage. Each version

produces

programs

which

can

be used

on

any

size

of

709,

provided

sufficient stor-

age is available for

the

object

program.

Object

pro-

grams

that

are

too large

for

the

709

on

which

they

are

to

be

used

must

be

subdivided

by

the

user.

Object

programs

produced

by

Fortran

will gener-

ally be as efficient as those

written

by

experienced

programmers.

The

Fortran

language

is

intended

to

provide

facili-

ties for expressing

any

problem

of

numerical

compu-

tation.

In

particular,

problems

containing

large sets

of

formulas

and

many

variables

can

be

dealt

with

46 IBM 709·7090

easily,

and

any

variable

may

have

up

to

three

inde·

pendent

subscripts.

The

language

of

Fortran

may

be

expanded

by

the

use

of

subprograms.

These

subprograms

may

be

writ-

ten

in

Fortran

language,

and

may

be

called

by

other

Fortran

programs, as well as subprograms.

The

lan-

guage

may

be

expanded

by

the

use

of

subprograms

to

any

desired

depth.

For

problems

in

which

machine

words

have

a logi-

cal

rather

than a numerical

meaning,

the

language is

less satisfactory,

and

difficulties

may

arise

in

express-

ing

such problems. Nevertheless,

many

logical opera-

tions

not

directly expressible

in

the

Fortran

language

can

be

carried

out

by

making

use

of

the

provisions

for

incorporating

library

routines.

Prewritten

routines, to

evaluate

functions

of

any

number

of

arguments,

can

be

made

available

for

in-

corporation

into

object

programs

by

the

use

of

any

of

several different facilities

provided

for

this purpose.

Certain

statements

in

the

Fortran

language

cause

the

inclusion

in

the

object

program

of

the

necessary

input

and

output

routines.

Those

which

deal

with

decimal

information

include

conversion to

or

from

the

internal

machine

language,

and

permit

consider-

able

freedom

of

format

in

data

input

and

output.

Arithmetic

in

an

object

program

will generally

be

performed

with

single-precision floating

point

num-

bers.

These

numbers

provide

about

8 decimal digits

of

precision,

and

may

be

zero

or

have

magnitudes

between

approximately

10-

38

and

10

38

•

Fixed

point

arithmetic

for integers is also

provided.

CODE

ACL ADD

ADM

ALS ANA ANS

ARS

AXC AXT BSFA

BSFB BSFC BSFD BSFE BSFF

uBSFG

··BSFH BSR BSRA

BSRB

BSRC BSRD BSRE BSRF

··BSRG

··BSRH BTTA BTTB BTTC BTTD BTTE BTTF

··BTTG

··BTTH

·CAD CAL CAQ CAS CFF CHS

CLA

CLM

CLS

COM

·CPY CRQ CVR DCT DVH DVP ENK ERA

ETM ETTA ETTB

ETTC ETTD ETTE ETTF

uETTG

··ETTH FAD

COMMENT

Add

and

Carry

Logical

Word

................ X

Add

........................................................... X

Add

Magnitude

......................................... X

Accumulator

Left

Shift

............................ X

AND

to

Accumulator

................................ X

AND

to

Storage .......................................... X

Accumulator

Right

Shift

........................ X

Address

to

Index,

Complemented

......... .

Address to

Index,

True

........................... .

Backspace File,

Ch.

A ................................ X

Backspace File, Ch. B ................................ X

Backspace File,

Ch.

C ................................ X

Backspace File, Ch. D ................................ X

Backspace File,

Ch.

E ................................ X

Backspace File,

Ch.

F ................................ X

Backspace File,

Ch.

G .............................. X

Backspace File,

Ch.

H .............................. X

Backspace

Record

................................ X

Backspace

Record,

Ch.

A ........................ X

Backspace

Record,

Ch. B ........................ X

Backspace

Record,

Ch. C ........................ X

Backspace

Record,

Ch. D ........................ X

Backspace

Record,

Ch.

E .......... .............. X

Backspace

Record,

Ch.

F ........................ X

Backspace

Record,

Ch. G ............ ............ X

Backspace

Record,

Ch.

H .......... .............. X

Beginning

of

Tape

Test,

Ch.

A .............. X

Beginning

of

Tape

Test,

Ch.

B .............. X

Beginning

of

Tape

Test,

Ch.

C .............. X

Beginning

of

Tape

Test,

Ch.

D .............. X

Beginning

of

Tape

Test,

Ch.

E .............. X

Beginning

of

Tape

Test,

Ch.

F .............. X

Beginning

of

Tape

Test,

Ch.

G .............. X

Beginning

of

Tape

Test,

Ch.

H .............. X

Copy

and

Add

Logical

Word

.................. X

Clear

and

Add

Logical

Word

................ X

Convert

by

Addition

from

MQ

............

..

Compare

Accumulator

with

Storage

....

X

Change

Film

Frame

.................................. X

Change

Sign ................................... X

Clear

and

Add

...................... ...... ................ X

Clear

Magnitude

........................................ X

Clear

and

Subtract

.................................... X

Complement

Magnitude

.......................... X

Copy

..................................... X

Convert

by

Replacement

from

MQ

....

..

Convert

by

Replacement

from

AC

......

..

Divide

Check

Test

...................................... X

Divide

or

Halt

............................................ X

Divide

or

Proceed

...................................... X

Enter

Keys ................................................ X

Exclusive

OR

to

Accumulator

................ X

Enter

Trapping

Mode

.............................. X

End

of

Tape

Test,

Ch.

A ........................ X

End

of

Tape

Test,

Ch.

B ........................ X

End

of

Tape

Test,

Ch. C ........................ X

End

of

Tape

Test,

Ch. D ........................ X

End

of

Tape

Test,

Ch. E ........................ X

End

of

Tape

Test,

Ch.

F ....................... X

End

of

Tape

Test,

Ch.

G ........................ X

End

of

Tape

Test,

Ch.

H ........................ X

Floating

Add

................ ........................ ...... X

-709

system

instruction

only.

u7090

system

instruction

only.

X

X X

X

X X

X

CODE

FAM FDH FDP FMP FOR

FRN FSB FSM

FVE

HPR HTR IIA

ilL

IIR

liS

IOCD IOCDN

IOCP IOCPN

IOCT IOCTN

IORP IORPN IORT

IORTN IOSP IOSPN lOST IOSTN lOT

LAC

LAS LBT

LCHA

LCHB

LCHC

LCHD

LCHE

LCHF

··LCHG

·-LCHH

-LDA LDC

LDI

LDQ LFT LGL

LGR

LLS LNT

LRS

LTM

LXA

LXD MON MPR MPY MSE

SCAT

Mnemonic

Operation

Codes

COMMENT

Floating

Add

Magnitude

.......................... X X

Floating

Divide

or

Halt

.......................... X X

Floating

Divide

or

Proceed .................... X X

Floating

Multiply

...................................... X X

Four

............................................................. .

Floating

Round

.......................................... X

Floating

Subtract

........

..

........................ X X

Floating

Subtract

Magnitude...

..

.......... X X

Five

....

..

.................................... .

Halt

and

Proceed

.................................... .

Halt

and

Transfer

.................................... X X

Invert

Indicators

from

Accumulator

....

..

Invert

Indicators

of

the

Left

Half

......

..

Invert

Indicators

of

the

Right

Half

....

..

Invert

Indicators

from

Storage .............. X X

Input-Output

under

Count

Control

and

Disconnect ..............................................

..

(IOCD

with

No

Transmission)

............

..

Input-Output

under

Count

Control

and

Proceed

....

..

..................

..

(IOCP

with

No

Transmission)

............

..

Input-Output

under

Count

Control

and

Transfer

..................................................

..

(IOCT

with

No

Transmission)

..........

..

Input-Output

of a Record

and

Proceed

(IORP

with

No

Transmission)

Input-Output

of a Record

and

Transfer

(IORT

with

No

Transmission)

..........

..

Input-Output

until

Signal

and

Proceed

(IOSP

with

No

Transmission)

..........

..

Input-Output

until

Signal

and

Transfer

(lOST

with

No

Transmission)

............

..

Input-Output

Check

Test

........................ X

Load

Complement

of

Address

in

Index

..

Logical

Compare

Accumulator

with

Storage ........................................ X X

Low-Order

Bit

Test

.................................. X

Load

Channel

A ...................................... X X

Load

Channel

B ................................. X X

Load

Channel

C ...................................... X X

Load

Channel

D ........................................ X X

Load

Channel

E ........................................ X X

Load

Channel

F ...................................... X X

Load

Channel

G ........................................ X

Load

Channel

H ........................................ X

Locate

Drum

Address ......... ............ ......... X X

Load

Complement

of

Decrement

in

XR

Load

Indicators

.......................................... X X

Load

the

MQ

.............................................. X X

Left

Half

Indicators,

Off

Test

..............

..

Logical

Left

Shift

...................................... X

Logical

Right

Shift

.................................... X

Long

Left

Shift

...................... .................... X

Left

Half

Indicators,

On

Test

..............

..

Long

Right

Shift

................. ....................... X

Leave

Trapping

Mode

.............................. X

Load

Index

from

Address ......................

..

Load

Index

from

Decrement

................

..

Minus

One

..................................................

..

MUltiply

and

Round

................................ X X

Multiply

...................................................... X X

Minus

Sense .................................................. X

Appendix

47

SCAT

Mnemonic

Operation

Codes

(Con~d)

CODE

ORA ORS

OSI

PAC PAl PAX

PBT

PDC

PDX

PIA

PON PSE PTH

PTW

PXA

PXD PZE RCDA

""RCDB

RCDC

"RCDD

RCDE

""RCDF

""RCDG URCDH

RCHA

RCHB

RCHC

RCHD

RCHE

RCHF

··RCHG uRCHH

""RDCA

RDCB RDCC RDCD RDCE

RDCF

RDCG

"·RDCH

·RDR RDS REWA REWB R£WC REWD REWE REWF

uREWG UREWH

RIT

RIA

RIL RIR RIS RND RNT RPRA

·"RPRB RPRC

uRPRD

RPRE

URPRF

··RPRG

··RPRH RQL RTBA RTBB

COMMENT

OR

to

Accumulator

.................................. X X

OR

to

Storage ............................................ X X

OR

Storage

to

Indicators

........................ X X

Place

Complement

of

Address

in

XR

...

.

Place

Accumulator

in

Indicators

......... .

Place Address ill

Index

........................... .

P-Bit

Test

.................................................... X

Place

Complement

of

Decrement

in

XR

Place

Decrement

in

Index

....................... .

Place

Indicators

in

Accumulator

........... .

Plus

One

..................................................... .

Plus

Sense .................................................... X

Plus

Three

................................................... .

Plus

Two

..................................................... .

Place

Index

in

Address ........................... .

Place

Index

in

Decrement

....................... .

Plus Zero

...

. ........................................ .

Read

Card

Reader,

Ch. A ........................ X

Read

Card

Reader,

Ch. B ........................ X

Read

Card

Reader,

Ch. C ........................ X

Read

Card

Reader,

Ch.

D ........................ X

Read

Card

Reader,

Ch. E ........................ X

Read

Card

Reader,

Ch. F ........................ X

Read

Card

Reader,

Ch.

G ........................ X

Read

Card

Reader,

Ch. H ........................ X

Reset

and

Load, Ch. A .......................... X X

Reset

and

Load,

Ch. B............................ X X

Reset

and

Load, Ch. C.......... .................. X X

Reset

and

Load,

Ch.

D ............................ X X

Reset

and

Load,

Ch.

E....

...... ......

........

....

X X

Reset

and

Load,

Ch.

F ............................ X X

Reset

and

Load,

Ch.

G ............................ X X

Reset

and

Load,

Ch.

H ............................ X X

Reset

Data

Channel A ..... .....

......

...... X

Reset

Data

Channel

B ...... ,

...

...... ............ X

Reset

Data

Channel

C ............................ X

Reset

Data

Channel

D ............................ X

Reset

Data

Channel

E....

...... ...... ............ X

Reset

Data

Channel

F....

...... ...... ............ X

Reset

Data

Channel

G ............................ X

Reset

Data

Channel

H ............................ X

Read

Drum

.................................................. X

Read

Select ................................................ X

Rewind,

Ch. A ............................................ X

Rewind,

Ch. B ............................................ X

Rewind,

Ch. C ............................................ X

Rewind,

Ch. D ............................................ X

Rewind,

Ch. E ............................................ X

Rewind,

Ch. F ............................................ X

Rewind,

Ch. G ............................................ X

Rewind,

Ch. H ............................................ X

Right

Half

Indicators, Off

Test

............. .

Reset

Indicators

from

the

Accumulator

Reset Indicators

of

the

Left

Half

......... .

Reset

Indicators

of

the

Right

Half

....... .

Reset

Indicators

from Storage ................ X X

Round

............................................................ X

Right

Half

Indicators,

On

Test

............. .

Read

Printer,

Ch.

A .................................. X

Read

Printer,

Ch.

B .................................. X

Read

Printer,

Ch.

C .................................. X

Read

Printer,

Ch.

D .................................. X

Read

Printer,

Ch.

E .................................. X

Read

Printer,

Ch.

F .................................. X

Read

Printer,

Ch.

G .................................. X

Read

Printer,

Ch.

H .................................. X

Rotate

MQ

Left

........................................ X

Read

Tape

Binary,

Ch.

A ........................ X

Read

Tape

Binary, Ch. B ........................ X

·709

system

instruction

only.

u7090

system

instruction

only.

48

IBM

709-7090

CODE

RTBC RTBD RTBE

RTBF uRTBG uRTBH

RTDA

RTDB

RTDC

RTDD

RTDE

RTDF

·"RTDG "RTDH

RUNA RUNB RUNC RUND

RUNE

RUNF "RUNG "RUNII

SBM

SCHA

SCHB

SCHC

SCHD

SCHE

SCHF

"·SCHG "SCHH

SDHA SDHB

SDHC

SDHD SDHE SDHF

·SDHG

·SDHH

SDLA SDLB SDLC SDLD SDLE SDLF

"SDLG

·SDLH

SIL SIR SIX SLF SLN SLQ SLT SLW

SPRA

"SPRB

SPRC

""SPRD

SPRE

··SPRF "SPRG "SPRH

SPTA

"SPTB

SPTC

"SPTD

SPTE "SPTF "SPTG

r.I

-1

"

<

~

r.I

co

COMMENT

~

Read

Tape

Binary,

Ch.

C ........................ X

Read

Tape

Binary,

Ch.

D ........................ X

Read

Tape

Binary,

Ch.

E ........................ X

Read

Tape

Binary,

Ch.

F ........................ X

Read

Tape

Binary,

Ch.

G ........................ X

Read

Tape

Binary,

Ch.

H ........................ X

Read

Tape

Decimal, Ch. A ...................... X

Read

Tape

Decimal, Ch. B................

......

X

Read

Tape

Decimal, Ch. C...................... X

Read

Tape

Decimal, Ch. D ...................... X

Read

Tape

Decimal, Ch. E...................... X

Read

Tape

Decimal, Ch.

I'

.................... X

Read

Tape

Decimal,

Ch.

G...................... X

Read

Tape

Decimal,

Ch.

H...................... X

Rewind

and

Unload,

Channel

A X

Rewind

and

Unload,

Channel

B X

Rewind

and

Unload,

Channel

C X

Rewind

and

Unload,

Channel

D X

Rewind

and

Unload,

Channel

E X

Rewind

and

Unload,

Channel

F X

Rewind

and

Unload,

Channel

G X

Rewind

and

Unload,

Channel

H X

Subtract

Magnitude

...

............. ....... ....... ...... X

Store, Ch.

A..........

............................ .......... X

Store,

Ch.

B .................................... ,

...

........ X

Store, Ch. C ............ '

...

.................. .............. X

Store, Ch. D ...... ,

...

............ ............ .............. X

Store, Ch. E

...

..................

.....

......... X

Store, Ch.

I'

................................... X

Store, Ch. G ....... ...................... ....... X

Store, Ch. H ...... ................... .............. X

Set Density

High.

Channel

A....

X

Set Density

High,

Channel

B...

)(

Set

Density

High,

Channel

C X

Set Density

High,

Channel

D X

Set Density

High,

Channel

E X

Set Density

High,

Channel

F .............. X

Set Density

High,

Channel

G X

Set Density

High.

Channel

H X

Set Density Low,

Channel

A X

Set Density Low,

Channel

B X

Set Density Low.

Channel

C X

Set Density Low,

Channel

D X

Set Density Low,

Channel

E X

Set Density Low,

Channel

F X

Set Density Low,

Channel

G X

Set Density Low,

Channel

H X

Set Indicators

of

the

Left

Half

.,

........... .

Set

Indicators

of

the

Right

Half

........... .

Six

.....

. .........................................

..

Sense Lights Off

....

......

............ .............. X

Sense Lights

On

.......................................... X

Store

Left

Half

MQ

.................................. X

Sense

Light

Test

.....

......

...... ...... ............ X

Store Logical

Word

.................................... X

Sense

Printer,

Ch. A .................................. X

Sense

Printer,

Ch.

B .................................. X

Sense

Printer.

Ch.

C .................................. X

Sense

Printer,

Ch.

D .................................. X

Sense

Printer,

Ch. E .................................. X

Sense

Printer,

Ch. F.................................. X

Sense

Printer,

Ch. G.......... ........................ X

Sense

Printer,

Ch.

H .................................. X

Sense

Printer

Test,

Ch.

A ........................ X

Sense

Printer

Test,

Ch. B........................ X

Sense

Printer

Test,

Ch. C ........................ X

Sense

Printer

Test,

Ch. D ........................ X

Sense

Printer

Test,

Ch. E ........................ X

Sense

Printer

Test,

Ch. F ........................ X

Sense

Printer

Test,

Ch. G ........................ X

i

I:i

~

X

X X

X X X X

X

(lODE

"SPTH

SPUA

"SPUB

SPUC

"SPUD

SPUE "SPUF "SPUG "SPUH

SSM

SSP

STA

STD

STI

STL

STO

STP

STQ

STR

STT

STZ SUB SVN SWT

SXA SXD

TCH TCNA

TCNB TCNC TCND

TCNE

TCNF

"TCNG

--TCNH TCOA TCOB TCOC TCOD TCOE TCOF

--TCOG

"TCOH

TEFA TEFB TEFC TEFD TEFE TEFF

-"TEFG

·"TEFH TIF

TIO

TIX TLQ TMI TNO TNX TNZ TOV TPL TQP

TRA

TRCA TRCB TRCC TRCD TRCE TRCF

"TRCG

··TRCH

COMMENl

Sense

Printer

Test,

Ch.

H ........................ X

Sense

Punch,

Ch.

A .................................. X

Sense

Punch,

Ch. B

....

...

..... .......... ..... ....... X

Sense

Punch,

Ch.

C .................................. X

Sense

Punch,

Ch.

D .................................. X

Sense

Punch,

Ch. E ..... .........

....

............

....

X

Sense

Punch,

Ch.

F . ...... ............... ....... ..... X

Sense

Punch,

Ch.

G .................................. X

Sense

Punch,

Ch.

H . ....... ...................... .... X

Set Sign

Minus

...................................... X

Set Sign

Plus

................................... X

Store

Address ....... .............

.....

...... X

Store

Decrement

.................................. X

Store

Indicators

................................. X

Store

Instruction

Location

Counter

...... X

Store .............................................. X

Store Prefix ..................................... X

Store

MQ

...................................... X

Store

Location

and

Trap

......................... .

Store

Tag

...................... X

Store Zero ....................................... X

Subtract

........................................................ X

Seven ................................................ .

Sense Switch

Test

...................................... X

Store

Index

in

Address ............................. .

Store

Index

in

Decrement

....................... .

Transfer

in

Channel

................................. .

Transfer

on

Ch. A Not

in

Operation

....

X

Transfer

on

Ch. B

Not

in

Operation

....

X

Transfer

on

Ch. C Not

in

Operation

....

X

Transfer

on

Ch. D Not

in

Operation

....

X

Transfer

on

Ch. E Not

in

Operation

....

X

Transfer

on

Ch. F Not

in

Operation

....

X

Transfer

on

Ch. G Not

in

Operation

....

X

Transfer

on

Ch. H Not

in

Operation

....

X

Transfer

on

Ch. A in

Operation

.........

...

X

Transfer

on

Ch. B in

Operation

............ X

Transfer

on

Ch. C in

Operation

............ X

Transfer

on

Ch. D in

Operation

............ X

Transfer

on

Ch. E in

Operation

............ X

Transfer

on

Ch. F

in

Operation

............ X

Transfer

on

Ch. G in

Operation

.....

...

....

X

Transfer

on

Ch. H in

Operation

........

....

X

Transfer

on

End

of

File,

Ch.

A X

Transfer

on

End

of

File,

Ch.

B X

Transfer

on

End

of

File,

Ch.

C X

Transfer

on

End

of

File, Ch. D X

Transfer

on

End

of

File,

Ch.

E X

Transfer

on

End

of

File, Ch. F X

Transfer

on

End

of

File,

Ch.

G X

Transfer

on

End

of

File,

Ch.

H X

Transfer

if

Indicators

Off ........................ X

Transfer

if

Indicators

On

....

....... ...... ....... X

Transfer

on

Index

.................................... .

Transfer

on

Low

MQ

.............................. X

Transfer

on

Minus

.................................... X

Transfer

on

No

Overflow ........................ X

Transfer

on

No

Index

............................. .

Transfer

on

No

Zero ................................ X

Transfer

on

Overflow .... ............ .............. X

Transfer

on

Plus

.... ...... ............ ...... ...... ...... X

Transfer

on

MQ

Plus

................................ X

Transfer

.................................... X

Transfer

on

Redun.

Check,

Ch.

A ........ X

Transfer

on

Redun.

Check, Ch. B ........ X

Transfer

on

Redun.

Check,

Ch.

C ........ X

Transfer

on

Redun.

Check,

Ch.

D ........ X

Transfer

on

Redun.

C1!eck, Ch. E ........ X

Transfer

on

Redun.

Check, Ch. F

........

X

Transfer

on

Redun.

Check,

Ch.

G ........ X

Transfer

on

Redun.

Check,

Ch.

H ........ X

-709 system

instruction

only.

"7090

system

instruction

only.

X

X X X

X

X X X

x x

X X X

X X X X X X

X X

X

X X

X

X X

X

X X X

X X

X X X X X X X

X

SCAT

Mnemonic

Operation

Codes

(Conrd)

CODE

TSX TTR TXH TXI TXL TZE

UAM UFA UFM UFS

USM VDH VDP VLM

"WDR

WEF WEFA

WEFB

WEFC

WEFD

WEFE

WEFF

""WEFG ""WEFH

WPBA

"WPBB

WPBC

....

WPBD

WPBE ""WPBF "WPBG

....

WPBH

WPDA

....

WPDB

WPDC "WPDD

WPDE ""WPDF

"'WPDG

....

WPDH

WPUA

"WPUB

WPUC

....

WPUD

WPUE ""WPUF ""WPUG "'WPUH

WRS

WTBA

WTBB

WTBC

WTBD

WTBE

WTBF

....

WTBG ""WTBH

WTDA

WTDB

WTDC

WTDD

WTDE

WTDF "WTDG "·WTDH

"WTV

XCA XCL XEC ZET

COMMENT

Transfer

and

Set

Index

........................... .

Trap

Transfer

................................ X X

Transfer

on

Index

High

......................... .

Transfer

with

Index

Incremented

....... .

Transfer

on

Index

Low

........ .

Transfer

on

Zero

...

. ................... X X

Un

normalized

Add

Magnitude

.............. X X

Unnormalized

Floating

Add

.................. X X

Unnormalized

Floating

Multiply

.......... X X

Un

normalized

Floating

Subtract

............ X X

Unnormalized

Subtract

Magnitude

...... X X

Variable

Length

Divide

or

Halt

............ X

Variable

Length

Divide

or

Proceed ...... X

Variable

Length

Multiply............ X

Write

Drum

................................................ X

Write

End

of

File .............................. X

Write

End

of

File,

Ch.

A .......................... X

Write

End

of

File,

Ch.

B .......................... X

Write

End

of

File,

Ch.

C .......................... X

Write

End

of

File, Ch. D ...... X

Write

End

of

File,

Ch.

E ....... X

Write

End

of

File,

Ch.

F X

Write

End

of

File,

Ch.

G .......................... X

Write

End

of

File, Ch. H .......................... X

Write

Printer

Binary,

Ch.

A .................... X

Write

Printer

Binary, Ch. B .................... X

Write

Printer

Binary,

Ch.

C .................... X

Write

Printer

Binary,

Ch.

D .................... X

Write

Printer

Binary, Ch. E .................... X

Write

Printer

Binary,

Ch.

F .................... X

Write

Printer

Binary,

Ch.

G .................... X

Write

Printer

Binary,

Ch.

H .................... X

Write

Printer

Decimal, Ch. A X

Write

Printer

Decimal,

Ch.

B ................ X

Write

Printer

Decimal,

Ch.

C ................ X

Write

Printer

Decimal, Ch. D ................ X

Write

Printer

Decimal, Ch. E ................ X

Write

Printer

Decimal, Ch. F ................ X

Write

Printer

Decimal, Ch. G ................ X

Write

Printer

Decimal,

Ch.

H ................ X

Write

Punch,

Ch.

A .................................. X

Write

Punch,

Ch.

B .................................. X

Write

Punch,

Ch.

C .................................. X

Write

Punch,

Ch.

D .................................. X

Write

Punch,

Ch.

E .................................. X

Write

Punch,

Ch.

F............................. X

Write

Punch,

Ch.

G .................................. X

Write

Punch,

Ch.

H .................................. X

Write

Select .................................... X

Write

Tape

Binary,

Ch.

A ........................ X

Write

Tape

Binary,

Ch.

B ........................ X

Write

Tape

Binary,

Ch.

C ........................ X

Write

Tape

Binary,

Ch.

D ........................ X

Write

Tape

Binary,

Ch.

E ........................ X

Write

Tape

Binary,

Ch.

F ........................ X

Write

Tape

Binary,

Ch.

G ........................ X

Write

Tape

Binary,

Ch.

H ........................ X

Write

Tape

Decimal, Ch. A .................... X

Write

Tape

Decimal, Ch. B .................... X

Write

Tape

Decimal, Ch. C .................... X

Write

Tape

Decimal,

Ch.

D .................... X

Write

Tape

Decimal, Ch. E ................... X

Write

Tape

Decimal, Ch. F .................... X

Write

Tape

Decimal,

Ch.

G .................... X

Write

Tape

Decimal,

Ch.

H .................... X

Write

Cathode

Ray

Tube

........................ X

Exchange

Accumulator

and

MQ

........ .

Exchange

Logical

Accumulator

and

MQ

Execute

............................................ X X

Storage Zero

Test

...................................... X X

Appendix

49

..

:

1,1

·.