Assembly Programs - IBM 709 General Information Manual

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
next storage reference cycle. All numbers in Figure
21 are shown in octal notation.
The next cycle, 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 next sequential 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 .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