Data Format .
Guard Digit .
Number Representation Normalization Instructions .
ADD NORMALIZED
ADD UNNORMALIZED
COMPARE DIVIDE HALVE . LOAD LOAD AND TEST LOAD COMPLEMENT. LOAD NEGATIVE. LOAD POSITIVE LOAD ROUNDED MULTIPLY. STORE .
SUBTRACT NORMALIZED
SUBTRACT UN NORMALIZED
Contents
The floating-point instructions are used to perform
calculations on operands with a wide range of mag­
nitude and to yield results scaled to preserve preci­
sion.
A floating-point number consists of a signed ex­
ponent, represented by the characteristic, and a
signed fraction. The quantity expressed by this
number is the product of the fraction and the num­
ber 16 raised to the power of the exponent. The
exponent is expressed in excess-64 binary notation
(see "Number Representation"); the fraction is ex­
pressed as a hexadecimal number having a radix
point to the left of the high-order digit.
To avoid unnecessary storing and loading opera­
tions for results and operands, four floating-point
registers are provided. The floating-point instruc­
tions provide for the loading, rounding, adding, sub­
tracting, comparing, multiplying, dividing, and stor­
ing, as well as the sign control, of short, long, and
extended operands. Short operands generally provide
faster processing and require less storage than long
or extended operands. On the other hand, long and
extended operands provide greater precision in com­
putation. Operations may be either register to regis­
ter or storage to register.
For addition, subtraction, multiplication, and divi­
sion, instructions are provided that generate normal­
ized results. Normalized results preserve the highest
Floating-Point Instructions
157
158
159
159 160 160 162
163
163
164
165
165
165
166
166
166
167
168
169
169
precision in the operation. For addition and subtrac­
tion, instructions are also provided that generate
unnormalized results. Normalized and unnormalized
operands may be used in any floating-point opera­
tion.
The condition code is set as a result of all sign­
control, add, subtract, and compare operations.
The rounding and extended-operand instructions
are part of the extended-precision floating-point
feature. The other floating-point instructions and the
registers are part of the floating-point feature.
Data Format
Floating-point data occupies a fixed-length format,
which may be either a four-byte (short) format, an
eight-byte (long) format, or a 16-byte (extended)
format. The short and long formats may be designat­
ed as operands both in main storage and in the
floating-point registers, whereas the extended for­
mats can be designated only in the floating-point
registers.
The floating-point registers are numbered 0, 2, 4,
and 6. Designation of an odd-numbered register in
the Rt or R2 field of a floating-point instruction
causes the operation to be suppressed and a program
interruption for specification exception to occur.
Floating-Point Instructions 157
Previous Page Next Page