interruption for the significance exception does not

occur; rather, the characteristic is made zero, yielding

a true zero result. Exponent underflow does not occur

for a zero fraction.

The sign of the sum is derived by the rules of

algebra. The sign of a sum with zero result fraction

is always positive.

ResultingCondition Code: o Result fraction is zero

1 Result is less than zero

2 Result is greater than zero

3 Result exponent overflows

Program Interruptions:

Operation (if floating-point feature is not in-

stalled)

Addressing (AE and AD only)

Specification

Significance

Exponent overflow

Exponent underflow

Programming Note

Interchanging the two operands in a floating-point

addition does not affect the value of the sum.

AddUnnormalized AUR RR (Short Operands) I 3E Rl I R2 I 0 78 11 12 15

AU RX (Short Operands)I 7E Rl I X

2I B2 D2 0 78 11 12 1516 1920 31

AWR RR (Long Operands)I 2E Rl I R2 I 0 78 11 12 15

AW RX (Long Operands)I 6E Rl I X

2I B2 0 78 11 12 1516 1920 31

The second operand is added to the first operand, and

the unnormalized sum is placed in the first operand

location.

In short-precision, the low-order halves of the float

ing-point registers are ignored and remain unchanged.

After the addition the intermediate sum is truncated

to the proper fraction length.

When the resulting fraction is zero and the signifi

cance mask bit is one, a significance exception exists

and a program interruption takes place. When the

resulting fraction is zero and the significance mask

bit is zero, the program interruption for the signifi

cance exception does not occur; rather, the character

istic is made zero, yielding a true zero result.

Leading zeros in the result are not eliminated by

normalization, and an exponent underflow cannot

occur.

The sign of the sum is derived by the rules of

algebra. The sign of a sum with zero result fraction is

always positive.

ResultingCondition Code: o Result fraction is zero

1 Result is less than zero

2 Result is greater than zero

3 Result exponent overflows

Program Interruptions:

Operation (if floating-point feature is not in-

stalled)

Addressing(AU and A w

only)

Specification

Significance

Exponent overflow

SubtractNormalized SER RR (Short Operands) I 3B Rl I R2 I 0 78 11 12 15 SE RX (Short Operands) I 7B Rl I X

2I B2 D2 0 78 11 12 1516 1920 31 SDR RR (Long Operands) I 2B Rl I R2 I 0 78 11 12 15 SD RX (Long Operands)

6B

7 8 11 12 15161920 31

The second operand is subtracted from the first op

erand, and the normalized difference is placed in thefi'rst operand location.

Floating-Point Arithmetic 45

occur; rather, the characteristic is made zero, yielding

a true zero result. Exponent underflow does not occur

for a zero fraction.

The sign of the sum is derived by the rules of

algebra. The sign of a sum with zero result fraction

is always positive.

Resulting

1 Result is less than zero

2 Result is greater than zero

3 Result exponent overflows

Program Interruptions:

Operation (if floating-point feature is not in-

stalled)

Addressing (AE and AD only)

Specification

Significance

Exponent overflow

Exponent underflow

Programming Note

Interchanging the two operands in a floating-point

addition does not affect the value of the sum.

Add

AU RX (Short Operands)

2

AWR RR (Long Operands)

AW RX (Long Operands)

2

The second operand is added to the first operand, and

the unnormalized sum is placed in the first operand

location.

In short-precision, the low-order halves of the float

ing-point registers are ignored and remain unchanged.

After the addition the intermediate sum is truncated

to the proper fraction length.

When the resulting fraction is zero and the signifi

cance mask bit is one, a significance exception exists

and a program interruption takes place. When the

resulting fraction is zero and the significance mask

bit is zero, the program interruption for the signifi

cance exception does not occur; rather, the character

istic is made zero, yielding a true zero result.

Leading zeros in the result are not eliminated by

normalization, and an exponent underflow cannot

occur.

The sign of the sum is derived by the rules of

algebra. The sign of a sum with zero result fraction is

always positive.

Resulting

1 Result is less than zero

2 Result is greater than zero

3 Result exponent overflows

Program Interruptions:

Operation (if floating-point feature is not in-

stalled)

Addressing

only)

Specification

Significance

Exponent overflow

Subtract

2

6B

7 8 11 12 1516

The second operand is subtracted from the first op

erand, and the normalized difference is placed in the

Floating-Point Arithmetic 45