Method and system for processing exponents in floating-point multiplication

Claim 2

An exponent processing method in floating-point multiplication determining a product of two operands of floating-point format, said exponent processing method being for determining a sum Z of exponents X and Y of n bits of the two operands, and comprising: applying an inversion of the MSB, i.e., (n-1)-th bit of one of the exponents (e.g., X) to the n-th bit and the (n-1)-th bit of a first input terminal set of an adder capable of performing an addition of (n+1) bits, applying all the bits except the MSB, i.e., the (n-2)-th to 0-th bits of the exponent (X) to the (n-2)-th to 0-th bits of said first input terminal set, applying a signal "0" to the n-th bit of a second input terminal set of the adder, applying all the bits, i.e., the (n-1)-th to the 0-th bits of the other exponent (Y) to the (n-1)-th to the 0-th bits of said second input terminal set, applying a signal "1" to a carry input terminal of the adder, and using the (n-1)-th to the 0-th bits of the output of the adder as bits constituting the sum Z.

Claim 3

A method according to claim 1, further comprising: judging that underflow or overflow is taking place when the n-th bit of the output of the adder is "1".

Claim 4

An exponent processing system in a floating-point multiplication system determining a product of two operands of floating-point format, said exponent processing system being for determining a sum Z of exponents X and Y of n bits of the two operands, and comprising: an adder capable of performing addition of at least (n+1) bits, a carry signal generator for supplying a signal "1" to a carry input terminal of said adder, and an inverter for inverting the MSB i.e., the (n-1)-th bit of the exponent (e.g., X) of a first operand, said adder receiving, at the n-th and the (n-1)-th bits of its first input terminal set, the output of said inverter, and, at the (n-2)-th to the 0-th bits of its first input terminal set, the (n-2)-th to the 0-th bits of the exponent (X) of said first operand, said adder receiving, at the n-th bit of its second input terminal set, a signal "0", and at the (n-1)-th to the 0-th bits of its second input terminal set, all the bits, i.e., the (n-1)-th to the 0-th bits of the exponent (Y) of a second operand, the (n-1)-th to the 0-th bits of the output of said adder being used to form the sum Z.

Claim 5

An exponent processing system in a floating-point multiplication system determining a product of two operands of floating-point format, said exponent processing system being for determining a sum Z of exponents X and Y of n bits of the two operands, and comprising: an adder capable of performing addition of at least (n+1) bits, a carry signal generator for supplying a signal "1" to a carry input terminal of said adder, and an inverter for inverting the MSB i.e., the (n-1)-th bit of the exponent (e.g., X) of a first operand, said adder receiving, at the n-th and the (n-1)-th bits of its first input terminal set, the output of said inverter, and, at the (n-2)-th to the 0-th bits of its first input terminal set, the (n-2)-th to the 0-th bits of the exponent (X) of said first operand, said adder receiving, at the n-th bit of its second input terminal set, a signal "0", and at the (n-1)-th to the 0-th bits of its second input terminal set, all the bits, i.e., the (n-1)-th to the 0-th bits of the exponent (Y) of a second operand, the (n-1)-th to the 0-th bits of the output of said adder being used to form the sum Z.

Claim 6

An exponent processing method in floating-point multiplication determining a product of two operands of floating-point format, said exponent processing method being for determining a sum Z of exponents X and Y of n bits of the two operands, and comprising: inputting the exponents X and Y to the (n-1)-th to the 0-th bits of two input terminal sets of an adder capable of performing addition of at least n bits, inputting a signal "1" to a carry input terminal of the adder, and using the n-th bit (Z 0n ) of the output of the adder as the MSB, i.e., the (n-1)-th bit (Z n-1 ) of the sum Z, and the (n-2)-th to the 0-th bits (Z 0n-2 . . . Z 00 ) of the output of the adder as the (n-2)-th to the 0-th bits (Z n-2 . . . Z 0 ) of the sum Z.

Claim 7

An exponent processing method in floating-point multiplication determining a product of two operands of floating-point format, said exponent processing method being for determining a sum Z of exponents X and Y of n bits of the two operands, and comprising: inputting the exponents X and Y to the (n-1)-th to the 0-th bits of two input terminal sets of an adder capable of performing addition of at least n bits, inputting a signal "1" to a carry input terminal of the adder, and using the n-th bit (Z 0n ) of the output of the adder as the MSB, i.e., the (n-1)-th bit (Z n-1 ) of the sum Z, and the (n-2)-th to the 0-th bits (Z 0n-2 . . . Z 00 ) of the output of the adder as the (n-2)-th to the 0-th bits (Z n-2 . . . Z 0 ) of the sum Z.

Claim 8

A method according to claim 4, further comprising: judging that underflow or overflow is taking place when the n-th and the (n-1)-th bits Z 0n , Z 0n-1 of the output of the adder coincide with each other.

Claim 9

An exponent processing system in a floating-point multiplication system determining a product of two operands of floating-point format, said exponent processing system being for determining a sum Z of exponents X and Y of n bits of the two operands, and comprising: an adder capable of performing addition of at least n bits, and receiving at the (n-1)-th to the 0-th bits of its input terminal sets, the exponents X and Y, and a carry signal generator for supplying a signal "1" to a carry input terminal of said adder, wherein the n-th bit Z 0n of the output of the adder is used as the MSB, i.e., the (n-1)-th bit Z n-1 of the sum Z, and the (n-2)-th to the 0-th bits Z 0n-2 . . . Z 00 of the output of the adder are used as the (n-2)-th to the 0-th bits of the sum Z.

Claim 10

An exponent processing system in a floating-point multiplication system determining a product of two operands of floating-point format, said exponent processing system being for determining a sum Z of exponents X and Y of n bits of the two operands, and comprising: an adder capable of performing addition of at least n bits, and receiving at the (n-1)-th to the 0-th bits of its input terminal sets, the exponents X and Y, and a carry signal generator for supplying a signal "1" to a carry input terminal of said adder, wherein the n-th bit Z 0n of the output of the adder is used as the MSB, i.e., the (n-1)-th bit Z n-1 of the sum Z, and the (n-2)-th to the 0-th bits Z 0n-2 . . . Z 00 of the output of the adder are used as the (n-2)-th to the 0-th bits of the sum Z.

Claim 11

A system according to claim 6, further comprising: an Exclusive-NOR gate receiving the n-th and the (n-1)-th bits of the output of the adder, and providing, at its output, a signal indicative of occurrence of underflow or overflow.