A method and apparatus for performing integer and floating-point divide operations using a single modified SRT divider in a data processor. The floating-point and integer division is performed using SRT division on normalized positive mantissas (dividend and divisor). Integer division shares portions of the floating point circuitry, however, the sequence of operations is modified during the performance of an integer divide operation. The SRT divider performs a sequence of operations before and after an iteration loop to re-configure an integer divisor and dividend into a data path representation which the SRT algorithm requires for floating-point mantissas. During the iteration loop, quotient bits are selected and used to generate intermediate partial remainders. The quotient bits are also input to quotient registers which accumulate the final quotient mantissa. A full mantissa adder is used to generate a final remainder.

Method and apparatus for performing integer and floating point division using a single SRT divider in a data processor

A radix-16 divider, generally comprising two improved radix-4 dividers cascaded combinatorially, also utilizes improved overlapping quotient bit selection, together with concurrent quotient bit selection and rounding techniques, to produce four bits of quotient per divider cycle with reduced hardware requirements.

Radix-16 divider using overlapped quotient bit selection and concurrent quotient rounding and correction

A floating-point division circuit for performing division on floating-point data using a non-recovery type division method is disclosed. The floating-point division circuit includes a circuit portion for conducting a pre-division processing and pattern determination on a dividend and a divisor, an exponent operation portion, a mantissa division portion, and a quotient generating portion, further including either or both of an exception/non-operation detecting portion and a control portion. The exception/non-operation detecting portion generates a stop signal when detecting a non-operation pattern so as to stop a repetition of operations in the mantissa division portion. The control portion generates either a non-executional signal or a control signal so as to stop a latch operation during a period when no instruction for division is executed.

Floating-point division circuit

A divider/multiplier circuit (10) is disclosed. In a divider mode, numerator terms are coupled to a normalizer (14) which generates normalized numerator values and corresponding numerator exponent values therefrom. Denominator terms are coupled to a look-up normalizer (20) which generates normalized denominator inverse values and corresponding denominator exponent values therefrom. The numerator and denominator exponent values are summed in an adder circuit (18) to generate a sum exponent value. The normalized numerator and inverse denominator values are multiplied in a multiplier circuit (16) to generate a normalized quotient value. The normalized quotient value is denormalized according to the sum exponent value. In a multiply mode of operation first and second multiplicands are coupled to the multiplier circuit (16). In a high precision divide mode, a sequence of numerator and inverse denominator values are coupled to the multiplier circuit (16) to generate a sequence of partial product terms. The partial product terms are accumulated in a high precision loop (24) to provide a high precision division value. Negative multiplicands and numerator values are handled by a leading absolute value generator (12) which generates the absolute value of the multiplicand or numerator value. A trailing signed value generator (22) additively inverts the product or quotient if the multiplicand or numerator value was negative.

Divider/multiplier circuit having high precision mode

A method to adjust the divisor and dividend, for application to a divider, so that the mantissa part of the divisor is transformed to be within a known limited range. The limiting of the transformed divisor range enables the complexity of the quotient select logic to be reduced accordingly. Once the divisor is restricted to the selected range, the dividend is adjusted proportionally so the quotient is unchanged.

Transformation of divisor and dividend in digital division

In dividing a pair of binary coded, hexidecimal floating point numbers, leading zero quotient bits are eliminated by comparing the magnitudes of the most significant digits of the fractional parts of the dividend and divisor after the dividend and divisor have been normalized

Stephen J. Rawlinson

Papers

Saving cycles in floating point division