摘要 |
PROBLEM TO BE SOLVED: To realize an arithmetic operation unit having sufficient arithmetic operation accuracy on a floating point number product sum arithmetic operation by an addition of a few circuit scale. SOLUTION: For performing a product sum arithmetic operation for adding third data to the product of first data and second data, the first data and the second data are multiplied by a floating point multiplier 114, and a binary digit string for expressing a mantissa part in a multiplication result is divided into a string for expressing a superordinate digit in the mantissa part and a string for expressing a subordinate digit in the mantissa part. A floating point adder 113 is allowed to first perform addition of third data and subordinate multiplication result data 114L with the binary digit string for expressing the subordinate digit as the mantissa part, and a floating point adder 113 is allowed to thereafter perform addition of an addition result and superordinate multiplication result data 114H with the binary digit string for expressing the superordinate digit as the mantissa part, and rounding processing is applied to the addition result thereafter, and is used as a result of the product sum arithmetic operation. COPYRIGHT: (C)2003,JPO
|