Apparatus and method of performing product-sum operation

An apparatus for performing a product-sum operation by multiplying and adding floating point number data representing a floating point number using a bit string, comprising:

• multiplication means for performing a multiplication on the floating point number data;
• addition means for performing an addition on the floating point number data;
• rounding means for performing a rounding process on the floating point number data obtained as a result of the addition performed by said addition means;
• result storage means for storing a result of the product-sum operation by adding third data which is floating point number data to a product of first data and second data which are floating point number data;
• multiplication control means for allowing said multiplication means to compute multiplication result data which is a result of the multiplication of the first data and the second data;
• first addition control means for allowing said addition means to compute first addition result data obtained by adding the third data to less significant digit multiplication result data having a fixed-point part formed by a bit string representing the less significant digits in two divisions obtained by dividing a bit string representing the fixed-point part of the multiplication result data into more significant digits and less significant digits;
• and
• second addition control means for allowing said addition means to compute second addition result data obtained by adding more significant digit multiplication result data having a fixed-point part formed by a bit string representing more significant digits to the first addition result data, wherein
• said result storage means stores first product-sum operation result data which is floating point number data obtained by said rounding means performing a rounding process on the second addition result data.

A method for performing a product-sum operation by adding third data which is floating point number data to a product of first data and second data which are floating point number data representing a floating point number by a bit string, comprising:

• allowing a multiplication unit multiplying floating point number data to multiply first data by second data;
• allowing an addition unit to add floating point number data by adding the third data to less significant digit multiplication result data having as a fixed-point part a bit string representing less significant digits in two divisions obtained by dividing a bit string representing a fixed-point part of more significant digit multiplication result data obtained as a result of the multiplication into two divisions, that is, a bit string indicating more significant digits and less significant digits;
• allowing the addition unit to compute the second addition result data obtained by adding more significant digit multiplication result data having as a fixed-point part the bit string representing more significant digits to the first addition result data which is a result of the addition;
• and
• obtaining as a result of the product-sum operation the data obtained by performing a rounding process on the second addition result data.