| 摘要 |
PROBLEM TO BE SOLVED: To provide a data arithmetic processor suitable for simplifying an arithmetic processing when performing addition or subtraction by a block floating system. SOLUTION: When the addition using a plurality of data blocks is performed, the data arithmetic processor performs a correction processing before the arithmetic operation. In a scale correction processing, a block scale factor of a data block including data with the maximum absolute value is selected as a common block scale factor among block scale factors of a plurality of data blocks. Then, scaling is performed to the respective pieces of data of data groups based on scale factors consisting of difference between the group scale factors and the common block scale factor of the data group by every data group of the data blocks. |
