| 摘要 |
PROBLEM TO BE SOLVED: To solve the problem that since an arithmetic processing unit is individually configured of circuits when a plurality of arithmetic classifications are required, the circuit scale is likely to be large, and that since the arithmetic processing unit is individually configured of circuits when arithmetic precision is different according to arithmetic parts even though there is only one arithmetic classification, the circuit scale is likely to be increased in the same way. SOLUTION: The process of arithmetic processing is divided into respective processes, that is, pre-processing, approximation polynomial selection, approximation arithmetic processing and post-processing, and the pre-processing, approximation polynomial group and post-processing suitable for an arithmetic classification signal to be inputted is selected, and the approximation polynomial of precision equivalent to the inputted arithmetic precision signal is selected from the selected approximation polynomial group, and approximation arithmetic processing is performed. COPYRIGHT: (C)2006,JPO&NCIPI
|
