FLOATING POINT COMPUTING ELEMENT

摘要

PURPOSE:To execute an arithmetic without generating overflow or underflow by constituting a computing element which can execute the four rules of arithmetic by means of dealing with a floating point data system defined by the means of a specified expression. CONSTITUTION:The computing element which can execute the operation of the four rules of arithmetic by the data system shown by an expression I is constituted. An exponential part operation part 3 obtains the number of shift in the mantissa part of one input data at the time of addition/subtraction, and obtains an exponential part operation result at the time of multiplication/division by (m) and (n) when the exponential part is expressed by the form of (2) and the parameter 1. An exponential part parameter decision circuit 4 adjusts the exponential part given by the exponential part operation part 3 by the number of shift by normalization given from an addition/ subtraction mantissa part operation part 1 or a multiplication/division mantissa part operation part 2 and obtains (m) and (n) which express the new exponential part of the result of the operation of the four rules of arithmetic and the adjustment parame ter 1. Thus, the operation of the four rules of arithmetic can be executed without being restricted by the overflow and the underflow owing to numerical expression by the floating point data system.