EFFICIENT DETECTION SYSTEM OF FLOATING-POINT OVERFLOW AND UNDERFLOW

摘要

PURPOSE: To provide a processing system to judge whether or not an underflow or an overflow state takes place at the same time as deciding the result of an exponent of a floating decimal point. CONSTITUTION: The processing system that judges whether or not an underflow or an overflow state takes place at the same time as deciding the result of an exponent of a floating decimal point, uses a latched constant group to be added to an exp int and an exp adjust to judge an out-range state in all cases. A proper one among the latched constants (exponent range check exp range chk) is added to the exp int and the exp adjust to obtain a changed value depending whether or not an exponent result (exp result) is at the outside of the range.