FLOATING POINT ARITHMETIC UNIT

摘要

PURPOSE:To shorten the time needed for the floating point arithmetic by performing the overflow processing for both the exponent and mantissa parts before the overflow of the mantissa part is produced. CONSTITUTION:An arithmetic circuit 1 performs the arithmetic operations for exponent parts of both floating point data. Then an arithmetic circuit 2 performs an increment operation to the output of the circuit 1. An arithmetic circuit 3 performs the operations for the mantissa parts of both floating point data based on the arithmetic result of the circuit 1. Then a correcting circuit 4 performs an overflow correcting operation to the output of the circuit 3. A multiplexer 5 selects the output of the circuit 1 or 2 based on the arithmetic result of the circuit 3. Then a multiplexer 6 selects the output of the circuit 3 or 4 based on the arithmetic result of the circuit 3 respectively.