IEEE 754-2008 compliant decimal floating point divider / Ahmed Hamdy Ahmed Khalil ; Supervised Hossam A. H. Fahmy

Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Electronics and Communication

The binary numbering system is the most suitable for electronic computers, and the decimal numbering system is the most used in human life especially in commercial applications. But the binary system can not represent decimal fraction numbers accurately, so the IEEE 754-2008 standard defines the decimal floating numbers and decimal floating arithmetic operations. The decimal floating point division is one of these operations. This thesis presents new design and implementation for decimal floating point divider based on Accurate Quotient Approximation algorithm which can be considered as very high radix decimal divider. Accurate Quotient Approximations is an iterative division algorithm based on look-up table that gives an approximation to the reciprocal of the divisor. The algorithm iterates by using the reciprocal to find an approximated value for the quotient. Then the approximated quotient is multiplied by the divisor and the result is subtracted from the partial remainder to find the new partial remainder. At first iteration the partial remainder is the dividend. The divider iterates until the quotient reaches the desired accuracy. This divider can produce three digits per iteration. The operation of this divider is compliant to IEEE 754-2008 standard. Also it supports the five rounding modes that are defined in the IEEE 754- 2008 standard, and another two famous rounding methods. This divider is extended to execute decimal floating point division operations

Issued also as CD