Conditional Estimation Of Residuals With Prescaling For Use In Low-Energy Division Units

This paper discusses digit-recurrence division with operand scaling focusing on its implementation for energy and delay. The implementations with are compared with implementations without scaling and with multiplicative division methods. Results demonstrate the ability of overlapping and operand scaling to enhance delay and energy-driven implementations. Results are given in cmos12soi 45nm SOI technology using ARM-based standard-cells and commercial EDA toolsets.