Efficient FPGA-implementation of twos complement digit-serial/parallel multipliers

This paper presents an efficient implementation of digit-serial/parallel multipliers on 4-input look-up table (LUT)-based field programmable gate arrays (FPGAs). This subset of FPGA devices hide individual gate delays and add important wiring delay. These two facts produce important changes over the theoretical advantages of each topology. Architectural transformations are applied to obtain topologies with minimum logic depth and where the maximum clock speed is limited by the FPGA technology. The main results of applying those transformations to the different multipliers have been quantified for Altera FLEX10K family, and the conclusions have been extrapolated to other FPGA families.