Skip to Main Content
A radix-4 scheme is proposed for computing the square s=x2 using left-to-right multiplication algorithm and on-the-fly conversion. The MS and LS parts are computed in parallel so that the final result is obtained with reduced extra delay compared to a carry-propagate adder required in conventional schemes. The input and the output are in the two's complement fixed-point representation. The most-significant half of the result can also be used the most-significant digit first to allow overlap with successive operations. The scheme can be adapted to compute x2+y2. The critical path is kept small by the use of redundant arithmetic. The proposed algorithm, its implementation and performance are discussed.