An architecture for high-performance/small-area multipliers for use in digital filtering applications

A multiplier architecture and encoding scheme well suited for programmable digital filtering applications is described. The multiplier's partial product recoding scheme uses only simple multiplexers and takes advantage of a RAM that stores filter coefficients. We use an optimized 20-transistor full-adder cell in the carry-save adder array, and a carry-select vector-merge adder produces the final output. An integrated circuit comprising an ll-b by ll-b multiplier using second-order recoding has been fabricated in 2-/spl mu/m CMOS technology. It operates in 22 ns and its core occupies 1.53 mm/sup 2/. Also, an ll-b by 16-b multiplier using third-order recoding has been fabricated through MOSIS in 1.2-/spl mu/m CMOS technology. Its core occupies 0.9 mm/sup 2/ and it operates in 19 ns.<
>