In this letter, we present a novel single-precision floating-point multiply-accumulator (FNA-MAC) to achieve lower hardware resource, reduced computing latency and improved computing accuracy for continuous dot product operations. By further fusing the normalization and alignment in the traditional FMA algorithm, the proposed architecture eliminates the first N-1 normalization and rounding operations for an N-point dot product, and preserves the precision of interim results in a significant bit size that is twice of that in the traditional methods. The normalization and rounding of the final result is processed at the cost of consuming an additional multiply-add operation. The simulation results show that the improvement in computational accuracy is significant. Meanwhile, when comparing to a recently published FMA design, the proposed FNA-MAC can reduce the slice look-up table/flip-flop resource and computing latency by a fact of 18%, 33.3%, respectively.

This paper presents a well-structured modified Booth encoding (MBE) multiplier which is applied in the design of a reconfigurable multiply-accumulator (MAC) core. The multiplier adopts an improved Booth encoder and selector to achieve an extra-row-removal and uses a hybrid approach in the two's complementation circuit to reduce the area and improve the speed. The multiplier is used to form a 32-bit reconfigurable MAC core which can be flexibly configured to execute one 3232, two 1616 or four 88 signed multiply-accumulation. Experimentally, when implemented with a 130 nm CMOS single-Vt standard cell library, the multiplier achieved a 15.8% area saving and 11.7% power saving over the classical design, and the reconfigurable MAC achieved a 4.2% area and a 7.4% power saving over the MAC design published so far if implemented with a mixed-Vt standard cell library.