Abstract: A computing device for fast weighted sum calculation in neural networks is disclosed. The computing device comprises an array of processing elements configured to accept an input array. Each processing element comprises a plurality of multipliers and a multiple levels of accumulators. A set of weights associated with the inputs and a target output are provided to a target processing element to compute the weighted sum for the target output. The device according to the present invention reduces the computation time from M clock cycles to log2M, where M is the size of the input array.

Abstract: A method and apparatus for calculating a quotient from a dividend and a divisor, wherein the divisor can be represented as (2N+2M) where N is greater than M, and wherein the dividend comprises an X-bit binary number divisible by the divisor without a remainder. The values of N and M for the dividend are determined such that the divisor is equal to the value (2N+2M). The M-th through the (N−1)-th bits of the dividend are selected as lower order bits of the quotient. The (N−1)-th and the (2N−M−1)-th bits of the dividend are examined. If the (N−1)-th bit of the dividend is “1” and if the (2N−M−1)-th bit of the dividend is “0”, then one is subtracted from a value represented by the (2N−M)-th through the (X−1)-th bits of the dividend to obtain a result as higher order bits of the quotient. Otherwise, the (2N−M)-th through the (X−1)-th bits of the dividend are selected as higher order bits of the quotient.