METHODS AND SYSTEMS FOR USING SHIFTER CIRCUIT AND 3X3 CONVOLVER UNITS TO EMULATE FUNCTIONALITY OF LARGER SIZED CONVOLVER UNITS

Convolution with a 5x5 kernel involves computing the dot product of a 5x5 data block with a 5x5 kernel. Instead of computing this dot product as a single sum of 25 products, the dot product is computed as a sum of four partial sums, where each partial sum is computed as a dot product of a 3x3 data block with a 3x3 kernel. The four partial sums may be computed by a single 3x3 convolver unit over four time periods. During each time period, at least some of the weights received by the 3x3 convolver unit may correspond to a quadrant of weights from the 5x5 kernel. A shifter circuit provides shifted columns (left or right shifted) of the input data to the 3x3 convolver unit, allowing the 3x3 convolver unit access to the 3x3 data block that spatially corresponds to a particular quadrant of weights from the 5x5 kernel.