Abstract: Disclosed is a power-optimized distributed arithmetic (DA)-based feed forward equalizer (FFE) that performs on-demand equalization processing of a data sample. Specifically, a data stream is represented by digital words, which indicate signal levels at taps on a transmission medium. A screener applies formulas to selected taps as opposed to all taps (e.g., to the main cursor tap, which corresponds to the current data sample, and to specific pre-cursor and post-cursor taps, which correspond to immediately proceeding and following data samples) to determine whether the current data sample (which should indicate a specific two-bit symbol) has degraded during transmission to a point where equalization processing is required. If so, a bypass flag is set to a first level so that the data sample is subjected to equalization processing. If not, the bypass flag is set to a second level so that such processing is bypassed. Also disclosed is a corresponding method.