Extended precision accumulation of floating-point data for digital signal generation and processing.