[摘要]:A new rounding method to reduce the required precision of the multiplier for Goldschmidt division is presented. It applies special truncation methods at the final iteration step. This requires a minor modification to the rounding constants of the multipli