Abstract: Exact eight-bit floating-point multiplication performed using simple integer operations is considered in this work. For two-bit mantissa formats, correct rounding for all considered rounding ...