计算机浮点运算的尾数处理

On the Mantissa Processing of Computer Floating Point Arithmetic

浮点数在本质上仍然是以定点数表现出来的,通过浮点数中的指数使数据的表示范围得到扩展。浮点运算的尾数处理包括规格化、舍入处理和溢出处理。规格化就是使尾数部分最高有效位是1。舍入处理就是按照一定的规则确定一致的位数,然后舍去某些数字后面多余的尾数。溢出就是计算结果大于寄存器或内存所能储存或表示的能力。

The idea of floating-point representation over intrinsically integer fixed-point numbers, which consists purely of significand, is that expanding it with the exponent component achieves greater range.The mantissa handling including normalization,rounding rules and overflow handling.The floating point will be normalized such that the most significant bit will be a one. Rounding handling aims to turn a given value into another value with a specified number of significant digits.Overflow is that occurs when a calculation produces a result that is greater than what a given register can store or represent.