Principle of computer composition--floating point representation method

In order to represent floating-point numbers, the number is divided into two parts: the integer part and the fractional part. For example, floating-point 14.234 has an integer part of 14 and a fractional part of 0.234. First, the floating-point number is converted to binary, and the steps are as follows: 1 converts the integer part to binary. 2 converts fractional parts to binary. 3 adds a decimal point between the two sections. Floating-point numbers can also be normalized, Floating-point numbers can be represented by single-precision notation and double-precision notation. Normalization stores only the three-part information for this number: symbols, pointers, and mantissa. If +1000111.0101 is normalized to
+    2^6   *   1.0001110101
Sign  index        mantissa
Single-precision representations of normalized numbers such as +2^6*1.01000111001 solutions:
Since the sign is positive, it is represented by 0. The exponent is 6, and in excess_127 notation, the exponent plus 127 gets 133. In binary notation, 10000101. The mantissa is 01000111001. When the number of digits is increased to 32 bits, Get 01000111001000000000000. Note that you cannot miss the left 0 because it is a decimal. Missing that 0 is the equivalent of multiplying this number by 2. This number is stored in memory in 32 digits. As shown below
Sign      index              mantissa
0     10000101      01000111001000000000000

Principle of computer composition--floating point representation method