it is difficult to understand the accuracy of floating-point numbers, online interpretation is always 2 ^ 23 =. Of course, it will be 7 bits, but the Mantissa of 23bit memory is converted from binary to 10, why the precision is 7, even if it is 7, there are exponential bit constraints. The maximum number of float is 3.4 * 10 ^ 38, but the precision is only 7 digits. Isn"t such a large number itself more than 7 digits? How to understand what this precision means