Division of two floating point numbers is about the same as division for two fixed point numbers. In this case, we can assume the implied binary point is between bit 22 and bit 23 for the representation of the mantissa of a single precision floating point number.
The exponent of the result is pe = ae − be, assuming a is the dividend, and b is the divisor (ae and be are the exponent values of a and b, respectively).
The result of the division may need to be normalized.