Embedded Systems Group (ES)

Floating Point Conversion

This tool converts a given rational or fixed point number to an approximated floating point number. The input number should be of the form ±p/q or ±p.q with radix-B numbers p and q given to the specified radix B. For the floating point number that shall be generated, one has to specify the base and the numbers of digits of the mantissa and the exponent.

The tool then converts the given number to a floating point number that is normalized if possible. Even in case of target base 2, the tool does not generate a IEEE 754 floating point number, since we do not make use of hidden bits here. You can however specify one additional bit for the mantissa, and have to adjust the exponent in case of denormalized numbers to obtain a IEEE 754 denormalized number.

radix of given number
input number
radix of floating point number
digits of mantissa
digits of exponent