Floating Point Representation

[Foxtildawn]

Floating Point Representation:

I have to convert 21.426304 into an Unnormalized number. It has to be 16 bits long with a 1 bit sign, 5 bit exponent, and a 10 bit sinificand. I understand that you have to take the fraction of the original number or something along those lines but i have no clue how. Any help in understanding how to do this so i can continue on with the rest of the homework problem would be great...Thanks

[murschech]

I don't know what "unnormalized" means, but here's how to write it as a normalized floating pt. number: Since the smallest pwer of 2 bigger than the given number is 32, write it as
21.426304 = 32*0.669572, so the exponent is 5 and the significand is .669572. You've got to write this last number to base 2.

.669572 = a1/2+a2/4+a3/8+... Multiply both sides by 2 and take integer parts. You conclude a1=1. Now procede at your own risk.