1

I have a quick question about a problem I'm trying to solve. For this problem, I have to convert (0.0A)16 into IEEE754 half precision floating point standard. I converted it to binary (0000.0000 1010), normalized it (1.010 * 2^5), encoded the exponent (which came out to be 01010), but now I'm lost on how to put it into the actual form. What should I do with the fractional part? The answer comes out to be 0 01010 01 0000 0000.

I know there's something to do with adding a omit 1, but I'm not entirely sure on where that happens either.

Any help is appreciated!

asavyy
  • 31
  • 1
  • 9

1 Answers1

2

The 1 you have to omit is the first one of the mantissa, since we know the significant part always starts with 1 (this way, IEEE-754 gains one bit of space). The mantissa is 1.010, so you will represent only "010".

The solution 0 01010 0100000000 means: 0 is the sign; 01010 is the exponent; 01000000 is the mantissa, omitting the first one.

luisp
  • 91
  • 5