How to convert from two's complement to single precision?

[oldrich.florian]

Hello,
Does anybody know how can I convert a number coded in two's complement to single precision?

I have
1111111111110011
in two's complement

and I need to convert it to single precision.

Thank you for your replies.

 

[dale]

Use the first character from the left as sign indicator (in this case, negative).

Then convert 111111111110011 as 16384 + 8192 + 4096 ... 16 + 2 + 1, which yields 32755.

 

[oldrich.florian]

Use the first character from the left as sign indicator (in this case, negative).

Then convert 111111111110011 as 16384 + 8192 + 4096 ... 16 + 2 + 1, which yields 32755.

Isn't it 1 bit for sign indicator, 8 bit for exponent and the rest for mantissa?
Like this:

 

[dale]

Ack, I messed up. Let's try again.

So you have this number: 1111111111110011

Invert it first because the leftmost digit denotes negative sign: 0000000000001100

Plus 1, i.e.: 0000000000001101

Conversion to decimal yields 13. So the result is -13.

 

[dale]

BTW, in response to the image file you posted. It refers to float I believe. Two's complement is all about int.