Floating point adder algorithm

[panda1234]

Hi,
Below picture shows algorithm of adding floating point number but i have a question why we use 29 bit adder and what is sticky bit ?

 

[barry]

Why shouldn't it be 29 bits? Without knowing the width of your data, it's impossible to answer your question, but my GUESS is that it has to do with overflow. Sticky bits are flags for special cases, e.g., overflow, underflow, divide by zero,etc.

 

[panda1234]

width of a floating number single precision in IEEE 754 is 32 bits.
Exactly my question is:we have 23 bits fraction part and 1 hidden bit and 2 bit for R and G and 1 bit for Sticky bit with sign extension we have two 28 bit number and need 28 bit adder but this adder is 29 bit why?
and what is sticky bit exactly?

 

[barry]

First of all, you said nothing about IEEE 754 initially; it's not the only FP representation there is. If you add two 28-bit numbers, your result can require 29 bits (or a 28 bits and carry bit).

 

[panda1234]

Thanks,and what is sticky bit exactly?

 

[ads-ee]

Thanks,and what is sticky bit exactly?

Just two links from many links displayed when using the search string "ieee 754 sticky bit".

**broken link removed**


Why can't posters use google, follow links, and read the information found on those sites?

 

[barry]

How do I search for google?