Basics
Extension of sign
|
|
|||||||||||||||||||||||||||||||||||||||||||||
|
This question always comes on when the quantity of source information is less than destination's capacity. This is a very typical situation because we often want to save "short" information into "large" register (for example, one byte or wyde into octabyte register). The problem is what to do with the "unused" left bits of the destination. It first comes to our mind that they must be reset to zero. But this evident decision is correct only for unsign numbers. But if the number contains sign bit, the picture becomes more complex. In negative numbers the leftmost bits are usually 1s, but not 0s. So in the signed case all left "unused" bits are filled by the value of the source sign bit. In other words for positive numbers when the sign is reset, zero value is used for filling, but for negative ones the most significant bits will be set to 1. Here is the picture how it works: 
Some concrete examples:
Related topics: |
||||||||||||||||||||||||||||||||||||||||||||||
| (C) 2001, Evgeny Eremin. rEd-MMI project documentation | ||||||||||||||||||||||||||||||||||||||||||||||