ripple carry adder vs carry lookahead adder

[hby01]

What is the difference between ripple carry adder and carry lookahead adder?

 

[DTFT]

Succinctly, a ripple carry adder passes its carry bit through a long logic chain, which is very straightforward to design, but can have a very large delay.

A carry-lookahead adder uses a clever algorithm to cut that logic to only a few layers, and thus keeps the delay pretty short even for a worst-case situation.

**broken link removed** of the carry lookahead algorithm is something that often shows up on places like Digital Logic exams, in case you're curious.

 

[Analog Ground]

Here is a more formal definition. A ripple carry adder is implemented purely with a half-adder and multiple full adders. The maximum delay is through the carry signal path from the LSB to the MSB which "ripples" through all the adders. There is pretty much only one way to implement this type of adder. A carry look-ahead adder includes additional logic which decodes the inputs directly to determine the carry output of a group of the adders. This special decoding provides an alternate and faster path for the carry information. There are various ways of implementing the carry function. Of course, a web search provides examples of both types of adders. The basic idea is to get the benefit of more speed at the cost of more logic.