The simplest way is splitting it to the upper 4 bit addition and lower 4 bit addition. THe lower 4 bits are simply added. Upper 4 bits have 2 sets of adder, one has 0 as an carry in, another has 1 as an carry-in. and the result from one of those 2 adders are selected by carry-out from lower 4 bits. THe critical path should be roughly with 10 levels of logic.
There are many addition algorithms, but having only 8 bits, i don't think they make much difference in speed. I guess carry select adder like above should be enough.