[Sharlin]

They’ve both had the exact same cost (both throughput and latency) on just about any ALU designed in the last several decades. There are no separate steps involved in the function of a full adder.