No gate is instant. When an input changes, the output follows a short time later: that's its propagation delay. To compare designs without worrying about exact nanoseconds, we count in gate delays: the number of gates a signal passes through on its way from an input to an output.
The model used in this course's adder lessons:
- every gate, XOR included, takes 1 gate delay
- all inputs (every A and B bit, and C0) arrive at time 0
- a gate's output is ready one delay after its latest input is ready
The slowest input-to-output path is the critical path, and its length is the circuit's worst-case delay.
The model is a simplification. Real gates differ (an XOR is slower than a NAND), and wide gates are slower than narrow ones. But it captures the key comparison: a ripple carry adder needs about 2 gate delays per bit, so its delay grows with the width, while a carry lookahead adder needs a fixed number of levels.
Read each question for what it counts. Some count only the carry path from C0 (2 per stage, 2n in total). Others count from the A and B inputs, which adds 1 for the first XOR (2n + 1).