We describe a high performance clocking methodology for domino pipelines. Our technique maximizes the clock rate of the circular pipeline (“ring”) while maintaining the ring cycle time to be the worst-case combinational logic delay around the ring. It is relatively immune to global clock skew, incurs no latch overhead, allows up to 50% time borrowing, and offers a robust way of preventing race-through problems, adjusted for the worst-case time borrowing.
Kenneth Y. Yun, Ayoob E. Dooply