I write hardware descriptions for FPGAs. In a normal week I lose a day to a path that misses timing by 80 picoseconds and gain it back by moving one register.
Interested in verification, clock domain crossings and the many ways a design that simulates perfectly can fail on the board.