This is a minor detail, but why do we see the peaks and dips in time as we increase the number of processors? For example, why does it take less time to acquire a lock when there are 6, 7, or 8 processors compared to when there are 5 processors. Does the amount of cache coherence traffic in the bus strictly increase?

Never mind, I just got to that point in lecture - Kayvon’s answer is that real-world measurements can be noisy!