Can someone explain why the cache timing pattern gets such a strong signal for something so seemingly distant? Is this about memory locality or just the effects

Can someone explain why the cache timing pattern gets such a strong signal for something so seemingly distant? Is this about memory locality or just the effects of a different “CPU workload” in general? Also, what JS APIs are used to carry out such high resolution time measurements?