Where does cache collide?

A small, fast store that holds copies of data from a larger, slower store, so that repeated or nearby accesses are served quickly instead of paying the full cost every time. Smith's 1982 survey Cache Memories set out the design of the CPU cache, where recently used memory lines are kept close to the processor and a hit is served far faster than a trip to main memory; Wang's 1999 A Survey of Web Caching Schemes for the Internet carried the same idea to the web, where proxies keep copies of documents near the user. The referent is a hardware or software buffer, and the unit is a cache holding cached entries. What makes it work is locality of reference, not any hidden intent: nothing is concealed, and no animal is involved.[1][2]

Unit

A cache holding cached copies of data

Recovery means

Retrieval on a hit; on a miss the item is fetched from the backing store and copied in

Measured by

Assessed by hit rate, miss rate, and access latency against a workload

Food caching, the behaviour of storing food in one or more places and retrieving it later, seen in many birds and mammals. Smith and Reichman's 1984 review The Evolution of Food Caching by Birds and Mammals treats it as an adaptation to a variable food supply, and Vander Wall's 1990 Food Hoarding in Animals draws the standard split between larder hoarding, one defended store, and scatter hoarding, many small hidden stores an animal relocates from memory. The referent is an animal's behaviour and the caches it makes, and the unit is one cache or one caching bout. Retrieval depends on spatial memory and on the food surviving pilferage. The store holds nuts and seeds, not data, and the aim is surviving lean seasons, not fast access.[3][4]

Unit

One cache of stored food, or one caching bout

Recovery means

Cache recovery: the animal relocating and retrieving stored food, often from spatial memory

Measured by

Observed as caching and recovery rates, cache spacing, and pilferage in the field