Least Recently Used algorithm

From Computer History Wiki
Revision as of 13:01, 15 January 2024 by Jnc (talk | contribs) (Used in more places than just replacement)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

The Least Recently Used algorithm (usually abbreviated as LRU) is an algorithm for choosing which entity to deal with first; be it the entry to 'recycle' in an array of resources, when a need arises for a free resource (and one has to be replaced); or for some other situation.

In replacement situations it says, as the name implies, that when the entries are all already in use, one should recycle the entry which was most recently used the longest time ago, since the ones used since then are more likely to be used again soon.

The reasoning behind minimizing the number of times entries are freed is that freeing a resource likely involves some effort (e.g. writing a modified disk block from a buffer back onto secondary storage), so that effort should be expended are rarely as possible.

Providing the information needed to determine which entry is the least recently used entry may require some effort, in both hardware and software systems. There are thus algorithms which approximate 'true' LRU, but use less overhead.

LRU replacement algorithms are used in many circumstances: disk buffers (as above); page frames (in virtual memory systems); and caches of all kinds.

In other situations (such as arbitration for use of a bus), a similar algorithm is used - pick whichever was most recently picked the longest time ago.