Skip to content
scrambled
Parallel And Distributed Programming Concept

Consensus Number

Depends on JourneyParallel And Distributed Programming Statuslearning Tags
  • parallel programming

The maximum number of processes for which a given synchronization primitive can solve the consensus problem in a wait-free manner. Defined by Herlihy's hierarchy.

Examples: Read/write registers have consensus number 1. Fetch-and-Add and Fetch-and-Set have consensus number 2. Compare-and-Swap and Load-Link/Store-Conditional have consensus number ∞, making them universal.