Amber condition

The Bernstein condition is a term from computer science , especially from the field of multiprocessing , and describes the conditions under which two program sections produce the same result when executed in parallel as when executed sequentially.

Two program sections and are given . The set of variables that the section has read access to is given. Analogous to this denotes the quantities of variables that are changed by Section during execution. ${\displaystyle P_{1}}$ ${\displaystyle P_{2}}$ ${\displaystyle P_{i}}$ ${\displaystyle I_{i}}$ ${\displaystyle O_{i}}$ ${\displaystyle P_{i}}$

The Bernstein condition now states that the sections and can be executed in parallel without changing the result of this or subsequent calculations, if ${\displaystyle P_{1}}$ ${\displaystyle P_{2}}$

• ${\displaystyle I_{1}\cap O_{2}=\emptyset }$,
• ${\displaystyle I_{2}\cap O_{1}=\emptyset }$, and
• ${\displaystyle O_{1}\cap O_{2}=\emptyset }$

applies.

See also

• Concurrency
• Parallel programming

literature

• AJ Bernstein: Analysis of Programs for Parallel Processing . In: IEEE Transactions on Electronic Computers . EC-15, no. 5 , October 1, 1966, ISSN  0367-7508 , p. 757-763 , doi : 10.1109 / PGEC.1966.264565 .

Individual evidence

1. ↑ B. Chapman, GR Gao, M. Sato, E. Ayguadé, D. Wang (eds.): A Practical Programming Model for the Multi-Core Era - Springer . Springer-Verlag, 2008, p. 200 , doi : 10.1007 / 978-3-540-69303-1 ( limited preview in Google Book search).