Amdahl's law


Also found in: Dictionary, Thesaurus, Legal, Wikipedia.

Amdahl's law

[′am‚dälz ‚lȯ]
(computer science)
A law stating that the speed-up that can be achieved by distributing a computer program over p processors cannot exceed 1/{f + [1 -f)/ p ]}, where f is the fraction of the work of the program that must be done in serial mode.

Amdahl's Law

(parallel)
(Named after Gene Amdahl) If F is the fraction of a calculation that is sequential, and (1-F) is the fraction that can be parallelised, then the maximum speedup that can be achieved by using P processors is 1/(F+(1-F)/P).

[Gene Amdahl, "Validity of the Single Processor Approach to Achieving Large-Scale Computing Capabilities", AFIPS Conference Proceedings, (30), pp. 483-485, 1967].

Amdahl's law

"Overall system speed is governed by the slowest component," coined by Gene Amdahl, chief architect of IBM's first mainframe series and founder of Amdahl Corporation and other companies. Amdahl's law applied to networking. The slowest device in the network will determine the maximum speed of the network. See laws.
Mentioned in ?
References in periodicals archive ?
However, traditional Amdahl's law only considers the cost of instruction execution.
Therefore, we can extend Amdahl's law to be a new equation called "Enhanced Amdahl's law".
In conclusion, communication overhead should be considered in the Extended Amdahl's Law.
7, we find that data synchronization and communication affect Amdahl's Law in the hetergeneous architecture.
2 Extending Amdahl's law in the multicore era [11], [12]
Amdahl's law assumes that the problem size is fixed and provides a fixed-size speed-up model.
5 Amdahl's law for predicting the future of multicores considered harmful [17]
Zhou's central thesis is that Amdahl's law is the pessimistic extreme, and Sandia's law is the optimistic extreme.
The fallacy of assuming linear speedup is the ultimate goal that underlies Amdahl's law and two decades of misguided papers written on parallel processor performance evaluation.
The results show our proposed load balancing technique can overcome the performance limitations of Amdahl's law.
Based on the estimated results, we confirmed that the proposed approach can provide better speedup than Amdahl's law computed with typical load balancing as the number of cores increased.
By distributing the first module asymmetrically and keeping the data dependency in the second module, we can provide better performance than Amdahl's law computed with typical load balancing.