Exploiting multicore systems with Cilk

Authors:
Stephen Lewin-Berlin
Affiliations:
Quanta Research Cambridge, Cambridge, MA
Venue:
Proceedings of the 4th International Workshop on Parallel and Symbolic Computation
Year:
2010

Citing 4
Cited 0

Space-efficient scheduling of multithreaded computations

STOC '93 Proceedings of the twenty-fifth annual ACM symposium on Theory of computing
Scheduling multithreaded computations by work stealing

Journal of the ACM (JACM)
Reducers and other Cilk++ hyperobjects

Proceedings of the twenty-first annual symposium on Parallelism in algorithms and architectures
The Cilkview scalability analyzer

Proceedings of the twenty-second annual ACM symposium on Parallelism in algorithms and architectures

Quantified Score

Hi-index	0.00

Visualization

Abstract

The increasing prevalence of multicore processors has led to a renewed interest in parallel programming. Cilk is a language extension to C and C++ designed to simplify programming shared-memory multiprocessor systems. The workstealing scheduler in Cilk is provably efficient and maintains well-defined space bounds. [1, 2] A deterministic program (that is, a race-free Cilk program that uses no lock constructs) maintains serial semantics, and such a Cilk program running on P processors will use no more than P times the stack space required by the corresponding serial program.