An adaptive task creation strategy for work-stealing scheduling

Wang, Lei; Cui, Huimin; Duan, Yuelu; Lu, Fang; Feng, Xiaobing; Yew, Pen-Chung

doi:10.1145/1772954.1772992

Cited by 30 publications

(14 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SLAW [18] is a work-stealing scheduler with adaptive scheduling policies based on locality information. AdaptiveTC [36] improves system performance through adaptive thread management at thread creation time. BWS [14] improves system throughput and fairness in time-sharing multicore systems.…”

Section: Related Workmentioning

confidence: 99%

Energy-efficient work-stealing language runtimes

Ribic

Liu

2014

Proceedings of the 19th International Conference on Architectural Support for Programming Languages and Operating Systems

View full text Add to dashboard Cite

Work stealing is a promising approach to constructing multithreaded program runtimes of parallel programming languages. This paper presents HERMES, an energy-efficient work-stealing language runtime. The key insight is that threads in a work-stealing environment -thieves and victims -have varying impacts on the overall program running time, and a coordination of their execution "tempo" can lead to energy efficiency with minimal performance loss. The centerpiece of HERMES is two complementary algorithms to coordinate thread tempo: the workpath-sensitive algorithm determines tempo for each thread based on thief-victim relationships on the execution path, whereas the workload-sensitive algorithm selects appropriate tempo based on the size of work-stealing deques. We construct HERMES on top of Intel Cilk Plus's runtime, and implement tempo adjustment through standard Dynamic Voltage and Frequency Scaling (DVFS). Benchmarks running on HERMES demonstrate an average of 11-12% energy savings with an average of 3-4% performance loss through meter-based measurements over commercial CPUs.

show abstract

Section: Related Workmentioning

confidence: 99%

Energy-efficient work-stealing language runtimes

Ribic

Liu

2014

Proceedings of the 19th International Conference on Architectural Support for Programming Languages and Operating Systems

View full text Add to dashboard Cite

show abstract

“…These right nodes may be good candidate to become BOB-node. This idea has been already proposed in [19]. The goal is to be able to create BOB-node as close as possible to the root of the search tree.…”

Section: B Anticipated Dynamic Load Balancingmentioning

confidence: 99%

Anticipated Dynamic Load Balancing Strategy to Parallelize Constraint Programming Search

Menouer

Cun

2013

2013 IEEE International Symposium on Parallel &Amp; Distributed Processing, Workshops and PHD Forum

View full text Add to dashboard Cite

This paper 1 presents a parallelization of the Constraint Programming solver OR-Tools, using the parallel framework Bobpp. The principle of the parallelization is to assign one sequential OR-Tools solver per core and to communicate work or nodes using the Bobpp global priority queue. As usual the communication or migration of work between solvers is a kind of Work Stealing. But here we have to deal with a specific overcost. By using OR-Tools, it is possible to record the path from the root of the tree to a node so as to stop the search at a precise node. However, to start the search on a sub-tree, the entire path from the root of the main tree to the root of the sub-tree is to be replayed. This leads to additional costs to the parallel search. This paper presents new strategies using a Work Stealing technique to choose good nodes of the searchspace and assigns each sub-tree during the execution of the algorithm in order to reduce the extra costs due to OR-Tools solver.These strategies are tested with problem modeled in ORTools for imbalanced search tree, despite the additional costs add by OR-Tools a speed-up of 8.04 must to be reached with N-Queens problem using 12 cores and speed-up of 37.73 with Golomb Ruler problem using 48 cores.

show abstract

“…Work-stealing Deques Several prior studies [4,5,14,20] use cut-off strategies to control the recursion depth of function calls during the task generation. This is intended to reduce the overhead of task creation and deque operations.…”

Section: Related Workmentioning

confidence: 99%

Work-stealing without the baggage

Kumar

Frampton

Blackburn

et al. 2012

Proceedings of the ACM International Conference on Object Oriented Programming Systems Languages and Applications

View full text Add to dashboard Cite

Work-stealing is a promising approach for effectively exploiting software parallelism on parallel hardware. A programmer who uses work-stealing explicitly identifies potential parallelism and the runtime then schedules work, keeping otherwise idle hardware busy while relieving overloaded hardware of its burden. Prior work has demonstrated that work-stealing is very effective in practice. However, workstealing comes with a substantial overhead: as much as 2× to 12× slowdown over orthodox sequential code.In this paper we identify the key sources of overhead in work-stealing schedulers and present two significant refinements to their implementation. We evaluate our workstealing designs using a range of benchmarks, four different work-stealing implementations, including the popular fork-join framework, and a range of architectures. On these benchmarks, compared to orthodox sequential Java, our fastest design has an overhead of just 15%. By contrast, fork-join has a 2.3× overhead and the previous implementation of the system we use has an overhead of 4.1×. These results and our insight into the sources of overhead for workstealing implementations give further hope to an already promising technique for exploiting increasingly available hardware parallelism.

show abstract

An adaptive task creation strategy for work-stealing scheduling

Cited by 30 publications

References 11 publications

Energy-efficient work-stealing language runtimes

Energy-efficient work-stealing language runtimes

Anticipated Dynamic Load Balancing Strategy to Parallelize Constraint Programming Search

Work-stealing without the baggage

Contact Info

Product

Resources

About