Deque-Free Work-Optimal Parallel STL Algorithms

Traoré, Daouda; Roch, Jean‐Louis; Maillard, Nicolas; Gautier, Thierry; Bernard, Julien

doi:10.1007/978-3-540-85451-7_95

Cited by 22 publications

(15 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Adapting the number of created parallel tasks to dynamically fit the number of available resources is the key point to reach high performance. With an other approach for implementing this adaptation, we have proposed this on demand task creation to build coarse grain parallel adaptive algorithms for most of the STL algorithms [27]. Here, the proposed solution extends the task model for a much more finer integration with the scheduler.…”

Section: Adaptive Task Modelmentioning

confidence: 99%

See 1 more Smart Citation

X-kaapi: A Multi Paradigm Runtime for Multicore Architectures

Gautier

Lementec

Faucher

et al. 2013

2013 42nd International Conference on Parallel Processing

Self Cite

View full text Add to dashboard Cite

Abstract-The paper presents X-KAAPI, a compact runtime for multicore architectures that brings multi parallel paradigms (parallel independent loops, fork-join tasks and dataflow tasks) in a unified framework without performance penalty. Comparisons on independent loops with OpenMP and on dense linear algebra with QUARK/PLASMA confirm our design decisions. Applied to EUROPLEXUS, an industrial simulation code for fast transient dynamics, we show that X-KAAPI achieves high speedups on multicore architectures by efficiently parallelizing both independent loops and dataflow tasks.

show abstract

Section: Adaptive Task Modelmentioning

confidence: 99%

“…It allows for simple and efficient synchronization protocols. Moreover, for applications developers, a set of higher parallel algorithms, like those of the STL [27], are proposed on top of the adaptive task model. Next section focuses on the parallel foreach algorithm.…”

Section: Adaptive Task Modelmentioning

confidence: 99%

X-kaapi: A Multi Paradigm Runtime for Multicore Architectures

Gautier

Lementec

Faucher

et al. 2013

2013 42nd International Conference on Parallel Processing

Self Cite

View full text Add to dashboard Cite

show abstract

“…This assumption is not restrictive as the description of a large number of tasks can be very short. In the case of independent tasks, a whole subpart of an array of tasks can be represented in a compact way by the range of the corresponding indices, each cell containing the effective description of a task (a STL transform in Traoré et al (2008)). For more general cases with precedence constraints, it is usually enough to transfer a task which represents a part of the DAG.…”

Section: Model Of the Distributed Listmentioning

confidence: 99%

Decentralized list scheduling

2012

View full text Add to dashboard Cite

Classical list scheduling is a very popular and efficient technique for scheduling jobs for parallel and distributed platforms. It is inherently centralized. However, with the increasing number of processors, the cost for managing a single centralized list becomes too prohibitive. A suitable approach to reduce the contention is to distribute the list among the computational units: each processor only has a local view of the work to execute. Thus, the scheduler is no longer greedy and standard performance guarantees are lost. The objective of this work is to study the extra cost that must be paid when the list is distributed among the computational units. We first present a general methodology for computing the expected makespan based on the analysis of an adequate potential function which represents the load imbalance between the local lists. We obtain an equation giving the evolution of the potential by computing its expected decrease in one step of the schedule. Our main theorem shows how to solve such equations to bound the makespan. Then, we apply this method to several scheduling problems, namely, for unit independent tasks, for weighted independent tasks and for tasks with precedence constraints. More precisely, we prove that the time for scheduling a global workload W composed of independent unit tasks on m processors is equal to W/m plus an additional term proportional to log 2 W . We provide a lower bound which shows that this is optimal up to a constant. This result is extended to the case of weighted independent tasks. In the last setting, precedence task graphs, our analysis leads to an improvement on the bound of Arora et al. (2001). We end with some experiments using a simulator. The distribution of the makespan is shown to fit existing probability laws. Moreover, the simulations give a better insight into the additive term whose value is shown to be around 3 log 2 W confirming the precision of our analysis.

show abstract

“…This assumption is not restrictive: for the case of independent tasks, the description of a large number of tasks can be very short. For instance a whole subpart of an array of tasks can be represented in a compact way by the range of the corresponding indices, each cell containing the effective description of a task (a STL transform in [13]). For more general cases with dependencies, it is usually enough to transfer a task which represents a part of the graph [7].…”

Section: Model Of the Distributed Listmentioning

confidence: 99%

A Tighter Analysis of Work Stealing

Tchiboukdjian

Gast

Trystram

et al. 2010

Algorithms and Computation

Self Cite

View full text Add to dashboard Cite

Abstract. Classical list scheduling is a very popular and efficient technique for scheduling jobs in parallel platforms. However, with the increasing number of processors, the cost for managing a single centralized list becomes prohibitive. The objective of this work is to study the extra cost that must be paid when the list is distributed among the processors. We present a general methodology for computing the expected makespan based on the analysis of an adequate potential function which represents the load unbalance between the local lists. A bound on the deviation from the mean is also derived. Then, we apply this technique to show that the expected makespan for scheduling W unit independent tasks on m processors is equal to W/m with an additional term in 3.65 log 2 W . Moreover, simulations show that our bound is very close to the exact value, approximately 50% off. This new analysis also enables to study the influence of the initial repartition of tasks and the reduction of the number of steals when several thieves can simultaneously steal work in the same processor's list.

show abstract

Deque-Free Work-Optimal Parallel STL Algorithms

Cited by 22 publications

References 10 publications

X-kaapi: A Multi Paradigm Runtime for Multicore Architectures

X-kaapi: A Multi Paradigm Runtime for Multicore Architectures

Decentralized list scheduling

A Tighter Analysis of Work Stealing

Contact Info

Product

Resources

About