Hierarchical task mapping for parallel applications on supercomputers

Wu, Jingjin; Xiong, Xuanxing; Lan, Zhiling

doi:10.1007/s11227-014-1324-5

Cited by 18 publications

(7 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, in these existing works, only the network topology is taken into account and the node's internal architecture is left unaddressed when performance gains are expected from exploiting the memory hierarchy. Wu et al (2015) introduced a hierarchical task mapping strategy for modern supercomputers based on generic recursive algorithms for both fat-tree and torus network topologies showing very good performance with low overhead. Rashti et al (2011) proposed a weighted graph model for the whole physical topology of the computing system, including both the inter-and intranode topologies.…”

Section: Related Work and Discussionmentioning

confidence: 99%

Topology-aware job mapping

Georgiou

Jeannot

Mercier

et al. 2017

The International Journal of High Performance Computing Applica

View full text Add to dashboard Cite

A Resource and Job Management System (RJMS) is a crucial system software part of the HPC stack. It is responsible for e ciently delivering computing power to applications in supercomputing environments. Its main intelligence relies on resource selection techniques to find the most adapted resources to schedule the users' jobs. This paper introduces a new method that takes into account the topology of the machine and the application characteristics to determine the best choice among the available nodes of the platform, based upon the network topology and taking into account the applications communication pattern. To validate our approach, we integrate this algorithm as a plugin for Slurm, a well-known and widespread RJMS. We assess our plugin with di↵erent optimization schemes by comparing with the default topologyaware Slurm algorithm, using both emulation and simulation of a large-scale platform and by carrying out experiments in a real cluster. We show that transparently taking into account a job communication pattern and the topology allows for relevant performance gains.

show abstract

Section: Related Work and Discussionmentioning

confidence: 99%

Topology-aware job mapping

Georgiou

Jeannot

Mercier

et al. 2017

The International Journal of High Performance Computing Applica

View full text Add to dashboard Cite

show abstract

“…They used some application sets to produce some interesting evaluation results, which however are not tested with real workload traces or integrated with real resource and job management systems (RJMSs). Jingjin Wu et al [24] used generic recursive algorithms and proposed a hierarchical application mapping algorithm for both the torus and fat-tree network topologies of modern supercomputers. In the evaluation, they showed good performance of the proposal with low overhead.…”

Section: Topology-specific Application Mappingmentioning

confidence: 99%

Application Mapping and Scheduling of Uncertain Communication Patterns onto Non-Random and Random Network Topologies

Koibuchi

2020

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

Due to recent technology progress based on big-data processing, many applications present irregular or unpredictable communication patterns among compute nodes in high-performance computing (HPC) systems. Traditional communication infrastructures, e.g., torus or fat-tree interconnection networks, may not handle well their matchmaking problems with these newly emerging applications. There are already many communication-efficient application mapping algorithms for these typical non-random network topologies, which use nearby compute nodes to reduce the network distances. However, for the above unpredictable communication patterns, it is difficult to efficiently map their applications onto the non-random network topologies. In this context, we recommend using random network topologies as the communication infrastructures, which have drawn increasing attention for the use of HPC interconnects due to their small diameter and average shortest path length (ASPL). We make a comparative study to analyze the impact of application mapping performance on non-random and random network topologies. We propose using topology embedding metrics, i.e., diameter and ASPL, and list several diameter/ASPL-based application mapping algorithms to compare their job scheduling performances, assuming that the communication pattern of each application is unpredictable to the computing system. Evaluation with a large compound application workload shows that, when compared to nonrandom topologies, random topologies can reduce the average turnaround time up to 39.3% by a random connected mapping method and up to 72.1% by a diameter/ASPL-based mapping algorithm. Moreover, when compared to the baseline topology mapping method, the proposed diameter/ASPLbased topology mapping strategy can reduce up to 48.0% makespan and up to 78.1% average turnaround time, and improve up to 1.9x system utilization over random topologies.

show abstract

“…Jingjin Wu et al [19] proposed a strategy for the mapping of the hierarchical task that implements inter and intra node mapping. They considered supercomputers with torus network and fat-tree topologies, additionally providing two mapping algorithms.…”

Section: Background and Related Workmentioning

confidence: 99%

Topology-Aware Mapping Techniques for Heterogeneous HPC Systems: A Systematic Survey

Alotaibi¹,

alboraei²

2018

ijacsa

View full text Add to dashboard Cite

At the present time, the modern platforms of highperformance computing (HPC) consists of heterogeneous computing devices which are connected through complex hierarchical networks. Moreover, it is moving towards the Exascale era and which makes the number of nodes to increase as well as the number of cores within a node to increase. As a consequence, the communication costs and the data movement are increasing. Given that, the efficient topology-aware process mapping has become vital to efficiently optimize the data locality management in order to improve the system performance and energy consumption. It will also decrease the communication cost of the processes by matching the application virtual topology (exploited by the system for assigning the processes to the physical processor) to the target underlying hardware architecture called physical topology. Additionally, improving the locality problem which is one of the most challenging issues faced by the current parallel applications. In this survey paper, we have studied various topology-aware mapping techniques and algorithms.

show abstract

Hierarchical task mapping for parallel applications on supercomputers

Cited by 18 publications

References 32 publications

Topology-aware job mapping

Topology-aware job mapping

Application Mapping and Scheduling of Uncertain Communication Patterns onto Non-Random and Random Network Topologies

Topology-Aware Mapping Techniques for Heterogeneous HPC Systems: A Systematic Survey

Contact Info

Product

Resources

About