Studying the Impact of Power Capping on MapReduce-based, Data-intensive Mini-applications on Intel KNL and KNM Architectures

Joshua Hoke Davis; Tao Gao; Sunita Chandresekaran; Michela Taufer

Studying the Impact of Power Capping on MapReduce-based, Data-intensive Mini-applications on Intel KNL and KNM Architectures

Distributed, Parallel, and Cluster Computing 2019-03-29 v1

Authors: Joshua Hoke Davis , Tao Gao , Sunita Chandresekaran , Michela Taufer

Abstract

In this poster, we quantitatively measure the impacts of data movement on performance in MapReduce-based applications when executed on HPC systems. We leverage the PAPI 'powercap' component to identify ideal conditions for execution of our applications in terms of (1) dataset characteristics (i.e., unique words); (2) HPC system (i.e., KNL and KNM); and (3) implementation of the MapReduce programming model (i.e., with or without combiner optimizations). Results confirm the high energy and runtime costs of data movement, and the benefits of the combiner optimization on these costs.

Keywords

power capping mapreduce data-intensive mini-applications intel knl intel knm data movement papi

Cite

@article{arxiv.1903.11694,
  title  = {Studying the Impact of Power Capping on MapReduce-based, Data-intensive Mini-applications on Intel KNL and KNM Architectures},
  author = {Joshua Hoke Davis and Tao Gao and Sunita Chandresekaran and Michela Taufer},
  journal= {arXiv preprint arXiv:1903.11694},
  year   = {2019}
}

Comments

Extended abstract submitted for the ACM Student Research Competition at SC18. 2nd place undergraduate poster

Studying the Impact of Power Capping on MapReduce-based, Data-intensive Mini-applications on Intel KNL and KNM Architectures

Abstract

Keywords

Cite

Comments

Related papers