Multi objective optimization of HPC kernels for performance, power, and energy

Prasanna Balaprakash, Ananta Tiwari, Stefan M. Wild

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

38 Scopus citations

Abstract

Code optimization in the high-performance computing realm has traditionally focused on reducing execution time. The problem, in mathematical terms, has been expressed as a single objective optimization problem. The expected concerns of next-generation systems, however, demand a more detailed analysis of the interplay among execution time and other metrics. Metrics such as power, performance, energy, and resiliency may all be targeted together and traded against one another. We present a multi objective formulation of the code optimization problem. Our proposed framework helps one explore potential tradeoffs among multiple objectives and provides a significantly richer analysis than can be achieved by treating additional metrics as hard constraints. We empirically examine a variety of metrics, architectures, and code optimization decisions and provide evidence that such tradeoffs exist in practice.

Original languageEnglish
Title of host publicationHigh Performance Computing Systems
Subtitle of host publicationPerformance Modeling, Benchmarking and Simulation - 4th International Workshop, PMBS 2013, Revised Selected Papers
EditorsStephen A. Jarvis, Steven A. Wright, Simon D. Hammond
PublisherSpringer Verlag
Pages239-260
Number of pages22
ISBN (Electronic)9783319102139
DOIs
StatePublished - 2014
Externally publishedYes
Event4th International Workshop on Performance Modeling, Benchmarking and Simulation of High-Performance Computing Systems, PMBS 2013 - Denver, United States
Duration: Nov 18 2013Nov 18 2013

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8551
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference4th International Workshop on Performance Modeling, Benchmarking and Simulation of High-Performance Computing Systems, PMBS 2013
Country/TerritoryUnited States
CityDenver
Period11/18/1311/18/13

Fingerprint

Dive into the research topics of 'Multi objective optimization of HPC kernels for performance, power, and energy'. Together they form a unique fingerprint.

Cite this