A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms

Hamidreza Khaleghzadeh; Ravi Reddy Manumachu; Alexey Lastovetsky

doi:10.1007/978-3-031-06156-1_14

A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms

Hamidreza Khaleghzadeh^*, Ravi Reddy Manumachu, Alexey Lastovetsky

^*Corresponding author for this work

School of Computing

University College Dublin

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

37 Downloads (Pure)

Abstract

Performance and energy are the two most important objectives for optimization on heterogeneous HPC platforms. In this work, we study a mathematical problem motivated by the bi-objective optimization of a matrix multiplication application on such platforms for performance and energy. We formulate the problem and propose an algorithm of polynomial complexity solving the problem for the case where all the application profiles of objective type one are continuous and strictly increasing, and all the application profiles of objective type two are linear increasing. We solve the problem for the matrix multiplication application employing five heterogeneous processors that include two Intel multicore CPUs, an Nvidia K40c GPU, an Nvidia P100 PCIe GPU, and an Intel Xeon Phi. Based on our experiments, a dynamic energy saving of 17% is gained while tolerating a performance degradation of 5% (a saving of 106 Joules for an execution time increase of 0.05 seconds).

Original language	English
Title of host publication	Euro-Par 2021: Parallel Processing Workshops
Subtitle of host publication	Euro-Par 2021 International Workshops, Lisbon, Portugal, August 30-31, 2021, Revised Selected Papers
Editors	Ricardo Chaves, Dora B. Heras, Aleksandar Ilic, Didem Unat, Rosa M. Badia, Andrea Bracciali, Patrick Diehl, Anshu Dubey, Oh Sangyoon, Stephen L. Scott, Laura Ricci
Publisher	Springer
Pages	166-178
Number of pages	13
ISBN (Electronic)	9783031061561
ISBN (Print)	9783031061554
DOIs	https://doi.org/10.1007/978-3-031-06156-1_14
Publication status	Published - 9 Jun 2022
Event	International Workshop on Algorithms, Models and Tools for Parallel Computing on Heterogeneous Platforms - Lisbon, Portugal Duration: 31 Aug 2021 → 31 Aug 2021 http://heteropar2021.bsc.es/

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Volume	13098
ISSN (Print)	0302-9743

Workshop

Workshop	International Workshop on Algorithms, Models and Tools for Parallel Computing on Heterogeneous Platforms
Abbreviated title	HeteroPar'2021
Country/Territory	Portugal
City	Lisbon
Period	31/08/21 → 31/08/21
Internet address	http://heteropar2021.bsc.es/

Keywords

bi-objective optimization
min-max optimization
minsum optimization
performance optimization
energy optimization

Access to Document

10.1007/978-3-031-06156-1_14

HeteroPar_BOPPE_linear__2_
This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: http://dx.doi.org/10.1007/978-3-031-06156-1_14.
Accepted author manuscript (Post-print), 546 KB

http://heteropar2021.bsc.es/

Cite this

Khaleghzadeh, H., Manumachu, R. R., & Lastovetsky, A. (2022). A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms. In R. Chaves, D. B. Heras, A. Ilic, D. Unat, R. M. Badia, A. Bracciali, P. Diehl, A. Dubey, O. Sangyoon, S. L. Scott, & L. Ricci (Eds.), Euro-Par 2021: Parallel Processing Workshops: Euro-Par 2021 International Workshops, Lisbon, Portugal, August 30-31, 2021, Revised Selected Papers (pp. 166-178). (Lecture Notes in Computer Science; Vol. 13098). Springer. https://doi.org/10.1007/978-3-031-06156-1_14

Khaleghzadeh, Hamidreza ; Manumachu, Ravi Reddy ; Lastovetsky, Alexey. / A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms. Euro-Par 2021: Parallel Processing Workshops: Euro-Par 2021 International Workshops, Lisbon, Portugal, August 30-31, 2021, Revised Selected Papers. editor / Ricardo Chaves ; Dora B. Heras ; Aleksandar Ilic ; Didem Unat ; Rosa M. Badia ; Andrea Bracciali ; Patrick Diehl ; Anshu Dubey ; Oh Sangyoon ; Stephen L. Scott ; Laura Ricci. Springer, 2022. pp. 166-178 (Lecture Notes in Computer Science).

@inproceedings{dfe2c7cc2601423c89b0ce09e48f29aa,

title = "A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms",

abstract = "Performance and energy are the two most important objectives for optimization on heterogeneous HPC platforms. In this work, we study a mathematical problem motivated by the bi-objective optimization of a matrix multiplication application on such platforms for performance and energy. We formulate the problem and propose an algorithm of polynomial complexity solving the problem for the case where all the application profiles of objective type one are continuous and strictly increasing, and all the application profiles of objective type two are linear increasing. We solve the problem for the matrix multiplication application employing five heterogeneous processors that include two Intel multicore CPUs, an Nvidia K40c GPU, an Nvidia P100 PCIe GPU, and an Intel Xeon Phi. Based on our experiments, a dynamic energy saving of 17% is gained while tolerating a performance degradation of 5% (a saving of 106 Joules for an execution time increase of 0.05 seconds).",

keywords = "bi-objective optimization, min-max optimization, minsum optimization, performance optimization, energy optimization",

author = "Hamidreza Khaleghzadeh and Manumachu, {Ravi Reddy} and Alexey Lastovetsky",

year = "2022",

month = jun,

day = "9",

doi = "10.1007/978-3-031-06156-1_14",

language = "English",

isbn = "9783031061554",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "166--178",

editor = "Ricardo Chaves and Heras, {Dora B.} and Aleksandar Ilic and Didem Unat and Badia, {Rosa M.} and Andrea Bracciali and Patrick Diehl and Anshu Dubey and Oh Sangyoon and Scott, {Stephen L.} and Laura Ricci",

booktitle = "Euro-Par 2021: Parallel Processing Workshops",

note = "International Workshop on Algorithms, Models and Tools for Parallel Computing on Heterogeneous Platforms, HeteroPar'2021 ; Conference date: 31-08-2021 Through 31-08-2021",

url = "http://heteropar2021.bsc.es/",

}

Khaleghzadeh, H, Manumachu, RR & Lastovetsky, A 2022, A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms. in R Chaves, DB Heras, A Ilic, D Unat, RM Badia, A Bracciali, P Diehl, A Dubey, O Sangyoon, SL Scott & L Ricci (eds), Euro-Par 2021: Parallel Processing Workshops: Euro-Par 2021 International Workshops, Lisbon, Portugal, August 30-31, 2021, Revised Selected Papers. Lecture Notes in Computer Science, vol. 13098, Springer, pp. 166-178, International Workshop on Algorithms, Models and Tools for Parallel Computing on Heterogeneous Platforms, Lisbon, Portugal, 31/08/21. https://doi.org/10.1007/978-3-031-06156-1_14

A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms. / Khaleghzadeh, Hamidreza; Manumachu, Ravi Reddy; Lastovetsky, Alexey.
Euro-Par 2021: Parallel Processing Workshops: Euro-Par 2021 International Workshops, Lisbon, Portugal, August 30-31, 2021, Revised Selected Papers. ed. / Ricardo Chaves; Dora B. Heras; Aleksandar Ilic; Didem Unat; Rosa M. Badia; Andrea Bracciali; Patrick Diehl; Anshu Dubey; Oh Sangyoon; Stephen L. Scott; Laura Ricci. Springer, 2022. p. 166-178 (Lecture Notes in Computer Science; Vol. 13098).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms

AU - Khaleghzadeh, Hamidreza

AU - Manumachu, Ravi Reddy

AU - Lastovetsky, Alexey

PY - 2022/6/9

Y1 - 2022/6/9

N2 - Performance and energy are the two most important objectives for optimization on heterogeneous HPC platforms. In this work, we study a mathematical problem motivated by the bi-objective optimization of a matrix multiplication application on such platforms for performance and energy. We formulate the problem and propose an algorithm of polynomial complexity solving the problem for the case where all the application profiles of objective type one are continuous and strictly increasing, and all the application profiles of objective type two are linear increasing. We solve the problem for the matrix multiplication application employing five heterogeneous processors that include two Intel multicore CPUs, an Nvidia K40c GPU, an Nvidia P100 PCIe GPU, and an Intel Xeon Phi. Based on our experiments, a dynamic energy saving of 17% is gained while tolerating a performance degradation of 5% (a saving of 106 Joules for an execution time increase of 0.05 seconds).

AB - Performance and energy are the two most important objectives for optimization on heterogeneous HPC platforms. In this work, we study a mathematical problem motivated by the bi-objective optimization of a matrix multiplication application on such platforms for performance and energy. We formulate the problem and propose an algorithm of polynomial complexity solving the problem for the case where all the application profiles of objective type one are continuous and strictly increasing, and all the application profiles of objective type two are linear increasing. We solve the problem for the matrix multiplication application employing five heterogeneous processors that include two Intel multicore CPUs, an Nvidia K40c GPU, an Nvidia P100 PCIe GPU, and an Intel Xeon Phi. Based on our experiments, a dynamic energy saving of 17% is gained while tolerating a performance degradation of 5% (a saving of 106 Joules for an execution time increase of 0.05 seconds).

KW - bi-objective optimization

KW - min-max optimization

KW - minsum optimization

KW - performance optimization

KW - energy optimization

UR - http://heteropar2021.bsc.es/

U2 - 10.1007/978-3-031-06156-1_14

DO - 10.1007/978-3-031-06156-1_14

M3 - Conference contribution

SN - 9783031061554

T3 - Lecture Notes in Computer Science

SP - 166

EP - 178

BT - Euro-Par 2021: Parallel Processing Workshops

A2 - Chaves, Ricardo

A2 - Heras, Dora B.

A2 - Ilic, Aleksandar

A2 - Unat, Didem

A2 - Badia, Rosa M.

A2 - Bracciali, Andrea

A2 - Diehl, Patrick

A2 - Dubey, Anshu

A2 - Sangyoon, Oh

A2 - Scott, Stephen L.

A2 - Ricci, Laura

PB - Springer

T2 - International Workshop on Algorithms, Models and Tools for Parallel Computing on Heterogeneous Platforms

Y2 - 31 August 2021 through 31 August 2021

ER -

Khaleghzadeh H, Manumachu RR, Lastovetsky A. A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms. In Chaves R, Heras DB, Ilic A, Unat D, Badia RM, Bracciali A, Diehl P, Dubey A, Sangyoon O, Scott SL, Ricci L, editors, Euro-Par 2021: Parallel Processing Workshops: Euro-Par 2021 International Workshops, Lisbon, Portugal, August 30-31, 2021, Revised Selected Papers. Springer. 2022. p. 166-178. (Lecture Notes in Computer Science). doi: 10.1007/978-3-031-06156-1_14

A novel algorithm for bi-objective performance-energy optimization of applications with continuous performance and linear energy profiles on heterogeneous HPC platforms

Abstract

Publication series

Workshop

Keywords

Access to Document

Fingerprint

Cite this