Characterizing the performance and energy attributes of scientific simulations

Sayaka Akioka, Konrad Malkowski, Padma Raghavan, Mary Jane Irwin, Lois Curfman McInnes, Boyana Norris

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We characterize the performance and energy attributes of scientific applications based on nonlinear partial differential equations (PDEs). where the dominant cost is that of sparse linear system solution. We obtain performance and energy metrics using cycle-accurate emulations on a processor and memory system derived from the PowerPC RISC architecture with extensions to resemble the processor in the BlueGene/L. These results indicate that low-power modes of CPUs such as Dynamic Voltage Scaling (DVS) can indeed result in energy savings at the expense of performance degradation. We then consider the impact of certain memory subsystem optimizations to demonstrate that these optimizations in conjunction with DVS can provide faster execution time and lower energy consumption. For example, on the optimized architecture, if DVS is used to scale down the processor to 600MHz, execution times are faster by 45% with energy reductions of 75% compared to the original architecture at IGHz. The insights gained from this study can help scientific applications better utilize the low-power modes of processors as well as guide the selection of hardware optimizations in future power-aware, high-performance computers.

Original languageEnglish (US)
Title of host publicationComputational Science - ICCS 2006
Subtitle of host publication6th International Conference, Proceedings
PublisherSpringer Verlag
Pages242-249
Number of pages8
ISBN (Print)3540343792, 9783540343790
DOIs
StatePublished - Jan 1 2006
EventICCS 2006: 6th International Conference on Computational Science - Reading, United Kingdom
Duration: May 28 2006May 31 2006

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume3991 LNCS - I
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

OtherICCS 2006: 6th International Conference on Computational Science
CountryUnited Kingdom
CityReading
Period5/28/065/31/06

Fingerprint

Dynamic Voltage Scaling
Attribute
Execution Time
Optimization
Energy
Data storage equipment
Reduced instruction set computing
Sparse Linear Systems
Simulation
Emulation
Energy Saving
Nonlinear Partial Differential Equations
Partial differential equations
Program processors
Energy Consumption
Linear systems
Energy conservation
Subsystem
Degradation
Computer systems

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Akioka, S., Malkowski, K., Raghavan, P., Irwin, M. J., McInnes, L. C., & Norris, B. (2006). Characterizing the performance and energy attributes of scientific simulations. In Computational Science - ICCS 2006: 6th International Conference, Proceedings (pp. 242-249). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3991 LNCS - I). Springer Verlag. https://doi.org/10.1007/11758501_36
Akioka, Sayaka ; Malkowski, Konrad ; Raghavan, Padma ; Irwin, Mary Jane ; McInnes, Lois Curfman ; Norris, Boyana. / Characterizing the performance and energy attributes of scientific simulations. Computational Science - ICCS 2006: 6th International Conference, Proceedings. Springer Verlag, 2006. pp. 242-249 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{073666cafa014901952e0644a556fb2e,
title = "Characterizing the performance and energy attributes of scientific simulations",
abstract = "We characterize the performance and energy attributes of scientific applications based on nonlinear partial differential equations (PDEs). where the dominant cost is that of sparse linear system solution. We obtain performance and energy metrics using cycle-accurate emulations on a processor and memory system derived from the PowerPC RISC architecture with extensions to resemble the processor in the BlueGene/L. These results indicate that low-power modes of CPUs such as Dynamic Voltage Scaling (DVS) can indeed result in energy savings at the expense of performance degradation. We then consider the impact of certain memory subsystem optimizations to demonstrate that these optimizations in conjunction with DVS can provide faster execution time and lower energy consumption. For example, on the optimized architecture, if DVS is used to scale down the processor to 600MHz, execution times are faster by 45{\%} with energy reductions of 75{\%} compared to the original architecture at IGHz. The insights gained from this study can help scientific applications better utilize the low-power modes of processors as well as guide the selection of hardware optimizations in future power-aware, high-performance computers.",
author = "Sayaka Akioka and Konrad Malkowski and Padma Raghavan and Irwin, {Mary Jane} and McInnes, {Lois Curfman} and Boyana Norris",
year = "2006",
month = "1",
day = "1",
doi = "10.1007/11758501_36",
language = "English (US)",
isbn = "3540343792",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "242--249",
booktitle = "Computational Science - ICCS 2006",
address = "Germany",

}

Akioka, S, Malkowski, K, Raghavan, P, Irwin, MJ, McInnes, LC & Norris, B 2006, Characterizing the performance and energy attributes of scientific simulations. in Computational Science - ICCS 2006: 6th International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 3991 LNCS - I, Springer Verlag, pp. 242-249, ICCS 2006: 6th International Conference on Computational Science, Reading, United Kingdom, 5/28/06. https://doi.org/10.1007/11758501_36

Characterizing the performance and energy attributes of scientific simulations. / Akioka, Sayaka; Malkowski, Konrad; Raghavan, Padma; Irwin, Mary Jane; McInnes, Lois Curfman; Norris, Boyana.

Computational Science - ICCS 2006: 6th International Conference, Proceedings. Springer Verlag, 2006. p. 242-249 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3991 LNCS - I).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Characterizing the performance and energy attributes of scientific simulations

AU - Akioka, Sayaka

AU - Malkowski, Konrad

AU - Raghavan, Padma

AU - Irwin, Mary Jane

AU - McInnes, Lois Curfman

AU - Norris, Boyana

PY - 2006/1/1

Y1 - 2006/1/1

N2 - We characterize the performance and energy attributes of scientific applications based on nonlinear partial differential equations (PDEs). where the dominant cost is that of sparse linear system solution. We obtain performance and energy metrics using cycle-accurate emulations on a processor and memory system derived from the PowerPC RISC architecture with extensions to resemble the processor in the BlueGene/L. These results indicate that low-power modes of CPUs such as Dynamic Voltage Scaling (DVS) can indeed result in energy savings at the expense of performance degradation. We then consider the impact of certain memory subsystem optimizations to demonstrate that these optimizations in conjunction with DVS can provide faster execution time and lower energy consumption. For example, on the optimized architecture, if DVS is used to scale down the processor to 600MHz, execution times are faster by 45% with energy reductions of 75% compared to the original architecture at IGHz. The insights gained from this study can help scientific applications better utilize the low-power modes of processors as well as guide the selection of hardware optimizations in future power-aware, high-performance computers.

AB - We characterize the performance and energy attributes of scientific applications based on nonlinear partial differential equations (PDEs). where the dominant cost is that of sparse linear system solution. We obtain performance and energy metrics using cycle-accurate emulations on a processor and memory system derived from the PowerPC RISC architecture with extensions to resemble the processor in the BlueGene/L. These results indicate that low-power modes of CPUs such as Dynamic Voltage Scaling (DVS) can indeed result in energy savings at the expense of performance degradation. We then consider the impact of certain memory subsystem optimizations to demonstrate that these optimizations in conjunction with DVS can provide faster execution time and lower energy consumption. For example, on the optimized architecture, if DVS is used to scale down the processor to 600MHz, execution times are faster by 45% with energy reductions of 75% compared to the original architecture at IGHz. The insights gained from this study can help scientific applications better utilize the low-power modes of processors as well as guide the selection of hardware optimizations in future power-aware, high-performance computers.

UR - http://www.scopus.com/inward/record.url?scp=33746602614&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33746602614&partnerID=8YFLogxK

U2 - 10.1007/11758501_36

DO - 10.1007/11758501_36

M3 - Conference contribution

SN - 3540343792

SN - 9783540343790

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 242

EP - 249

BT - Computational Science - ICCS 2006

PB - Springer Verlag

ER -

Akioka S, Malkowski K, Raghavan P, Irwin MJ, McInnes LC, Norris B. Characterizing the performance and energy attributes of scientific simulations. In Computational Science - ICCS 2006: 6th International Conference, Proceedings. Springer Verlag. 2006. p. 242-249. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/11758501_36