ScaleML: Machine Learning based Heap Memory Object Scaling Prediction

Joongeon Park; Safdar Jamil; Awais Khan; Sangkeun Lee; Youngjae Kim

doi:10.1109/NVMSA51238.2020.9188162

ScaleML: Machine Learning based Heap Memory Object Scaling Prediction

Joongeon Park, Safdar Jamil, Awais Khan, Sangkeun Lee, Youngjae Kim

Discrete Algorithms

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

Abstract

Memory subsystem contributes 28-40% of total energy consumption. Several studies investigated energy prediction and consumption via profiling memory object access patterns. However, such profiling leads to higher energy consumption due to intense memory object-level profiling to achieve high prediction accuracy. Further, memory object access pattern prediction has been considered through analyzing the variation between memory object access patterns, referred to as scaling rate. The existing techniques for scaling rate prediction, such as Linear Scaling Rate (LSR), suffer from a high error rate in prediction with changes in access patterns, which leads to a high error rate of energy consumption prediction. In this paper, we compare and evaluate several memory object access pattern prediction models including LSR and machine learning (ML) models. Further, we propose ScaleML, a heap memory object scaling rate prediction mechanism that employs an ML model to achieve high access pattern prediction accuracy with variations in memory object access patterns. We evaluate ScaleML using various application benchmarks. The experimental results show that ScaleML achieves about 20% higher accuracy than the LSR model for predicting the scaling rate of object access patterns and energy estimation.

Original language	English
Title of host publication	Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781728184821
DOIs	https://doi.org/10.1109/NVMSA51238.2020.9188162
State	Published - Aug 2020
Event	9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020 - Virtual, Seoul, Korea, Republic of Duration: Aug 19 2020 → Aug 21 2020

Publication series

Name	Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020

Conference

Conference	9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020
Country/Territory	Korea, Republic of
City	Virtual, Seoul
Period	08/19/20 → 08/21/20

Funding

formation Computing Development Progr~ ithrough the National Research Foundation of Korea (Nltl ) funded by the Ministry of Science, ICT (2017M3C4A7080243). Y. Kim is the corresponding author. This manuscript has been authored by UT-Battelle, LLC under Contract No. DEAC05-000R22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a nonexclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloadsldoe-public-access-plan).

Keywords

Machine Learning
Memory Object Access Patterns
Scaling Rate

Access to Document

10.1109/NVMSA51238.2020.9188162

Cite this

Park, J., Jamil, S., Khan, A., Lee, S., & Kim, Y. (2020). ScaleML: Machine Learning based Heap Memory Object Scaling Prediction. In Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020 Article 9188162 (Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/NVMSA51238.2020.9188162

Park, Joongeon ; Jamil, Safdar ; Khan, Awais et al. / ScaleML : Machine Learning based Heap Memory Object Scaling Prediction. Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020. Institute of Electrical and Electronics Engineers Inc., 2020. (Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020).

@inproceedings{1c8e6b9c5fc14413b629c46b69935285,

title = "ScaleML: Machine Learning based Heap Memory Object Scaling Prediction",

abstract = "Memory subsystem contributes 28-40% of total energy consumption. Several studies investigated energy prediction and consumption via profiling memory object access patterns. However, such profiling leads to higher energy consumption due to intense memory object-level profiling to achieve high prediction accuracy. Further, memory object access pattern prediction has been considered through analyzing the variation between memory object access patterns, referred to as scaling rate. The existing techniques for scaling rate prediction, such as Linear Scaling Rate (LSR), suffer from a high error rate in prediction with changes in access patterns, which leads to a high error rate of energy consumption prediction. In this paper, we compare and evaluate several memory object access pattern prediction models including LSR and machine learning (ML) models. Further, we propose ScaleML, a heap memory object scaling rate prediction mechanism that employs an ML model to achieve high access pattern prediction accuracy with variations in memory object access patterns. We evaluate ScaleML using various application benchmarks. The experimental results show that ScaleML achieves about 20% higher accuracy than the LSR model for predicting the scaling rate of object access patterns and energy estimation.",

keywords = "Machine Learning, Memory Object Access Patterns, Scaling Rate",

author = "Joongeon Park and Safdar Jamil and Awais Khan and Sangkeun Lee and Youngjae Kim",

note = "Publisher Copyright: {\textcopyright} 2020 IEEE.; 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020 ; Conference date: 19-08-2020 Through 21-08-2020",

year = "2020",

month = aug,

doi = "10.1109/NVMSA51238.2020.9188162",

language = "English",

series = "Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020",

}

Park, J, Jamil, S, Khan, A, Lee, S & Kim, Y 2020, ScaleML: Machine Learning based Heap Memory Object Scaling Prediction. in Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020., 9188162, Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020, Institute of Electrical and Electronics Engineers Inc., 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020, Virtual, Seoul, Korea, Republic of, 08/19/20. https://doi.org/10.1109/NVMSA51238.2020.9188162

ScaleML: Machine Learning based Heap Memory Object Scaling Prediction. / Park, Joongeon; Jamil, Safdar; Khan, Awais et al.
Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020. Institute of Electrical and Electronics Engineers Inc., 2020. 9188162 (Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020).

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

TY - GEN

T1 - ScaleML

T2 - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020

AU - Park, Joongeon

AU - Jamil, Safdar

AU - Khan, Awais

AU - Lee, Sangkeun

AU - Kim, Youngjae

PY - 2020/8

Y1 - 2020/8

N2 - Memory subsystem contributes 28-40% of total energy consumption. Several studies investigated energy prediction and consumption via profiling memory object access patterns. However, such profiling leads to higher energy consumption due to intense memory object-level profiling to achieve high prediction accuracy. Further, memory object access pattern prediction has been considered through analyzing the variation between memory object access patterns, referred to as scaling rate. The existing techniques for scaling rate prediction, such as Linear Scaling Rate (LSR), suffer from a high error rate in prediction with changes in access patterns, which leads to a high error rate of energy consumption prediction. In this paper, we compare and evaluate several memory object access pattern prediction models including LSR and machine learning (ML) models. Further, we propose ScaleML, a heap memory object scaling rate prediction mechanism that employs an ML model to achieve high access pattern prediction accuracy with variations in memory object access patterns. We evaluate ScaleML using various application benchmarks. The experimental results show that ScaleML achieves about 20% higher accuracy than the LSR model for predicting the scaling rate of object access patterns and energy estimation.

AB - Memory subsystem contributes 28-40% of total energy consumption. Several studies investigated energy prediction and consumption via profiling memory object access patterns. However, such profiling leads to higher energy consumption due to intense memory object-level profiling to achieve high prediction accuracy. Further, memory object access pattern prediction has been considered through analyzing the variation between memory object access patterns, referred to as scaling rate. The existing techniques for scaling rate prediction, such as Linear Scaling Rate (LSR), suffer from a high error rate in prediction with changes in access patterns, which leads to a high error rate of energy consumption prediction. In this paper, we compare and evaluate several memory object access pattern prediction models including LSR and machine learning (ML) models. Further, we propose ScaleML, a heap memory object scaling rate prediction mechanism that employs an ML model to achieve high access pattern prediction accuracy with variations in memory object access patterns. We evaluate ScaleML using various application benchmarks. The experimental results show that ScaleML achieves about 20% higher accuracy than the LSR model for predicting the scaling rate of object access patterns and energy estimation.

KW - Machine Learning

KW - Memory Object Access Patterns

KW - Scaling Rate

UR - http://www.scopus.com/inward/record.url?scp=85091974049&partnerID=8YFLogxK

U2 - 10.1109/NVMSA51238.2020.9188162

DO - 10.1109/NVMSA51238.2020.9188162

M3 - Conference contribution

AN - SCOPUS:85091974049

T3 - Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020

BT - Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 19 August 2020 through 21 August 2020

ER -

Park J, Jamil S, Khan A, Lee S, Kim Y. ScaleML: Machine Learning based Heap Memory Object Scaling Prediction. In Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020. Institute of Electrical and Electronics Engineers Inc. 2020. 9188162. (Proceedings - 9th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2020). doi: 10.1109/NVMSA51238.2020.9188162

ScaleML: Machine Learning based Heap Memory Object Scaling Prediction

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this