Chip multiprocessor performance modeling for contention aware task migration and frequency scaling

Hao Shen, Qinru Qiu

Research output: Contribution to journalArticle

Abstract

Workload consolidation is usually performed in datacenters to improve server utilization for higher energy efficiency. One of the key issues in workload consolidation is the contention for shared resources. Dynamic voltage and frequency scaling (DVFS) of CPU is another effective technique that has been widely used to trade performance for power reduction. We have found that the degree of resource contention of a system affects its performance sensitivity to CPU frequency. Without detailed architecture level information, the complex relationship between contention, frequency and performance cannot be retrieved analytically. In this paper, we apply machine learning techniques to construct a model for chip multiprocessor (CMP) Performance Estimation under Fixed workload Scheduling (PEFS). It quantifies performance degradation of target process caused by resource contention and frequency scaling for current CMP workload with the assumption of a fixed task mapping. The model is further generalized for performance prediction with task migration (PPTM), which predicts the performance degradation after potential intra-processor task migration. Both models are tested on an SMT-enabled chip multi-processor with 10∼20% estimation error on average. Experimental results show that our PEFS model can keep the performance of those bottleneck tasks much closer to the performance threshold than all other techniques, which leads to almost no performance violation while achieves more energy savings, and task migration guided by our PPTM model produces 4%∼9% higher performance than conventional task migration guided by last level cache miss.

Original languageEnglish (US)
Pages (from-to)263-277
Number of pages15
JournalJournal of Low Power Electronics
Volume11
Issue number3
DOIs
StatePublished - Sep 1 2015

Fingerprint

Consolidation
Program processors
Scheduling
Degradation
Surface mount technology
Error analysis
Energy efficiency
Learning systems
Energy conservation
Servers
Dynamic frequency scaling
Voltage scaling

Keywords

  • Consolidation
  • Frequency scaling
  • Migration
  • Power management
  • Resource contention

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Cite this

Chip multiprocessor performance modeling for contention aware task migration and frequency scaling. / Shen, Hao; Qiu, Qinru.

In: Journal of Low Power Electronics, Vol. 11, No. 3, 01.09.2015, p. 263-277.

Research output: Contribution to journalArticle

@article{cf38f1c498b34bb4a0a5dd891f2b67ca,
title = "Chip multiprocessor performance modeling for contention aware task migration and frequency scaling",
abstract = "Workload consolidation is usually performed in datacenters to improve server utilization for higher energy efficiency. One of the key issues in workload consolidation is the contention for shared resources. Dynamic voltage and frequency scaling (DVFS) of CPU is another effective technique that has been widely used to trade performance for power reduction. We have found that the degree of resource contention of a system affects its performance sensitivity to CPU frequency. Without detailed architecture level information, the complex relationship between contention, frequency and performance cannot be retrieved analytically. In this paper, we apply machine learning techniques to construct a model for chip multiprocessor (CMP) Performance Estimation under Fixed workload Scheduling (PEFS). It quantifies performance degradation of target process caused by resource contention and frequency scaling for current CMP workload with the assumption of a fixed task mapping. The model is further generalized for performance prediction with task migration (PPTM), which predicts the performance degradation after potential intra-processor task migration. Both models are tested on an SMT-enabled chip multi-processor with 10∼20{\%} estimation error on average. Experimental results show that our PEFS model can keep the performance of those bottleneck tasks much closer to the performance threshold than all other techniques, which leads to almost no performance violation while achieves more energy savings, and task migration guided by our PPTM model produces 4{\%}∼9{\%} higher performance than conventional task migration guided by last level cache miss.",
keywords = "Consolidation, Frequency scaling, Migration, Power management, Resource contention",
author = "Hao Shen and Qinru Qiu",
year = "2015",
month = "9",
day = "1",
doi = "10.1166/jolpe.2015.1398",
language = "English (US)",
volume = "11",
pages = "263--277",
journal = "Journal of Low Power Electronics",
issn = "1546-1998",
publisher = "American Scientific Publishers",
number = "3",

}

TY - JOUR

T1 - Chip multiprocessor performance modeling for contention aware task migration and frequency scaling

AU - Shen, Hao

AU - Qiu, Qinru

PY - 2015/9/1

Y1 - 2015/9/1

N2 - Workload consolidation is usually performed in datacenters to improve server utilization for higher energy efficiency. One of the key issues in workload consolidation is the contention for shared resources. Dynamic voltage and frequency scaling (DVFS) of CPU is another effective technique that has been widely used to trade performance for power reduction. We have found that the degree of resource contention of a system affects its performance sensitivity to CPU frequency. Without detailed architecture level information, the complex relationship between contention, frequency and performance cannot be retrieved analytically. In this paper, we apply machine learning techniques to construct a model for chip multiprocessor (CMP) Performance Estimation under Fixed workload Scheduling (PEFS). It quantifies performance degradation of target process caused by resource contention and frequency scaling for current CMP workload with the assumption of a fixed task mapping. The model is further generalized for performance prediction with task migration (PPTM), which predicts the performance degradation after potential intra-processor task migration. Both models are tested on an SMT-enabled chip multi-processor with 10∼20% estimation error on average. Experimental results show that our PEFS model can keep the performance of those bottleneck tasks much closer to the performance threshold than all other techniques, which leads to almost no performance violation while achieves more energy savings, and task migration guided by our PPTM model produces 4%∼9% higher performance than conventional task migration guided by last level cache miss.

AB - Workload consolidation is usually performed in datacenters to improve server utilization for higher energy efficiency. One of the key issues in workload consolidation is the contention for shared resources. Dynamic voltage and frequency scaling (DVFS) of CPU is another effective technique that has been widely used to trade performance for power reduction. We have found that the degree of resource contention of a system affects its performance sensitivity to CPU frequency. Without detailed architecture level information, the complex relationship between contention, frequency and performance cannot be retrieved analytically. In this paper, we apply machine learning techniques to construct a model for chip multiprocessor (CMP) Performance Estimation under Fixed workload Scheduling (PEFS). It quantifies performance degradation of target process caused by resource contention and frequency scaling for current CMP workload with the assumption of a fixed task mapping. The model is further generalized for performance prediction with task migration (PPTM), which predicts the performance degradation after potential intra-processor task migration. Both models are tested on an SMT-enabled chip multi-processor with 10∼20% estimation error on average. Experimental results show that our PEFS model can keep the performance of those bottleneck tasks much closer to the performance threshold than all other techniques, which leads to almost no performance violation while achieves more energy savings, and task migration guided by our PPTM model produces 4%∼9% higher performance than conventional task migration guided by last level cache miss.

KW - Consolidation

KW - Frequency scaling

KW - Migration

KW - Power management

KW - Resource contention

UR - http://www.scopus.com/inward/record.url?scp=84957103494&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84957103494&partnerID=8YFLogxK

U2 - 10.1166/jolpe.2015.1398

DO - 10.1166/jolpe.2015.1398

M3 - Article

AN - SCOPUS:84957103494

VL - 11

SP - 263

EP - 277

JO - Journal of Low Power Electronics

JF - Journal of Low Power Electronics

SN - 1546-1998

IS - 3

ER -