A novel cache replacement policy via dynamic adaptive insertion and re-reference prediction

Xi Zhang, Chongmin Li, Zhenyu Liu, Haixia Wang, Dongsheng Wang, Takeshi Ikenaga

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Previous research illustrates that LRU replacement policy is not efficient when applications exhibit a distant re-reference interval. Recently RRIP policy is proposed to improve the performance for such kind of workloads. However, the lack of access recency information in RRIP confuses the replacement policy to make the accurate prediction. To enhance the robustness of RRIP for recency-friendly workloads, we propose an Dynamic Adaptive Insertion and Re-reference Prediction (DAI-RRP) policy which evicts data based on both re-reference prediction value and the access recency information. DAI-RRP makes adaptive adjustment on insertion position and prediction value for different access patterns, which makes the policy robust across different workloads and different phases. Simulation results show that DAI-RRP outperforms LRU and RRIP. For a single-core processor with a 1MB 16-way set last-level cache (LLC), DAIRRP reduces CPI over LRU and Dynamic RRIP by an average of 8.1% and 2.7% respectively. Evaluations on quad-core CMP with a 4MBshared LLC show that DAI-RRP outperforms LRU and Dynamic RRIP (DRRIP) on the weighted speedup metric by an average of 8.1% and 15.7% respectively. Furthermore, compared to LRU, DAI-RRP consumes the similar hardware for 16-way cache, or even less hardware for high-associativity cache. In summary, the proposed policy is practical and can be easily integrated into existing hardware approximations of LRU.

Original languageEnglish
Pages (from-to)468-476
Number of pages9
JournalIEICE Transactions on Electronics
VolumeE94-C
Issue number4
DOIs
Publication statusPublished - 2011 Apr

Fingerprint

Hardware
Cytidine Monophosphate
cyclopropapyrroloindole
Elvitegravir, Cobicistat, Emtricitabine, Tenofovir Disoproxil Fumarate Drug Combination

Keywords

  • Adaptive insertion
  • Cache replacement
  • Set dueling
  • Shared cache

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Electronic, Optical and Magnetic Materials

Cite this

A novel cache replacement policy via dynamic adaptive insertion and re-reference prediction. / Zhang, Xi; Li, Chongmin; Liu, Zhenyu; Wang, Haixia; Wang, Dongsheng; Ikenaga, Takeshi.

In: IEICE Transactions on Electronics, Vol. E94-C, No. 4, 04.2011, p. 468-476.

Research output: Contribution to journalArticle

Zhang, Xi ; Li, Chongmin ; Liu, Zhenyu ; Wang, Haixia ; Wang, Dongsheng ; Ikenaga, Takeshi. / A novel cache replacement policy via dynamic adaptive insertion and re-reference prediction. In: IEICE Transactions on Electronics. 2011 ; Vol. E94-C, No. 4. pp. 468-476.
@article{a10d20ce39b245a2996ae4691bc75c55,
title = "A novel cache replacement policy via dynamic adaptive insertion and re-reference prediction",
abstract = "Previous research illustrates that LRU replacement policy is not efficient when applications exhibit a distant re-reference interval. Recently RRIP policy is proposed to improve the performance for such kind of workloads. However, the lack of access recency information in RRIP confuses the replacement policy to make the accurate prediction. To enhance the robustness of RRIP for recency-friendly workloads, we propose an Dynamic Adaptive Insertion and Re-reference Prediction (DAI-RRP) policy which evicts data based on both re-reference prediction value and the access recency information. DAI-RRP makes adaptive adjustment on insertion position and prediction value for different access patterns, which makes the policy robust across different workloads and different phases. Simulation results show that DAI-RRP outperforms LRU and RRIP. For a single-core processor with a 1MB 16-way set last-level cache (LLC), DAIRRP reduces CPI over LRU and Dynamic RRIP by an average of 8.1{\%} and 2.7{\%} respectively. Evaluations on quad-core CMP with a 4MBshared LLC show that DAI-RRP outperforms LRU and Dynamic RRIP (DRRIP) on the weighted speedup metric by an average of 8.1{\%} and 15.7{\%} respectively. Furthermore, compared to LRU, DAI-RRP consumes the similar hardware for 16-way cache, or even less hardware for high-associativity cache. In summary, the proposed policy is practical and can be easily integrated into existing hardware approximations of LRU.",
keywords = "Adaptive insertion, Cache replacement, Set dueling, Shared cache",
author = "Xi Zhang and Chongmin Li and Zhenyu Liu and Haixia Wang and Dongsheng Wang and Takeshi Ikenaga",
year = "2011",
month = "4",
doi = "10.1587/transele.E94.C.468",
language = "English",
volume = "E94-C",
pages = "468--476",
journal = "IEICE Transactions on Electronics",
issn = "0916-8524",
publisher = "Maruzen Co., Ltd/Maruzen Kabushikikaisha",
number = "4",

}

TY - JOUR

T1 - A novel cache replacement policy via dynamic adaptive insertion and re-reference prediction

AU - Zhang, Xi

AU - Li, Chongmin

AU - Liu, Zhenyu

AU - Wang, Haixia

AU - Wang, Dongsheng

AU - Ikenaga, Takeshi

PY - 2011/4

Y1 - 2011/4

N2 - Previous research illustrates that LRU replacement policy is not efficient when applications exhibit a distant re-reference interval. Recently RRIP policy is proposed to improve the performance for such kind of workloads. However, the lack of access recency information in RRIP confuses the replacement policy to make the accurate prediction. To enhance the robustness of RRIP for recency-friendly workloads, we propose an Dynamic Adaptive Insertion and Re-reference Prediction (DAI-RRP) policy which evicts data based on both re-reference prediction value and the access recency information. DAI-RRP makes adaptive adjustment on insertion position and prediction value for different access patterns, which makes the policy robust across different workloads and different phases. Simulation results show that DAI-RRP outperforms LRU and RRIP. For a single-core processor with a 1MB 16-way set last-level cache (LLC), DAIRRP reduces CPI over LRU and Dynamic RRIP by an average of 8.1% and 2.7% respectively. Evaluations on quad-core CMP with a 4MBshared LLC show that DAI-RRP outperforms LRU and Dynamic RRIP (DRRIP) on the weighted speedup metric by an average of 8.1% and 15.7% respectively. Furthermore, compared to LRU, DAI-RRP consumes the similar hardware for 16-way cache, or even less hardware for high-associativity cache. In summary, the proposed policy is practical and can be easily integrated into existing hardware approximations of LRU.

AB - Previous research illustrates that LRU replacement policy is not efficient when applications exhibit a distant re-reference interval. Recently RRIP policy is proposed to improve the performance for such kind of workloads. However, the lack of access recency information in RRIP confuses the replacement policy to make the accurate prediction. To enhance the robustness of RRIP for recency-friendly workloads, we propose an Dynamic Adaptive Insertion and Re-reference Prediction (DAI-RRP) policy which evicts data based on both re-reference prediction value and the access recency information. DAI-RRP makes adaptive adjustment on insertion position and prediction value for different access patterns, which makes the policy robust across different workloads and different phases. Simulation results show that DAI-RRP outperforms LRU and RRIP. For a single-core processor with a 1MB 16-way set last-level cache (LLC), DAIRRP reduces CPI over LRU and Dynamic RRIP by an average of 8.1% and 2.7% respectively. Evaluations on quad-core CMP with a 4MBshared LLC show that DAI-RRP outperforms LRU and Dynamic RRIP (DRRIP) on the weighted speedup metric by an average of 8.1% and 15.7% respectively. Furthermore, compared to LRU, DAI-RRP consumes the similar hardware for 16-way cache, or even less hardware for high-associativity cache. In summary, the proposed policy is practical and can be easily integrated into existing hardware approximations of LRU.

KW - Adaptive insertion

KW - Cache replacement

KW - Set dueling

KW - Shared cache

UR - http://www.scopus.com/inward/record.url?scp=79953318178&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79953318178&partnerID=8YFLogxK

U2 - 10.1587/transele.E94.C.468

DO - 10.1587/transele.E94.C.468

M3 - Article

AN - SCOPUS:79953318178

VL - E94-C

SP - 468

EP - 476

JO - IEICE Transactions on Electronics

JF - IEICE Transactions on Electronics

SN - 0916-8524

IS - 4

ER -