Multi-core Chip Dynamic Power Management Framework Based on Reinforcement Learning br

被引：0

作者：

Zhuo, Cheng ^{[2
,5
]}

Zeng, Xudong ^{[1
,5
]}

Chen, Yufei ^{[2
]}

Sun, Songyu ^{[2
]}

Luo, Guojie ^{[3
]}

He, Qing ^{[4
]}

Yin, Xunzhao ^{[2
]}

机构：

[1] Zhejiang Univ, Polytech Inst, Hangzhou 310015, Peoples R China

[2] Zhejiang Univ, Coll Informat Sci & Elect Engn, Hangzhou 310027, Peoples R China

[3] Peking Univ, Sch Elect Engn & Comp Sci, Beijing 100871, Peoples R China

[4] Hangzhou Xingxin Technol Co Ltd, Hangzhou 310052, Peoples R China

[5] Key Lab Collaborat Sensing & Autonomous Unmanned S, Hangzhou 310015, Peoples R China

来源：

JOURNAL OF ELECTRONICS & INFORMATION TECHNOLOGY | 2023年 / 45卷 / 01期

基金：

中国国家自然科学基金;

关键词：

Multi-core chip; Dynamic Power Management(DPM); Reinforcement Learning(RL); THERMAL MANAGEMENT; MOBILE;

D O I：

10.11999/JEIT220350

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

Multi-core chips can provide mighty computing capability for mobile intelligent terminals, but their performance is constraint by thermal and power issues. For this problem, this paper proposes a multi-core chipdynamic power management framework based on reinforcement learning. First, based on GEM5, a dynamicvoltage and frequency scaling simulation system of the multi-core chips is established. Second, a chip powermodel characterization method is adopted, which takes CMOS physical characteristics into consideration torealize online real-time power monitoring. Finally, a gradient reward method for the multi-core chips isdesigned, and a Deep Q Network (DQN) algorithm is used to learn the power management strategy for themulti-core chips. Compared with conventional Ondemand and MaxBIPS schemes, the simulation results showthat the proposed framework achieves 2.12% and 4.03% improvement in computational performance of the multi-core chips respectively

引用

页码：24 / 32

页数：9

共 29 条

[1] Bertran Ramon, 2010, 24th ACM International Conference on Supercomputing 2010, P147
[2] Algorithmic Optimization of Thermal and Power Management for Heterogeneous Mobile Platforms
Bhat, Ganapati
Singla, Gaurav
Unver, Ali K.
Ogras, Umit Y.
[J]. IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS, 2018, 26 (03) : 544 - 557
[3] Power-Temperature Stability and Safety Analysis for Multiprocessor Systems
Bhat, Ganapati
Gumussoy, Suat
Ogras, Umit Y.
[J]. ACM TRANSACTIONS ON EMBEDDED COMPUTING SYSTEMS, 2017, 16
[4] Bienia Christian, 2011, Benchmarking Modern Multiprocessors
[5] Binkert Nathan, 2011, Computer Architecture News, V39, P1, DOI 10.1145/2024716.2024718
[6] Cai EM, 2015, ICCAD-IEEE ACM INT, P500, DOI 10.1109/ICCAD.2015.7372611
[7] An Efficient and Flexible Learning Framework for Dynamic Power and Thermal Co-Management
Cao, Yuan
Shen, Tianhao
Zhang, Li
Yin, Xunzhao
Zhuo, Cheng
[J]. PROCEEDINGS OF THE 2020 ACM/IEEE 2ND WORKSHOP ON MACHINE LEARNING FOR CAD (MLCAD '20), 2020, : 117 - 122
[8] PAM: A Piecewise-Linearly-Approximated Floating-Point Multiplier With Unbiasedness and Configurability
Chen, Chuangtao
Qian, Weikang
Imani, Mohsen
Yin, Xunzhao
Zhuo, Cheng
[J]. IEEE TRANSACTIONS ON COMPUTERS, 2022, 71 (10) : 2473 - 2486
[9] Ge Y, 2011, DES AUT CON, P95
[10] A Dynamic Programming Framework for DVFS-Based Energy-Efficiency in Multicore Systems
Hajiamini, Shervin
Shirazi, Behrooz
Crandall, Aaron
Ghasemzadeh, Hassan
[J]. IEEE TRANSACTIONS ON SUSTAINABLE COMPUTING, 2020, 5 (01): : 1 - 12

← 1 2 3 →