Dynamic thermal management by greedy scheduling algorithm

Abstract

Chip multiprocessors (CMPs) allow thread level parallelism, thus increasing performance. However, this comes with the cost of temperature problem. CMPs require more power, creating non uniform power map and hotspots. Aiming at this problem, a thread scheduling algorithm, the greedy scheduling algorithm, was proposed to reduce the thermal emergencies and to improve the throughput. The greedy scheduling algorithm was implemented in the Linux kernel on Intel’s Quad-Core system. The experimental results show that the greedy scheduling algorithm can reduce 9.6%–78.5% of the hardware dynamic thermal management (DTM) in various combinations of workloads, and has an average of 5.2% and up to 9.7% throughput higher than the Linux standard scheduler.

Keywords

greedy scheduling algorithm / chip multiprocessor / thermal-aware

References

[1] SkadronK., StanM. R., HuangW., VelusamyS., SankaranarayananK., TarjanD.. Temperature-aware microarchitecture [C]. ISCA’03: The thirtieth International Symposium on Computer Architecture, 2003, New York, ACM: 2-13

[2] BrooksD., MartonosiM.. Dynamic thermal management for high-performance microprocessors [C]. HPCA’01: Proceedings of the Seventh International Symposium on High-Performance Computer Architecture, 2001, Nuevo Leone, Mexico, IEEE Computer Society: 0171

[3] Intel pentium 4 processor in the 478-pin package thermal design guidelines [R]. Intel Document, 2002: 48.

[4] GuntherS., BinnsF., CarmeanD. M., HallJ. C.. Managing the impact of increasing microprocessor power consumption [J]. Intel Technology Journal, 2001, 5(1): 1-9

[5] HeoS., BarrK., AsanovićK.. Reducing power density through activity migration [C]. ISLPED’03: The International Symposium on Low Power Electronics and Design, 2003, New York, ACM: 217-222

[6] SKADRON K. Hybrid architectural dynamic thermal management [R]. Automation and Test in Europe Conference and Exhibition, 2004: 10010.

[7] HansonH., KecklerS. W., GhiasiS., RajamaniK., RawsonF., RubioJ.. Thermal response to DVFS: Analysis with an intel pentium M [C]. ISLPED’07: The International Symposium on Low Power Electronics and Design, 2007, New York, ACM: 219-224

[8] YaoC., SalujaK. K., RamanathanP.. Thermal-aware test scheduling using on-chip temperature sensors [C]. 24th International Conference on VLSI Design, 2011, Madras, Chennai India, IEEE: 376-381

[9] ZhaoJ., DattaB., BurlesonW., TessierR.. Thermal-aware voltage droop compensation for multi-core architectures [C]. Proceedings of the 20th Symposium on Great Lakes on VLSI, 2010, New York, ACM: 335-340

[10] SrinivasanJ., AdveS. V.. Predictive dynamic thermal management for multimedia applications [C]. Proceedings of the 17th annual international conference on Supercomputing, 2003, San Francisco, ACM: 109-120

[11] ChoiJ., CherC. Y., FrankeH., HamannH., WegerA., BoseP.. Thermal-aware task scheduling at the system software level [C]. ISLPED’07. The International Symposium on Low Power Electronics and Design, 2007, New York, ACM: 213-218

[12] DonaldJ., MartonosiM.. Techniques for multicore thermal management: Classification and new exploration [C]. ISCA’06: 33rd International Symposium on Computer Architecture, 2006, Boston, Massachusetts, IEEE: 78-88

[13] KumarA., ShangL., PehL. S., JhaN. K.. HybDTM: A coordinated hardware-software approach for dynamic thermal management [C]. DAC’06: Design Automation Conference, 2006, New York, ACM: 548-553

[14] DengL., DouY.. Multi-core optimization for conjugate gradient benchmark on heterogeneous processors [J]. Journal of Central South University of Technology, 2011, 18(2): 170-179

[15] KURSUN E, CHER C Y, BUYUKTOSUNOGLU A, BOSE P. Investigating the effects of task scheduling on thermal behavior [C]// TACS’ 06: Workshop on TACS. Citeseer, 2006: 63–72.

[16] GomaaM., PowellM. D., VijaykumarT. N.. Heat-and-run: leveraging SMT and CMP to manage power density through the operating system [C]. ASPLOS-XI, 2004, New York, ACM: 260-270

[17] YangJ., ZhouX., ChrobakM., ZhangY., JinL.. Dynamic thermal management through task scheduling [C]. ISPASS’08: IEEE International Symposium on Performance Analysis of Systems and Software, 2008, Washington, IEEE: 191-201

[18] ABBASI Z, VARSAMOPOULOS G, GUPTA S K S. Thermal aware server provisioning and workload distribution for internet data centers [C]// Proceedings of the 19th ACM International Symposium on High Performance Distributed Computing. Chicago, Illinois, 2010: 130–141.

[19] ZhouX., YangJ., XuY., ZhangY., ZhaoJ.. Thermal-aware task scheduling for 3D multicore processors [J]. IEEE Transactions on Parallel and Distributed Systems, 2010, 21(1): 60-71

[20] ChaoC. H., JhengK. Y., WangH. Y., WuJ. C., WuA. Y.. Traffic-and thermal-aware run-time thermal management scheme for 3D NoC systems [C]. Preceedings of the Fourth ACM/IEEE International Symposium on Networks-on-Chip, 2010, Grenoble, France, IEEE: 223-230

[21] CUI J, MASKELL D L. High level event driven thermal estimation for thermal aware task allocation and scheduling [C]// Proceedings of the 2010 Asia and South Pacific Design Automation Conference. Taipei, Taiwan, 2010: 793–798.

[22] FisherN., ChenJ. J., WangS., ThieleL.. Thermal-aware global real-time scheduling and analysis on multicore systems [J]. Journal of Systems Architecture, 2010, 57(5): 547-560

[23] WangS., BettatiR.. Reactive speed control in temperature-constrained real-time systems [J]. Real-Time Systems, 2008, 39(1): 73-95

[24] MulasF., PittauM., ButtuM., CartaS., AcquavivaA., BeniniL., AtienzaD., MicheliG. D.. Thermal balancing policy for streaming computing on multiprocessor architectures [C]. DATE’08: Design, Automation and Test in Europe, 2008, Munich, Germany, IEEE: 734-739

[25] YEO I, KIM E J. Temperature-aware scheduler based on thermal behavior grouping in multicore systems [C]// DATE’09: Proceedings of the Conference on Design, Automation and Test in Europe. Leuven, Belgium: European Design and Automation Association, 2009: 946–951.

[26] CHROBAK M, D RR C, HURAND M, ROBERT J. Algorithms for temperature-aware task scheduling in microprocessor systems, [J]. Algorithmic Aspects in Information and Management, 2008(5034): 120–130.

[27] BovetD., CesatiM.Understanding the Linux kernel [M], 2002, Sebastopol, O’Reilly & Associates, Inc: 136-150

[28] LiL., ZhouX., YangJ.. ThresHot: An aggressive task scheduling approach in CMP thermal design [J]. Algorithmic Aspects in Information and Management, 2008, 5034: 92-99

[29] YeoI., LiuC. C., KimE. J.Predictive dynamic thermal management for multicore systems [C], 2008, Anaheim, ACM: 734-739