skip to main content
research-article

An Interpretable Machine Learning Model Enhanced Integrated CPU-GPU DVFS Governor

Published:18 October 2021Publication History
Skip Abstract Section

Abstract

Modern heterogeneous CPU-GPU-based mobile architectures, which execute intensive mobile gaming/graphics applications, use software governors to achieve high performance with energy-efficiency. However, existing governors typically utilize simple statistical or heuristic models, assuming linear relationships using a small unbalanced dataset of mobile games; and the limitations result in high prediction errors for dynamic and diverse gaming workloads on heterogeneous platforms. To overcome these limitations, we propose an interpretable machine learning (ML) model enhanced integrated CPU-GPU governor: (1) It builds tree-based piecewise linear models (i.e., model trees) offline considering both high accuracy (low error) and interpretable ML models based on mathematical formulas using a simulatability operation counts quantitative metric. And then (2) it deploys the selected models for online estimation into an integrated CPU-GPU Dynamic Voltage Frequency Scaling governor. Our experiments on a test set of 20 mobile games exhibiting diverse characteristics show that our governor achieved significant energy efficiency gains of over 10% (up to 38%) improvements on average in energy-per-frame with a surprising-but-modest 3% improvement in Frames-per-Second performance, compared to a typical state-of-the-art governor that employs simple linear regression models.

REFERENCES

  1. [1] Altman Naomi S.. 1992. An introduction to kernel and nearest-neighbor nonparametric regression. Am. Stat. 46, 3 (1992), 175185.Google ScholarGoogle Scholar
  2. [2] Bai Yu and Vaidya Priya. 2009. Memory characterization to analyze and predict multimedia perormance and power in embedded systems. In Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP’09). Google ScholarGoogle ScholarDigital LibraryDigital Library
  3. [3] Bhattacharya B. and Solomatine D. P.. 2003. Neural networks and M5 model treesin modeling water level-discharge relationshipfor an Indian river. In Proceedings of the European Symposium on Artificial Neural Networks (ESANN’03). 407412.Google ScholarGoogle Scholar
  4. [4] Breiman Leo, Friedman Jerome, Stone Charles J., and Olshen R. A.. 1984. Classification and Regression Trees. Wadsworth Inc.Google ScholarGoogle Scholar
  5. [5] Molnar Christoph. 2020. Interpretable Machine Learning: A Guide for Making Black Box Models Explainable. Retrieved from https://christophm.github.io/interpretable-ml-book/index.html.Google ScholarGoogle Scholar
  6. [6] Chuan Po-Kai, Chen Ya-Shu, and Huang Po-Hao. 2017. An adaptive on-line cpu-gpu governor for games on mobile devices. In Proceedings of the Asia and South Pacific Design Automation Conference (ASP-DAC’17).Google ScholarGoogle Scholar
  7. [7] Dietrich Benedikt and Chakraborty Samarjit. 2014. Lightweight graphics instrumentation for game state-specific power management in Android. In Multimedia Systems. Google ScholarGoogle ScholarDigital LibraryDigital Library
  8. [8] Dillard Logan. 2017. lmt.py. Retrieved from https://gist.github.com/logandillard/lmt.py.Google ScholarGoogle Scholar
  9. [9] Dwyer Tyler, Fedorova Alexandra, Blagodurov Sergey, Roth Mark, Gaud Fabien, and Pei Jian. 2012. A practical method for estimating performance degradation on multicore processors, and its application to HPC workloads. In Proceedings of the ACM/IEEE Supercomputing Conference (SC’12). Google ScholarGoogle ScholarDigital LibraryDigital Library
  10. [10] Gupta Ujjwal, Babu Manoj, Ayoub Raid, Kishinevsky Michael, Paterna Francesco, Gumussoy Suat, and Ogras Ümit Y.. 2018. An online learning methodology for performance modeling of graphics processors. IEEE Trans. Comput. 67, 12 (2018), 16771691.Google ScholarGoogle ScholarDigital LibraryDigital Library
  11. [11] Gupta Ujjwal, Campbell Joseph, Ayoub Raid, Kishinevsky Michael, and Gumussoy Suat. 2016. Adaptive performance prediction for integrated GPUs. In Proceedings of the International Conference on Computer Aided Design (ICCAD’16). Google ScholarGoogle ScholarDigital LibraryDigital Library
  12. [12] Gupta Ujjwal, Mandal Sumit K., Ma Manqing, Chakrabarti Chaitali, and Ogras Ümit Y.. 2019. A deep Q-learning approach for dynamic management of heterogeneous processors. Comput. Arch. Lett. 18, 1 (2019), 1417. Google ScholarGoogle ScholarDigital LibraryDigital Library
  13. [13] Hall M. A.. 1998. Correlation-based Feature Subset Selection for Machine Learning. Ph.D. Dissertation. University of Waikato, Hamilton, New Zealand.Google ScholarGoogle Scholar
  14. [14] Han Jiawei and Kamber Micheline. 2001. Data Mining Concepts and Techniques. Morgan Kaufmann, San Francisco, CA. Google ScholarGoogle ScholarDigital LibraryDigital Library
  15. [15] Hastie Trevor, Tibshirani Robert, and Friedman Jerome. 2008. The Elements of Statistical Learning: Data Mining, Inference, and Prediction. Springer.Google ScholarGoogle Scholar
  16. [16] Hsieh Chen-Ying, Park Jurn-Gyu, Dutt Nikil, and Lim Sung-Soo. 2018. Memcop: Memory-aware co-operative power management governor for mobile games. Des. Autom. Embed. Syst. 22 (Sep. 2018), 95–116. Google ScholarGoogle ScholarDigital LibraryDigital Library
  17. [17] Park Jurn-Gyu. 2020. Gaming Applications (benchmarks) and Datasets. Retrieved from https://github.com/OD-ML/MLGov_ext.Google ScholarGoogle Scholar
  18. [18] Kadjo David, Ayoub Raid, Kishinevsky Michael, and Gratz Paul V.. 2015. A control-theoretic approach for energy efficient CPU-GPU subsystem in mobile systems. In Proceedings of the Design Automation Conference (DAC’15). Google ScholarGoogle ScholarDigital LibraryDigital Library
  19. [19] Lee Jaehwan, Nam Sanghyuck, and Park Sangoh. 2019. Energy-efficient control of mobile processors based on long short-term memory. Access IEEE 7 (2019), 8055280560.Google ScholarGoogle ScholarCross RefCross Ref
  20. [20] Lipton Zachary C.. 2018. The mythos of model interpretability. Queue 16, 3 (2018), 8055280560. Google ScholarGoogle ScholarDigital LibraryDigital Library
  21. [21] Mandal Sumit K., Bhat Ganapati, Patil Chetan Arvind, Doppa Janardhan Rao, Pande Partha Pratim, and Ogras Umit Y.. 2019. Sumit K. Mandal, Ganapati Bhat, Chetan Arvind Patil, Janardhan Rao Doppa, Partha Pratim Pande, Umit Y. Ogras. IEEE Trans. VLSI Syst. 27, 12 (2019), 28422854.Google ScholarGoogle ScholarCross RefCross Ref
  22. [22] Nielsen Michael. 2019. Neural Networks and Deep Learning. Retrieved from https://neuralnetworksanddeeplearning.com/indexs.html.Google ScholarGoogle Scholar
  23. [23] Molnar Christoph, Casalicchio Giuseppe, and Bischl Bernd. 2019. Quantifying Model Complexity via Functional Decomposition for Better Post-Hoc Interpretability. arXiv: 1904.03867v2.Google ScholarGoogle Scholar
  24. [24] Murdocha W. James, Singh Chandan, Kumbiera Karl, Abbasi-Aslb Reza, and Yu Bin. 2019. Interpretable machine learning: definitions,methods, and applications. arXiv.Google ScholarGoogle Scholar
  25. [25] Park Jurn-Gyu, Dutt Nikil, and Lim Sung-Soo. 2017. ML-Gov: A machine learning enhanced integrated CPU-GPU DVFS governor for mobile gaming. In Proceedings of the Symposium on Embedded Systems for Real-Time Multimedia (ESTIMedia’17). 1221. Google ScholarGoogle ScholarDigital LibraryDigital Library
  26. [26] Park Jurn-Gyu, Hsieh Chen-Ying, Dutt Nikil, and Lim Sung-Soo. 2014. Quality-aware mobile graphics workload characterization for energy-efficient DVFS design. In Proceedings of the Symposium on Embedded Systems for Real-Time Multimedia (ESTIMedia’14).Google ScholarGoogle ScholarCross RefCross Ref
  27. [27] Park Jurn-Gyu, Hsieh Chen-Ying, Dutt Nikil, and Lim Sung-Soo. 2018. Synergistic CPU-GPU Frequency Capping for Energy-efficient Mobile Games. IEEE Trans. Embed. Comput. Syst. 17, 2 (2018). Google ScholarGoogle ScholarDigital LibraryDigital Library
  28. [28] Park Jurn-Gyu, Kim Hoyeonjiki, Dutt Nikil, and Lim Sung-Soo. 2016. HiCAP: Hierarchical FSM-based dynamic integrated CPU-GPU frequency capping governor for energy-efficient mobile gaming. In Proceedings of the International Symposium on Low Power Electronics and Design (ISLPED’16). Google ScholarGoogle ScholarDigital LibraryDigital Library
  29. [29] Pathania Anuj, Irimiea Alexandru Eugen, Prakash Alok, and Mitra Tulika. 2015. Power-performance modelling of mobile gaming workloads on heterogeneous MPSoCs. In Proceedings of the Design Automation Conference (DAC’15). Google ScholarGoogle ScholarDigital LibraryDigital Library
  30. [30] Pathania Anuj, Jiao Qing, Prakash Alok, and Mitra Tulika. 2014. Integrated CPU-GPU Power Management for 3D Mobile Games. In Proceedings of the Design Automation Conference (DAC’14). Google ScholarGoogle ScholarDigital LibraryDigital Library
  31. [31] Pedregosa F., Varoquaux G., Gramfort A., Michel V., Thirion B., Grisel O., Blondel M., Prettenhofer P., Weiss R., Dubourg V., Vanderplas J., Passos A., Cournapeau D., Brucher M., Perrot M., and Duchesnay E.. 2011. Scikit-learn: Machine Learning in Python. J. Mach. Learn. Res. 12 (2011), 28252830. Google ScholarGoogle ScholarDigital LibraryDigital Library
  32. [32] Quinlan Ross J.. 1992. Learning with Continuous Classes. In Proceedings of the Australian Joint Conference on Artificial Intelligence.Google ScholarGoogle Scholar
  33. [33] Rudin Cynthia. 2019. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nat. Mach. Intell. 1, 5 (2019), 206215.Google ScholarGoogle ScholarCross RefCross Ref
  34. [34] Russell Stuart and Norvig Peter. 2003. Artificial Intelligence: A Modern Approach. Prentice Hall. Google ScholarGoogle ScholarDigital LibraryDigital Library
  35. [35] Slack Dylan, Friedler Sorelle A., Scheidegger Carlos, and Roy Chitradeep Dutta. 2019. Assessing the local interpretability of machine learning models. arXiv: 1902.03501v2. Retrieved from https://arxiv.org/abs/1902.03501v2.Google ScholarGoogle Scholar
  36. [36] Wang Yong and Witten Ian H.. 1997. Induction of model trees for predicting continuous classes. In Proceedings of the European Conference on Machine Learning.Google ScholarGoogle Scholar
  37. [37] Willmott C. and Matsuura K.. 2005. Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance. Clim. Res. 30 (Dec. 2005), 7982.Google ScholarGoogle ScholarCross RefCross Ref
  38. [38] Wong Anson. 2018. Building Model Trees. Retrieved from https://github.com/ankonzoid/LearningX/tree/master/advanced_ML/model_tree.Google ScholarGoogle Scholar
  39. [39] Wu Gene, Greathouse Joseph L., Lyashevsky Alexander, Jayasena Nuwan, and Chiou Derek. 2015. GPGPU performance and power estimation using machine learning. In Proceedings of the IEEE International Symposium on High-Performance Computer Architecture (HPCA’15).Google ScholarGoogle ScholarCross RefCross Ref

Index Terms

  1. An Interpretable Machine Learning Model Enhanced Integrated CPU-GPU DVFS Governor

        Recommendations

        Comments

        Login options

        Check if you have access through your login credentials or your institution to get full access on this article.

        Sign in

        Full Access

        PDF Format

        View or Download as a PDF file.

        PDF

        eReader

        View online with eReader.

        eReader

        Full Text

        View this article in Full Text.

        View Full Text

        HTML Format

        View this article in HTML Format .

        View HTML Format
        About Cookies On This Site

        We use cookies to ensure that we give you the best experience on our website.

        Learn more

        Got it!