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.
- [1] . 1992. An introduction to kernel and nearest-neighbor nonparametric regression. Am. Stat. 46, 3 (1992), 175–185.Google Scholar
- [2] . 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 ScholarDigital Library
- [3] . 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) . 407–412.Google Scholar - [4] . 1984. Classification and Regression Trees. Wadsworth Inc.Google Scholar
- [5] . 2020. Interpretable Machine Learning: A Guide for Making Black Box Models Explainable. Retrieved from https://christophm.github.io/interpretable-ml-book/index.html.Google Scholar
- [6] . 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 Scholar - [7] . 2014. Lightweight graphics instrumentation for game state-specific power management in Android. In Multimedia Systems. Google Scholar
Digital Library
- [8] . 2017. lmt.py. Retrieved from https://gist.github.com/logandillard/lmt.py.Google Scholar
- [9] . 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 ScholarDigital Library
- [10] . 2018. An online learning methodology for performance modeling of graphics processors. IEEE Trans. Comput. 67, 12 (2018), 1677–1691.Google Scholar
Digital Library
- [11] . 2016. Adaptive performance prediction for integrated GPUs. In Proceedings of the International Conference on Computer Aided Design
(ICCAD’16) . Google ScholarDigital Library
- [12] . 2019. A deep Q-learning approach for dynamic management of heterogeneous processors. Comput. Arch. Lett. 18, 1 (2019), 14–17. Google Scholar
Digital Library
- [13] . 1998. Correlation-based Feature Subset Selection for Machine Learning. Ph.D. Dissertation. University of Waikato, Hamilton, New Zealand.Google Scholar
- [14] . 2001. Data Mining Concepts and Techniques. Morgan Kaufmann, San Francisco, CA. Google Scholar
Digital Library
- [15] . 2008. The Elements of Statistical Learning: Data Mining, Inference, and Prediction. Springer.Google Scholar
- [16] . 2018. Memcop: Memory-aware co-operative power management governor for mobile games. Des. Autom. Embed. Syst. 22 (
Sep. 2018), 95–116. Google ScholarDigital Library
- [17] . 2020. Gaming Applications (benchmarks) and Datasets. Retrieved from https://github.com/OD-ML/MLGov_ext.Google Scholar
- [18] . 2015. A control-theoretic approach for energy efficient CPU-GPU subsystem in mobile systems. In Proceedings of the Design Automation Conference
(DAC’15) . Google ScholarDigital Library
- [19] . 2019. Energy-efficient control of mobile processors based on long short-term memory. Access IEEE 7 (2019), 80552–80560.Google Scholar
Cross Ref
- [20] . 2018. The mythos of model interpretability. Queue 16, 3 (2018), 80552–80560. Google Scholar
Digital Library
- [21] . 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), 2842–2854.Google Scholar
Cross Ref
- [22] . 2019. Neural Networks and Deep Learning. Retrieved from https://neuralnetworksanddeeplearning.com/indexs.html.Google Scholar
- [23] . 2019. Quantifying Model Complexity via Functional Decomposition for Better Post-Hoc Interpretability. arXiv: 1904.03867v2.Google Scholar
- [24] . 2019. Interpretable machine learning: definitions,methods, and applications. arXiv.Google Scholar
- [25] . 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) . 12–21. Google ScholarDigital Library
- [26] . 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 ScholarCross Ref
- [27] . 2018. Synergistic CPU-GPU Frequency Capping for Energy-efficient Mobile Games. IEEE Trans. Embed. Comput. Syst. 17, 2 (2018). Google Scholar
Digital Library
- [28] . 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 ScholarDigital Library
- [29] . 2015. Power-performance modelling of mobile gaming workloads on heterogeneous MPSoCs. In Proceedings of the Design Automation Conference
(DAC’15) . Google ScholarDigital Library
- [30] . 2014. Integrated CPU-GPU Power Management for 3D Mobile Games. In Proceedings of the Design Automation Conference
(DAC’14) . Google ScholarDigital Library
- [31] . 2011. Scikit-learn: Machine Learning in Python. J. Mach. Learn. Res. 12 (2011), 2825–2830. Google Scholar
Digital Library
- [32] . 1992. Learning with Continuous Classes. In Proceedings of the Australian Joint Conference on Artificial Intelligence.Google Scholar
- [33] . 2019. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nat. Mach. Intell. 1, 5 (2019), 206–215.Google Scholar
Cross Ref
- [34] . 2003. Artificial Intelligence: A Modern Approach. Prentice Hall. Google Scholar
Digital Library
- [35] . 2019. Assessing the local interpretability of machine learning models. arXiv: 1902.03501v2. Retrieved from https://arxiv.org/abs/1902.03501v2.Google Scholar
- [36] . 1997. Induction of model trees for predicting continuous classes. In Proceedings of the European Conference on Machine Learning.Google Scholar
- [37] . 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), 79–82.Google ScholarCross Ref
- [38] . 2018. Building Model Trees. Retrieved from https://github.com/ankonzoid/LearningX/tree/master/advanced_ML/model_tree.Google Scholar
- [39] . 2015. GPGPU performance and power estimation using machine learning. In Proceedings of the IEEE International Symposium on High-Performance Computer Architecture
(HPCA’15) .Google ScholarCross Ref
Index Terms
An Interpretable Machine Learning Model Enhanced Integrated CPU-GPU DVFS Governor
Recommendations
ML-Gov: a machine learning enhanced integrated CPU-GPU DVFS governor for mobile gaming
ESTIMedia '17: Proceedings of the 15th IEEE/ACM Symposium on Embedded Systems for Real-Time MultimediaModern heterogeneous CPU-GPU based mobile architectures that execute intensive mobile games and other graphics applications use software governors to achieve high performance with energy-efficiency. For dynamic and diverse gaming workloads on ...
Accelerating Open vSwitch with Integrated GPU
KBNets '17: Proceedings of the Workshop on Kernel-Bypass NetworksWith the fast development of Software Defined Networking (SDN) and network virtualization, software-based network virtual switches have emerged as a critical component to provide network services to VMs. Among virtual switches, Open vSwitch (OvS) is an ...
HiCAP: Hierarchical FSM-based Dynamic Integrated CPU-GPU Frequency Capping Governor for Energy-Efficient Mobile Gaming
ISLPED '16: Proceedings of the 2016 International Symposium on Low Power Electronics and DesignContemporary mobile platforms use software governors to achieve high performance with energy-efficiency for heterogeneous CPU-GPU based architectures that execute mobile games and other graphics-intensive applications. Mobile games typically exhibit ...






Comments