Using Software-Defined Performance Counters to Construct a GPU Power Consumption Model
https://doi.org/10.15514/ISPRAS-2025-37(4)-16
Abstract
In the modern world, processor performance and energy efficiency play a key role in computer system design. Along with CPUs, GPUs are powerful computing devices used for computer graphics processing, machine learning, and more. Processors are equipped with built-in sensors accessible through specialized tools. The chip of a modern video card can operate in a fairly wide range of frequencies and power limits (PLs). Very often, when solving a computational task or rendering a scene, the video card can operate more optimally, without wasting excess power, which can significantly save energy on labor-intensive tasks. Therefore, it is important for a set of given tasks to find such parameters where the ratio of useful work per watt will be maximum. After conducting a large number of experiments, one can learn to predict the dependence of such a target function on the parameters. This paper examines obtaining current GPU parameter values using various tools. We present results of collecting raw data from NVIDIA GPUs and the subsequent construction of an optimal power consumption model.
About the Authors
Alexey Nikolaevich GULIN
Polzunov Altai State Technical University
Russian Federation
Russian Federation
Master student at the department of Applied Mathematics, Software Engineering direction, AltSTU. Research interests: energy-efficient computing, computational system optimization.
Sergey Mikhailovich STAROLETOV
Polzunov Altai State Technical University
Russian Federation
Russian Federation
Cand. Sci. (Phys.-Math.), associate professor. Research interests: formal verification, model checking, cyber-physical systems, operating systems.
References
1. Falk H. Personal computer graphic cards. The Electronic Library, vol. 12, no. 2, pp. 127–130, 1994.
2. Fatahalian K., Houston M. A closer look at GPUs. Communications of the ACM, vol. 51, no. 10, pp. 50-57, 2008.
3. Garland M, Kirk D. B. Understanding throughput-oriented architectures. Communications of the ACM, vol. 53, no. 11, pp. 58-66, 2010.
4. Gorlatch S., Garanina N, Staroletov S. Using the SPIN model checker for auto-tuning high-performance programs. Journal of Mathematical Sciences, pp. 1-13, 2025.
5. Кондратьев Д.А., Старолетов С.М., Шошмина И.В., Красненкова А.В., Зиборов К.В., Шилов Н.В., Гаранина Н.О., Черганов Т.Ю. Соревнования по формальной верификации VeHa-2024: накопленный в течение двух лет опыт и перспективы. Труды ИСП РАН, том 37, вып. 1, 2025 г., стр. 159-184. DOI: 10.15514/ISPRAS–2025–37(1)–10. / Kondratyev D.A., Staroletov S.M., Shoshmina I.V., Krasnenkova A.V., Ziborov K.V., Shilov N.V., Garanina N.O., Cherganov T.Y. VeHa-2024 Formal Verification Contest: Two Years of Experience and Prospects (in Russian). Trudy ISP RAN/Proc. ISP RAS, vol 37, issue. 1, pp. 159-184, 2025.
6. Muralidhar R., Borovica-Gajic R, Buyya R. Energy efficient computing systems: Architectures, abstractions and modeling to techniques and standards. ACM Computing Surveys (CSUR), vol. 54, no. 11s, pp. 1–37, 2022.
7. Pei Q., Yuan Y., Hu H., Wang L., Zhang D., Yan B., Yu C., Liu F. Working Smarter Not Harder: Hybrid Cooling for Deep Learning in Edge Datacenters. IEEE Transactions on Sustainable Computing, 2025.
8. Murdock K., Oswald D., Garcia F. D., Van Bulck J., Piessens F., Gruss D. Plundervolt: How a little bit of undervolting can create a lot of trouble. IEEE Security & Privacy, vol. 18, no. 5, pp. 28–37, 2020.
9. Colmant M., Rouvoy R., Kurpicz M., Sobe A., Felber P., Seinturier L. The next 700 CPU power models. Journal of Systems and Software, vol. 144, pp. 382–396, 2018.
10. Fieni G., Acero D.R., Rust P., Rouvoy R. PowerAPI: A Python framework for building software-defined power meters. Journal of Open Source Software, vol. 9, no. 98, p. 6670, 2024.
11. Savikov D., Shintyapin I., Staroletov S. Concept of building power models for central processors using the PowerAPI toolkit (in Russian). In Modern digital technologies: proceedings of the III All-Russian scientific-practical conference (03 June 2024). Barnaul: AltSTU, pp. 211–215, 2024. Available: https://elibrary.ru/item.asp?id=68573384.
12. Gulin A., Filimontsev I., Staroletov S. Analysis of data from central processors for building their power models using the PowerAPI toolkit (in Russian). In Modern digital technologies: proceedings of the III All-Russian scientific-practical conference (03 June 2024). Barnaul: AltSTU, pp. 138–142, 2024. Available: https://elibrary.ru/item.asp?id=68573361
13. Group S.R. et al. PyJoules: a Python library to capture the energy consumption of code snippets. University of Lille and Inria, 2021.
14. Zamani H., Tripathy D., Bhuyan L., Chen Z. SAOU: Safe adaptive overclocking and undervolting for energy-efficient GPU computing. In Proceedings of the ACM/IEEE International Symposium on Low Power Electronics and Design, pp. 205–210, 2020.
15. Zamani H., Liu Y., Tripathy D., Bhuyan L, Chen Z. GreenMM: energy efficient GPU matrix multiplication through undervolting. In Proceedings of the ACM International Conference on Supercomputing, pp. 308–318, 2019.
16. Dong Z., Wang Z., Yuan J. Research and application of GPU pass-through based on KVM in graphics workstation. In 2018 3rd International Workshop on Materials Engineering and Computer Sciences (IWMECS 2018). Atlantis Press, pp. 354–357, 2018.
17. Czarnul P., Proficz J., Krzywaniak A. Energy-aware high-performance computing: survey of state-of-the-art tools, techniques, and environments. Scientific Programming, vol. 2019, no. 1, p. 8348791, 2019.
18. Torres L.A., Barrios C.J., Denneulin Y. Computational resource consumption in convolutional neural network training–a focus on memory. Supercomputing Frontiers and Innovations, vol. 8, no. 1, pp. 45–61, 2021.
19. pynvraw, 2021. Available: https://github.com/JustAMan/pynvraw, accessed Jul. 15, 2025.
20. Shaqiri M., Iljazi T., Kamberi L., Ramani-Halili R. Differences Between The Correlation Coefficients Pearson, Kendall And Spearman. Journal of Natural Sciences and Mathematics of UT, vol. 8, no. 15-16, pp. 392–397, 2023.
21. De Winter J.C., Gosling S.D., Potter J. Comparing the Pearson and Spearman correlation coefficients across distributions and sample sizes: A tutorial using simulations and empirical data. Psychological methods, vol. 21, no. 3, p. 273, 2016.
22. Ke G., Meng Q., Finley T., Wang T., Chen W., Ma W., Ye Q., Liu T.-Y. Lightgbm: A highly efficient gradient boosting decision tree. Advances in neural information processing systems, vol. 30, 2017.
23. Frazier P.I. Bayesian optimization. In Recent advances in optimization and modeling of contemporary problems. Informs, pp. 255–278, 2018.
24. GPU Power Model, 2025. Available: https://github.com/GulinAlexey/GPU_Power_Model, accessed Jul. 15, 2025.
Review
For citations:
GULIN A.N., STAROLETOV S.M. Using Software-Defined Performance Counters to Construct a GPU Power Consumption Model. Proceedings of the Institute for System Programming of the RAS (Proceedings of ISP RAS). 2025;37(4):17-30. https://doi.org/10.15514/ISPRAS-2025-37(4)-16
Email this article 