DOI QR코드

DOI QR Code

Power Modeling Approach for GPU Source Program

  • Li, Junke (College of Computer Science, Sichuan University, China/School of Computer and Information, Qiannan Normal University for Nationalities) ;
  • Guo, Bing (College of Computer Science, Sichuan University) ;
  • Shen, Yan (School of Control Engineering, Chengdu University of Information Technology) ;
  • Li, Deguang (College of Computer Science, Sichuan University) ;
  • Huang, Yanhui (College of Computer Science, Sichuan University)
  • 투고 : 2017.03.05
  • 심사 : 2017.08.23
  • 발행 : 2018.01.01

초록

Rapid development of information technology makes our environment become smarter and massive high performance computers are providing powerful computing for that. Graphics Processing Unit (GPU) as a typical high performance component is being widely used for both graphics and general-purpose applications. Although it can greatly improve computing power, it also delivers significant power consumption and need sufficient power supplies. To make high performance computing more sustainable, the important step is to measure it. Current power technologies for GPU have some drawbacks, such as they are not applicable for power estimation at the early stage. In this article, we present a novel power technology to correlate power consumption and the characteristics at the programmer perspective, and then to estimate power consumption of source program without prerunning. We conduct experiments on Nvidia's GT740 platform; the results show that our power model is more accurately than regression model and has an average error of 2.34% and the maximum error of 9.65%.

키워드

E1EEFQ_2018_v13n1_181_f0001.png 이미지

Fig. 1. Different Approaches of energy measurement[6]

E1EEFQ_2018_v13n1_181_f0002.png 이미지

Fig. 2. Structure of BP neural network

E1EEFQ_2018_v13n1_181_f0003.png 이미지

Fig. 3. Comparison between estimated and measured power

E1EEFQ_2018_v13n1_181_f0004.png 이미지

Fig. 4. Relationship between Rreg and the power

E1EEFQ_2018_v13n1_181_f0005.png 이미지

Fig. 5. Relationship between Os and the power

E1EEFQ_2018_v13n1_181_f0006.png 이미지

Fig. 6. Relationship between Rsmem and the power

E1EEFQ_2018_v13n1_181_f0007.png 이미지

Fig. 7. Relationship between Rgmem and the power

E1EEFQ_2018_v13n1_181_f0008.png 이미지

Fig. 8. Relationship between CTMR and the power

E1EEFQ_2018_v13n1_181_f0009.png 이미지

Fig. 9. Power comparisons under different divergence andinput size

E1EEFQ_2018_v13n1_181_f0010.png 이미지

Fig. 10. Different input affect characteristics

E1EEFQ_2018_v13n1_181_f0011.png 이미지

Fig. 11. Power error under different approaches

Table 1. Resource specified by programmer

E1EEFQ_2018_v13n1_181_t0001.png 이미지

Table 2. Hardware resource limitation

E1EEFQ_2018_v13n1_181_t0002.png 이미지

참고문헌

  1. Top 500 Supercomputer Sites Webpage, November 2015. http://www.top500.org.
  2. Chen, J., Li, B., Zhang, Y., "Tree structured analysis on GPU power study," in Proceedings of IEEE International Conference on Computer Design, pp. 57-64, 2011.
  3. Ma, X., et al, "Statistical power consumption analysis and modeling for GPU-based computing," in Proceedings of ACM SOSP Workshop on Power Aware Computing and Systems, October 2009.
  4. Nagasaka, H., Maruyama, N., et al, "Statistical power modeling of GPU kernels using performance counters," in Proceedings of IEEE International Conference on Green Computing, pp. 115-122, 2010.
  5. Zhang, Y., Hu, Y., et al, "Performance and power analysis of ATI gpu: A statistical approach," in Proceedings of IEEE International Conference on NAS, pp. 115-122, 2011.
  6. Song, S., Su, C., et al, "A simplified and accurate model of power-performance efficiency on emergent gpu architectures," in Proceedings of IEEE International Symposium on IPDPS, pp. 676-686, 2013.
  7. Chen, H., Li, Y., & Shi, W., "Fine-grained power management using process-level profiling," Sustainable Computing: Informatics and Systems, vol. 2, no. 1, pp. 33-42, 2012. https://doi.org/10.1016/j.suscom.2012.01.002
  8. Li, K., Yang, W., Li, K, "Performance analysis and optimization for SpMV on GPU using probabilistic modeling," IEEE Trans. on Parallel and Distributed Systems, vol. 26, no. 1, pp. 196-205, 2015. https://doi.org/10.1109/TPDS.2014.2308221
  9. Kreutzer, M., Pieper, A., et al, "Performance Engineering of the Kernel Polynomal Method on Large-Scale CPU-GPU Systems," in Proceedings of IEEE International Conference on IPDPS, pp. 417- 426, 2015.
  10. Chitty, D. M., "Improving the performance of GPUbased genetic programming through exploitation of on-chip memory," Soft Computing, vol. 20, no. 2, pp. 661-680, 2016. https://doi.org/10.1007/s00500-014-1530-3
  11. Dastgeer, U., Kessler, C., "Performance-aware composition framework for GPU-based systems," The Journal of Supercomputing, vol. 71, no. 12, pp. 4646- 4662, 2015. https://doi.org/10.1007/s11227-014-1105-1
  12. Angerer, C. M., et al, "A fast, hybrid, power-efficient high-precision solver for large linear systems based on low-precision hardware," Sustainable Computing: Informatics and Systems, vol. 12, pp. 72-82, 2015.
  13. Luo, C., Suda, R., "A performance and energy consumption analytical model for GPU," in Proceedings of IEEE International Conference on DASC, pp. 658-665, 2011.
  14. Ramani, K., Ibrahim, A., Shimizu, D., "PowerRed: A flexible modeling framework for power efficiency exploration in GPUs," in Proceedings of the Workshop on General Purpose Processing on GPUs, October 2007.
  15. Lim, J., et al, "Power modeling for GPU architectures using McPAT," ACM Trans. on Design Automation of Electronic Systems, vol. 19, no. 3, pp. 26, 2014. https://doi.org/10.1145/2611758
  16. Leng, J., Hetherington, T., et al, "GPU Wattch: enabling energy optimizations in GPGPUs," ACM SIGARCH Computer Architecture News, vol. 41, no. 3, ACM, 2013.
  17. N. Goswami, Cao, B., Li, T., "Power-performance co-optimization of throughput core architecture using resistive memory," in Proceedings of IEEE International Conference on HPCA, pp. 342-353, 2013.
  18. N. Goswami, A. Verma, and T. Li, Gpu-powersim, 2012.http://www.ideal.ece.ufl.edu/main.php?action=gpupowersim.
  19. Tiwari, V., Malik, S., Wolfe, A., "Power analysis of embedded software: a first step towards software power minimization," IEEE Trans. on Very Large Scale Integration Systems, vol. 2, no. 4, pp. 437-445, 1994. https://doi.org/10.1109/92.335012
  20. Hong, S., Kim, H., "An integrated GPU power and performance model," ACM SIGARCH Computer Architecture News, vol. 38, no. 3, ACM, 2010.
  21. Kim, Y. G., Kim, M., Kim, et al, "A novel GPU power model for accurate smartphone power breakdown," ETRI Journal, vol. 37, no. 1, pp. 157-164, 2015. https://doi.org/10.4218/etrij.14.0113.1411
  22. Ma, X., Dong, M., et al, "Statistical power consumption analysis and modeling for GPU-based computing," in Proceedings of ACM SOSP Workshop on Power Aware Computing and Systems, 2009.
  23. Pool, J., Lastra, A., & Singh, M., "An energy model for graphics processing units," in Proceedings of IEEE International Conference on ICCD, pp. 409- 416, 2010.
  24. Ryoo, S., Rodrigues, C. I., et al, "Optimization principles and application performance evaluation of a multithreaded GPU using CUDA," in Proceedings of ACM SIGPLAN Symposium on Principles and practice of parallel programming, ACM, pp. 73-82, 2008.
  25. Hong, S., Kim, H., "An analytical model for a GPU architecture with memory-level and thread-level parallelism awareness," ACM SIGARCH Computer Architecture News, vol. 37. no. 3, ACM, 2009.
  26. Sarajedini, Amir, R. Hecht-Nielson, "The best of both worlds: Casasent networks integrate multilayer perceptrons and radial basis functions," in Proceedings of IEEE International Joint Conference on Neural Networks, vol. 3, pp.905-910, 1992.
  27. Wu, G., Greathouse, J. L., et al, "GPGPU performance and power estimate on using machine learning," in Proceedings of IEEE International Conference on HPCA, pp. 564-576, 2015.
  28. Leng, J., Hetherington, T., et al, "GPUWattch: Enabling Energy Optimizations in GPGPUs," ACM SIGARCH Computer Architecture News, vol. 41, no. 3, ACM, 2013.
  29. Bailey, P. E., Lowenthal, D. K., et al, "Adaptive Configuration Selection for Power-Constrained Heterogeneous Systems," in Proceedings of IEEE International Conference on ICPP, pp. 371-380, 2014.
  30. Ma, K., Li, X., et al, "GreenGPU: A Holistic Approach to Energy Efficiency in GPU-CPU Heterogeneous Architectures," in Proceedings of IEEE International Conference on ICPP, pp. 48-57, 2012.
  31. Baghsorkhi, S. S., Delahaye, M., et al, "An Adaptive Performance Modeling Tool for GPU Architectures," in Proceedings of IEEE International Conference on PPoPP, pp. 105-114, 2010.
  32. Madougou S, Varbanescu A L, De Laat C, et al. "The landscape of GPGPU performance modeling tools," Parallel Computing, vol. 56, pp. 18-33, 2016. https://doi.org/10.1016/j.parco.2016.04.002