Abstract
Purely analytical worst-case execution time (WCET) estimation approaches for Graphics Processor Units (GPUs) cannot go far because of insufficient public information for the hardware. Therefore measurement-based probabilistic timing analysis (MBPTA) seems the way forward. We recently demonstrated MBPTA for GPUs, based on Extreme Value Theory (EVT) of the “Block Maxima” paradigm. In this newer work, we formulate and experimentally evaluate a more robust MBPTA approach based on the EVT “Peak over Threshold” paradigm with a complete set of tests for verifying EVT applicability. It optimally selects parameters to best-fit the input measurements for more accurate probabilistic WCET estimates. Different system configuration parameters (cache arrangements, thread block size) and their effect on the pWCET are considered, enhancing models of worst-case GPU behavior.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Intra-warp control flow divergence is handled with predicates/masking and NOPs.
- 2.
A random variable is a variable whose value is subject to variations due to chance; it can take on a set of possible different values, each with an associated probability.
- 3.
Admittedly, then the execution time is that of the modified kernel.
- 4.
The same holds for deterministic approaches, which derive safe WCET estimates from incomplete system models or assumptions about the system behavior.
- 5.
By extreme execution time measurements we intend execution time relatively far from the average values or relatively separated in time.
- 6.
As stated in [29], p. 47: “There are many factors involved in selecting block size, and inevitably some experimentation is required.”
References
Bakhoda, A., Yuan, G.L., Fung, W.W., Wong, H., Aamodt, T.M.: Analyzing CUDA workloads using a detailed GPU simulator. In: Proceedings of the IEEE ISPASS (2009)
Bautin, M., Dwarakinath, A., Chiueh, T.: Graphics engine resource management. In: Proceedings of the 15th ACM/SPIE MMCN (2008)
Berezovskyi, K., Bletsas, K., Andersson, B.: Makespan computation for GPU threads running on a single streaming multiprocessor. In: Proceedings of the 24th ECRTS (2012)
Berezovskyi, K., Bletsas, K., Petters, S.M.: Faster makespan estimation for GPU threads on a single streaming multiprocessor. In: Proceedings of the ETFA (2013)
Berezovskyi, K., Santinelli, L., Bletsas, K., Tovar, E.: WCET measurement-based and EVT characterisation of CUDA kernels. In: Proceedings of the RTNS (2014)
Betts, A., Donaldson, A.F.: Estimating the WCET of GPU-accelerated applications using hybrid analysis. In: Proceedings of the 25th ECRTS, pp. 193–202 (2013)
Brock, W., Scheinkman, J., Dechert, W., LeBaron, B.: A test for independence based on the correlation dimension. Econometric Rev. 15(3), 197–235 (1996)
Chen, X., Chang, L.-W., Rodrigues, C.I., Lv, J., Wang, Z., Hwu, W.-M.: Adaptive cache management for energy-efficient gpu computing. In: Proceedings of the 47th IEEE/ACM International Symposium on Microarchitecture, pp. 343–355 (2014)
Cucu-Grosjean, L., Santinelli, L., Houston, M., Lo, C., Vardanega, T., Kosmidis, L., Abella, J., Mezzeti, E., Quinones, E., Cazorla, F.J.: Measurement-based probabilistic timing analysis for multi-path programs. In: Proceedings of the 23nd ECRTS (2012)
Davis, R.I., Santinelli, L., Altmeyer, S., Maiza, C., Cucu-Grosjean, L.: Analysis of probabilistic cache related pre-emption delays. In: Proceedings of the 25th IEEE Euromicro Conference on Real-Time Systems (ECRTS) (2013)
Díaz, J., Garcia, D., Kim, K., Lee, C., Bello, L., Lopez, J.M., Mirabella, O.: Stochastic analysis of periodic real-time systems. In: 23rd RTSS, pp. 289–300 (2002)
Elliott, G., Ward, B., Anderson, J.: GPUSync: architecture-aware management of GPUs for predictable multi-GPU real-time systems. In: Proceedings of the RTSS (2013)
Embrechts, P., Klüppelberg, C., Mikosch, T.: Modelling Extremal Events for Insurance and Finance. Applications of mathematics. Springer, Heidelberg (1997)
Gumbel, E.: Statistics of Extremes. Columbia University Press, New York (1958)
Hirvisalo, V.: On static timing analysis of GPU kernels. In: Proceedings of the WCET (2014)
Hsing, T.: On tail index estimation using dependent data. Ann. Stat. 19(3), 1547–1569 (1991)
Kato, S., Lakshmanan, K., Kumar, A., Kelkar, M., Ishikawa, Y., Rajkumar, R.: RGEM: a responsive GPGPU execution model for runtime engines RTSS (2011)
Kato, S., Lakshmanan, K., Rajkumar, R., Ishikawa, Y.:Timegraph: GPU scheduling for real-time multi-tasking environments. In: USENIX ATC (2011)
Kato, S., McThrow, M., Maltzahn, C., Brandt, S.: Gdev: First-class GPU resource management in the operating system. In: Proceedings of the USENIX ATC (2012)
Kwiatkowski, D., Phillips, P.C.B., Schmidt, P., Shin, Y.: Testing the null hypothesis of stationarity against the alternative of a unit root: how sure are we that economic time series have a unit root? J. Econometrics 54, 1–3 (1992)
Leadbetter, M.R., Lindgren, G., Rootzén, H.: Conditions for the convergence in distribution of maxima of stationary normal processes. Stoch. Process. Appl. 8(2), 131–139 (1978)
Leadbetter, M.R., Lindgren, G., Rootzén, H.: Extremes and Related Properties of Random Sequences and Processes. Springer, New York (1983)
Lu, Y., Nolte, T., Bate, I., Cucu-Grosjean, L.: A statistical response-time analysis of real-time embedded systems. In: Proceedings of the RTSS, pp. 351–362 (2012)
Majdandzic, I., Trefftz, C., Wolffe, G.: Computation of voronoi diagrams using a graphics processing unit. In: IEEE International Conference on Electro/Information Technolog (EIT) (2008)
Mangharam, R., Saba, A.A.: Anytime algorithms for GPU architectures. In: Proceedings of the 32nd IEEE Real-Time Systems Symposium (RTSS) (2011)
Mejzler, D.: On the problem of the limit distribution for the maximal term of a variational series. Lvov. Politehn. Inst. Naucn Zap. Ser. Fiz. Mat. 38, 90–109 (1956)
Membarth, R., Lupp, J.-H., Hannig, F., Teich, J., Körner, M., Eckert, W.: Dynamic task-scheduling and resource management for GPU accelerators in medical imaging. In: Herkersdorf, A., Römer, K., Brinkschulte, U. (eds.) ARCS 2012. LNCS, vol. 7179, pp. 147–159. Springer, Heidelberg (2012)
Northrop, P.: Semiparametric estimation of the extremal index using block maxima. Technical report, Dept of Statistical Science, UCL (2005)
NVIDIA Corp. CUDA C Best Practices Guide. DG-05603-001\(\_\)v5.5
NVIDIA Corp. Whitepaper: Kepler GK110 (2012). www.nvidia.com/content/PDF/kepler/NVIDIA-Kepler-GK110-Architecture-Whitepaper.pdf
Rossbach, C.J., Currey, J., Silberstein, M., Ray, B., Witchel, E.: Ptask: Operating system abstractions to manage GPUs as computedevices. ACM SOSP (2011)
Santinelli, L., Morio, J., Dufour, G., Jacquemart, D.: On the sustainability of the extreme value theory for WCET estimation. In: International WCET Workshop (2014)
Acknowledgements
Work partially supported by National Funds through FCT/MEC (Portuguese Foundation for Science and Technology) and co-financed by ERDF (European Regional Development Fund) under the PT2020 Partnership, within project UID/CEC/04234/2013 (CISTER); also by FCT/MEC and the EU ARTEMIS JU within projects ARTEMIS/0003/2012 - JU grant 333053 (CONCERTO) and ARTEMIS/0001/2013 - JU grant 621429 (EMC2); by FCT/MEC and ESF (European Social Fund) through POPH (Portuguese Human Potential Operational Program), under PhD grant SFRH/BD/82069/2011.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2016 Springer International Publishing Switzerland
About this paper
Cite this paper
Berezovskyi, K., Guet, F., Santinelli, L., Bletsas, K., Tovar, E. (2016). Measurement-Based Probabilistic Timing Analysis for Graphics Processor Units. In: Hannig, F., Cardoso, J.M.P., Pionteck, T., Fey, D., Schröder-Preikschat, W., Teich, J. (eds) Architecture of Computing Systems – ARCS 2016. ARCS 2016. Lecture Notes in Computer Science(), vol 9637. Springer, Cham. https://doi.org/10.1007/978-3-319-30695-7_17
Download citation
DOI: https://doi.org/10.1007/978-3-319-30695-7_17
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-30694-0
Online ISBN: 978-3-319-30695-7
eBook Packages: Computer ScienceComputer Science (R0)