Abstract
HPC developers aim to deliver the very best performance. To do so they constantly think about memory bandwidth, memory hierarchy, locality, floating point performance, power/energy constraints and so on. On the other hand, application scientists aim to write performance portable code while exploiting the rich feature set of the hardware. By providing adequate hints to the compilers in the form of directives appropriate executable code is generated. There are tremendous benefits from using directive-based programming. However, applications are also becoming more and more complex and we need sophisticated tools such as auto-tuning to better explore the optimization space. In applications, loops typically form a major and time-consuming portion of the code. Scheduling these loops involves mapping from the loop iteration space to the underlying platform - for example GPU threads. The user tries different scheduling techniques until the best one is identified. However, this process can be quite tedious and time consuming especially when it is a relatively large application, as the user needs to record the performance of every schedule’s run. This paper aims to offer a better solution by proposing an auto-tuning framework that adopts an analytical model guiding the compiler and the runtime to choose an appropriate schedule for the loops, automatically and determining the launch configuration for each of the loop schedules. Our experiments show that the predicted loop schedule by our framework achieves the speedup of 1.29x on an average against the default loop schedule chosen by the compiler.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
EPCC OpenACC Benchmarks (2015). https://www.epcc.ed.ac.uk/research/computing/performance-characterisation-and-benchmarking/epcc-openacc-benchmark-suite
KernelGen Performance Test Suite, December 2015. https://hpcforge.org/plugins/mediawiki/wiki/kernelgen/index.php/Performance_Test_Suite
OpenACC (2016). http://www.openacc.org
Almási, G., Caşcaval, C., Padua, D.A.: Calculating stack distances efficiently. In: ACM SIGPLAN Notices, vol. 38, pp. 37–43. ACM (2002)
Baghsorkhi, S.S., Delahaye, M., Gropp, W.D., Wen-mei, W.H..: Analytical performance prediction for evaluation and tuning of GPGPU applications. In: Workshop on EPHAM2009, in Conjunction with CGO, Citeseer (2009)
Beyls, K., Hollander, E.D.: Reuse distance as a metric for cache behavior. In: Proceedings of the IASTED Conference on Parallel and Distributed Computing and Systems, vol. 14, pp. 350–360 (2001)
Choi, K.H., Kim, S.W.: Study of cache performance on GPGPU. IEIE Trans. Smart Process. Comput. 4(2), 78–82 (2015)
Cui, X., Chen, Y., Zhang, C., Mei, H.: Auto-tuning dense matrix multiplication for GPGPU with cache. In: IEEE 16th International Conference on Parallel and Distributed Systems (ICPADS), pp. 237–242. IEEE (2010)
Grauer-Gray, S., Xu, L., Searles, R., Ayalasomayajula, S., Cavazos, J.: Auto-tuning a high-level language targeted to GPU codes. In: Innovative Parallel Computing (InPar), pp. 1–10. IEEE (2012)
Hu, Y., Koppelman, D.M., Brandt, S.R., Löffler, F.: Model-driven auto-tuning of stencil computations on GPUs. In: Proceedings of the 2nd International Workshop on High-Performance Stencil Computations, pp. 1–8 (2015)
Lee, H., Brown, K.J., Sujeeth, A.K., Rompf, T., Olukotun, K.: Locality-aware mapping of nested parallel patterns on GPUs. In: 47th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO), pp. 63–74. IEEE (2014)
Mametjanov, A., Lowell, D., Ma, C.-C., Norris, B.: Autotuning stencil-based computations on GPUs. In: IEEE International Conference on Cluster Computing (CLUSTER), pp. 266–274. IEEE (2012)
Montgomery, C., Overbey, J.L., Li, X.: Autotuning openACC work distribution via direct search. In: Proceedings of the 2015 XSEDE Conference: Scientific Advancements Enabled by Enhanced Cyberinfrastructure, p. 38. ACM (2015)
Nugteren, C., van den Braak, G.-J., Corporaal, H., Bal, H.: A detailed GPU cache model based on reuse distance theory. In: High Performance Computer Architecture (HPCA), pp. 37–48. IEEE (2014)
Picchi, J., Zhang, W.: Impact of L2 cache locking on GPU performance. In: SoutheastCon, pp. 1–4. IEEE (2015)
Siddiqui, S., AlZayer, F., Feki, S.: Historic learning approach for auto-tuning openACC accelerated scientific applications. VECPAR-2014. LNCS, vol. 8969, pp. 224–235. Springer, Heidelberg (2014)
Tang, T., Yang, X., Lin, Y.: Cache miss analysis for GPU programs based on stack distance profile. In: 31st International Conference on Distributed Computing Systems (ICDCS), pp. 623–634. IEEE (2011)
Tian, X., Xu, R., Yan, Y., Yun, Z., Chandrasekaran, S., Chapman, B.: Compiling a high-level directive-based programming model for GPGPUs. LCPC 2013. LNCS, vol. 8664, pp. 105–120. Springer International Publishing, New York (2014)
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
Xu, R., Chandrasekaran, S., Tian, X., Chapman, B. (2016). An Analytical Model-Based Auto-tuning Framework for Locality-Aware Loop Scheduling. In: Kunkel, J., Balaji, P., Dongarra, J. (eds) High Performance Computing. ISC High Performance 2016. Lecture Notes in Computer Science(), vol 9697. Springer, Cham. https://doi.org/10.1007/978-3-319-41321-1_1
Download citation
DOI: https://doi.org/10.1007/978-3-319-41321-1_1
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-41320-4
Online ISBN: 978-3-319-41321-1
eBook Packages: Computer ScienceComputer Science (R0)