Abstract
EULAG (Eulerian/semi-Lagrangian fluid solver) is an established computational model for simulating thermo-fluid flows across a wide range of scales and physical scenarios. The multidimensional positive defined advection transport algorithm (MPDATA) is among the most time-consuming components of EULAG.
The main aim of our work is to design an efficient adaptation of the MPDATA algorithm to the NVIDIA GPU Kepler architecture. We focus on analysis of resources usage in the GPU platform and its influence on performance results. In this paper, a performance model is proposed, which ensures a comprehensive analysis of the resource consumption including registers, shared, global and texture memories. The performance model allows us to identify bottlenecks of the algorithm, and shows directions of optimizations.
The group of the most common bottlenecks is considered in this work. They include data transfers between host memory and GPU global memory, GPU global memory and shared memory, as well as latencies and serialization of instructions, and GPU occupancy. We put the emphasis on providing a fixed memory access pattern, padding, reducing divergent branches and instructions latencies, as well as organizing computation in the MPDATA algorithm in order to provide efficient shared memory and register file reusing.
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
Similar content being viewed by others
References
Cecilia, J.M., GarcÃa, J.M., Ujaldón, M.: Cuda 2D stencil computations for the Jacobi method. In: Jónasson, K. (ed.) PARA 2010, Part I. LNCS, vol. 7133, pp. 173–183. Springer, Heidelberg (2012)
Ciznicki, M., Kopta, P., Kulczewski, M., Kurowski, K., Gepner, P.: Elliptic solver performance evaluation on modern hardware architectures. In: Wyrzykowski, R., Dongarra, J., Karczewski, K., Wasniewski, J. (eds.) PPAM 2013, Part I. LNCS, vol. 8384, pp. 155–165. Springer, Heidelberg (2014)
de la Cruz, R., Araya-Polo, M., Cela, J.M.: Introducing the semi-stencil algorithm. In: Wyrzykowski, R., Dongarra, J., Karczewski, K., Wasniewski, J. (eds.) PPAM 2009, Part I. LNCS, vol. 6067, pp. 496–506. Springer, Heidelberg (2010)
Hager, A., Wellein, G.: Introduction to High Performance Computing for Science and Engineers. CRC Press, Boca Raton (2011)
Kurowski, K., Kulczewski, M., Dobski, M.: Parallel and GPU based strategies for selected CFD and climate modeling models. Environ. Sci. Eng. 3, 735–747 (2011)
Nguyen, A., Satish, N., Chhugani, J., Changkyu, K., Dubey, P.: 3.5-D blocking optimization for stencil computations on modern CPUs and GPUs. In: Proceedings of the 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis, pp. 1–13 (2010)
NVIDIA Kepler Compute Architecture. http://www.nvidia.com/object/nvidia-kepler.html
Rojek, K., Szustak, L.: Parallelization of EULAG model on multicore architectures with GPU accelerators. In: Wyrzykowski, R., Dongarra, J., Karczewski, K., Waśniewski, J. (eds.) PPAM 2011, Part II. LNCS, vol. 7204, pp. 391–400. Springer, Heidelberg (2012)
Smolarkiewicz, P.: Multidimensional positive definite advection transport algorithm: an overview. Int. J. Numer. Meth. Fluids 50, 1123–1144 (2006)
Szustak, L., Rojek, K., Gepner, P.: Using Intel Xeon Phi coprocessor to accelerate computations in MPDATA algorithm. In: Wyrzykowski, R., Dongarra, J., Karczewski, K., Wasniewski, J. (eds.) PPAM 2013, Part I. LNCS, vol. 8384, pp. 582–592. Springer, Heidelberg (2014)
Wyrzykowski, R., Rojek, K., Szustak, L.: Using Blue Gene/P and GPUs to accelerate computations in the EULAG model. In: Lirkov, I., Margenov, S., Waśniewski, J. (eds.) LSSC 2011. LNCS, vol. 7116, pp. 670–677. Springer, Heidelberg (2012)
Wyrzykowski, R., Szustak, L., Rojek, K., Tomas, A.: Towards efficient decomposition and parallelization of MPDATA on hybrid CPU-GPU cluster. In: LSSC 2013. LNCS (in print)
Acknowledgments
This work was partly supported by the Polish National Science Centre under grant no. UMO-2011/03/B/ST6/03500.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2014 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Rojek, K., Szustak, L., Wyrzykowski, R. (2014). Performance Analysis for Stencil-Based 3D MPDATA Algorithm on GPU Architecture. In: Wyrzykowski, R., Dongarra, J., Karczewski, K., Waśniewski, J. (eds) Parallel Processing and Applied Mathematics. PPAM 2013. Lecture Notes in Computer Science(), vol 8384. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-55224-3_15
Download citation
DOI: https://doi.org/10.1007/978-3-642-55224-3_15
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-55223-6
Online ISBN: 978-3-642-55224-3
eBook Packages: Computer ScienceComputer Science (R0)