Parallel Simulation of Scramjet with Multilevel Hypergraph Partitioning

Abstract:

As one of the most considerable methods to study hypersonic flight vehicle, the numerical simulation of supersonic combustion ramjet (scramjet) has drawn an ever increasing attention at present. Nevertheless, the traditional serial simulation is ungratified for current research requirements because of high calculation precision, avaricious memory overhead and overlong computation time. Meanwhile, the efficiency of parallel simulation using the domain decomposition method is not very satisfactory. In this paper, we study on a general algorithm for scramjet design, and subdivide the computing domain by using a multilevel hypergraph partitioning algorithm. In order to reduce computation while enhancing the degree of parallelism, overlapping communication with computation and non blocking communication is adopted to decrease the communication time when dealing with global communication. Finally, experimental results testing on a China-made supercomputer show the smallest value of parallel efficiency is more than 48% when the number of processors is 256. In conclusion, the result indicates that our parallel algorithm is simple, effective and practical in scramjet simulation.

• Periodical: Advanced Materials Research (Volumes 706-708)
• Pages: 1479-1482
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.706-708.1479
• Online since: June 2013
• Authors: Yao Yuan Zeng, Wen Tao Zhao, Zheng Hua Wang
• Keywords: Global Communication, Multilevel Hypergraph Partitioning, Numerical Simulation, Parallel Computing, Scramjet

References

[1] Kindler, M., et al. , "TASCOM3D: A Scientific Code for Compressible Reactive Flows," High Performance Computing on Vector Systems 2011 , 121-141(2012)

DOI: 10.1007/978-3-642-22244-3_9

[2] Ziegler, F., "n-Level Hypergraph Partitioning," ,2012)

[3] Catalyurek, U.V., et al. , "Dynamic Load Balancing for Adaptive Scientific Computations via Hypergraph Repartitioning," ,2009)

[4] Bansal, N., et al. , "Min-max graph partitioning and small set expansion," Foundations of Computer Science (FOCS), 2011 IEEE 52nd Annual Symposium on 17-26(2011)

DOI: 10.1109/focs.2011.79

[5] Conan-Guez, B. and F. Rossi, "Dissimilarity Clustering by Hierarchical Multi-Level Refinement," arXiv preprint arXiv:1204.6509 ,2012)

[6] Rital, S., "Hypergraph cuts & unsupervised representation for image segmentation," fundamenta informaticae 96(1), 153-179(2009)

DOI: 10.3233/fi-2009-172

[7] Marjanović, V., et al. , "Overlapping communication and computation by using a hybrid MPI/SMPSs approach," Proceedings of the 24th ACM International Conference on Supercomputing 5-16(2010)

DOI: 10.1145/1810085.1810091

[8] Kandalla, K., et al. , "High-performance and scalable non-blocking all-to-all with collective offload on InfiniBand clusters: a study with parallel 3D FFT," Computer Science-Research and Development 26(3), 237-246(2011)

DOI: 10.1007/s00450-011-0170-4

[9] Haack, J., "A hybrid OpenMP and MPI implementation of a conservative spectral method for the Boltzmann equation," arXiv preprint arXiv:1301.4195 ,2013)