Implementation Of Adetran Language Using Message Passing Interface

DOI DOI Open Access

Description

{"references": ["R. Numrich and J. Reid, \"Co-array Fortran for parallel programming,\"\nACM SIGPLAN Fortran Forum, vol. 17, issue 2, pp. 1-31, 1998.", "P. Charles, C. Grothoff, V. Saraswat, C. Donawa, A. Kielstra, K. Ebcioglu,\nC. Praun and V. Sarkar \"X10: an object-oriented approach to non-uniform\ncluster computing,\" ACM SIGPLAN Notices, vol. 40, issue 10, pp.\n519-538, 2005.", "B. Chamberlain, D. Callahan and H. Zima, \"Parallel Programmability\nand the Chapel Language,\" International Journal of High Performance\nComputing Applications, vol. 21, no. 3, pp. 291-312, 2007.", "M. Nakao, H. Murai, T. Shimosaka and M. Sato, \"Productivity and\nPerformance of the HPC Challenge Benchmarks with the XcalableMP\nPGAS Language\", in Proc. of the 7th International Conference on PGAS\nProgramming Models, pp. 157-171, 2013.", "C. Yang, W. Bland, J. Mellor-Crummey, P. Balaji, \"Portable,\nMPI-interoperable coarray fortran,\" ACM SIGPLAN Notices, vol. 49, issue\n8, pp. 81-92, 2014.", "H. Kadota, K. Kaneko, I. Okabayashi, T. Okamoto, T. Mimura, Y.\nNakakura, A. Wakatani, M. Nakajima, J. Nishikawa, K. Zaiki and T.\nNogi \"Pallel computer ADENART - its architecture and application,\" in\nProc. of the 5th international conference on Supercomputing, pp. 1-8,\n1991.", "S. Odanaka and T. Nogi, \"Massively parallel computation using a\nsplitting-up operator method for three-dimensional device simulation,\"\nIEEE Transactions on Computer-Aided Design of Integrated Circuits and\nSystems, vol. 14, issue 7, pp. 824-832, 1995.", "\"SX-ACE super computer,\" http://www.hpc.cmc.osaka-u.ac.jp/en/sx-ace/\n(as of 3/2/2016)", "\"GNU bison,\" http://www.gnu.org/software/bison/ (as of 3/2/2016)\n[10] \"ADETRAN translator,\" http://pplinux.is.konan-u.ac.jp/atran.html (as of\n3/2/2016)"]}

This paper describes the Message Passing Interface<br> (MPI) implementation of ADETRAN language, and its evaluation<br> on SX-ACE supercomputers. ADETRAN language includes pdo<br> statement that specifies the data distribution and parallel computations<br> and pass statement that specifies the redistribution of arrays. Two<br> methods for implementation of pass statement are discussed and the<br> performance evaluation using Splitting-Up CG method is presented.<br> The effectiveness of the parallelization is evaluated and the advantage<br> of one dimensional distribution is empirically confirmed by using the<br> results of experiments.

Details 詳細情報について

Report a problem

Back to top