Advertisement

Analysis of a sparse hypermatrix Cholesky with fixed-sized blocking

  • José R. Herrero
  • Juan J. Navarro
Article

Abstract

We present the way in which we have constructed an implementation of a sparse Cholesky factorization based on a hypermatrix data structure. This data structure is a storage scheme which produces a recursive 2D partitioning of a sparse matrix. It can be useful on some large sparse matrices. Subblocks are stored as dense matrices. Thus, efficient BLAS3 routines can be used. However, since we are dealing with sparse matrices some zeros may be stored in those dense blocks. The overhead introduced by the operations on zeros can become large and considerably degrade performance. We present the ways in which we deal with this overhead. Using matrices from different areas (Interior Point Methods of linear programming and Finite Element Methods), we evaluate our sequential in-core hypermatrix sparse Cholesky implementation. We compare its performance with several other codes and analyze the results. In spite of using a simple fixed-size partitioning of the matrix our code obtains competitive performance.

Keywords

Sparse Cholesky Hypermatrix structure 2D partitioning Windows in submatrices Small Matrix Library 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Ashcraft C. and Grimes R.G. (1989). The influence of relaxed supernode partitions on the multifrontal method. ACM Trans. Math. Softw. 15: 291–309 zbMATHCrossRefGoogle Scholar
  2. 2.
    Ast, M., Barrado, C., Cela, J., Fischer, R., Laborda, O., Manz, H., Schulz, U.: Sparse matrix structure for dynamic parallelisation efficiency. In: Euro-Par 2000, LNCS, vol. 1900, pp. 519–526 (2000)Google Scholar
  3. 3.
    Ast, M., Fischer, R., Manz, H., Schulz, U.: PERMAS: user’s reference manual. INTES publication no. 450, rev.d (1997)Google Scholar
  4. 4.
    Badics, T.: RMFGEN generator (1991). ftp://dimacs.rutgers.edu/pub/netflow/generators/ network/genrmfGoogle Scholar
  5. 5.
    Carolan W., Hill J., Kennington J., Niemi S. and Wichmann S. (1990). An empirical evaluation of the KORBX algorithms for military airlift applications. Oper. Res. 38: 240–248 CrossRefGoogle Scholar
  6. 6.
    Castro J. (2000). A specialized interior-point algorithm for multicommodity network flows. SIAM J. Optim. 10(3): 852–877 zbMATHCrossRefMathSciNetGoogle Scholar
  7. 7.
    Czyzyk, J., Mehrotra, S., Wagner, M., Wright, S.J.: PCx user’s guide (version 1.1). Technical Report OTC 96/01, Optimization Technology Center, Mathematics and Computer Science Division, Argonne National Laboratory, Argonne (1997). http://www.mcs.anl.gov/otc/Tools/ PCxGoogle Scholar
  8. 8.
    Davis, T.: University of Florida Sparse Matrix Collection. NA Digest 97(23) (1997). http://www.cise.ufl.edu/research/sparse/matricesGoogle Scholar
  9. 9.
    Duff, I.S.: Full matrix techniques in sparse Gaussian elimination. In: Numerical Analysis (Dundee, 1981). Lecture Notes in Mathematics, vol. 912, pp. 71–84. Springer, Berlin (1982)Google Scholar
  10. 10.
    Fischer, R., Ast, M., Manz, H., Labarta, J.: A dynamic task graph parallelization approach. In: Fourth International Colloquium on Computation of Shell and Spatial Structures (2000)Google Scholar
  11. 11.
    Frangioni, A.: Multicommodity Min Cost Flow problems. Operations Research Group, Department of Computer Science, University of Pisa. http://www.di.unipi.it/di/groups/optimize/Data/Google Scholar
  12. 12.
    Fuchs G., Roy J. and Schrem E. (1972). Hypermatrix solution of large sets of symmetric positive-definite linear equations. Comp. Meth. Appl. Mech. Eng. 1: 197–216 CrossRefzbMATHGoogle Scholar
  13. 13.
    Goldberg, A.V., Oldham, J.D., Plotkin, S., Stein, C.: An implementation of a combinatorial approximation algorithm for minimum-cost multicommodity flow. In: Proceedings of the 6th International Conference on Integer Programming and Combinatorial Optimization, IPCO’98 (Houston, June 22–24, 1998), LNCS, vol. 1412, pp. 338–352. Springer, Berlin (1998)Google Scholar
  14. 14.
    Goldfarb, D., Grigoriadis, M.D.: A computational comparison of the Dinic and network simplex methods for maximum flow. In: Simeone, B., et al. (eds.) FORTRAN Codes for Network Optimization. Annals of Operations Research, vol. 13, pp. 83–124 (1988)Google Scholar
  15. 15.
    Herrero, J.R., Navarro, J.J.: Improving performance of hypermatrix Cholesky factorization. In: Euro-Par’03, LNCS, vol. 2790, pp. 461–469. Springer, Berlin (2003)Google Scholar
  16. 16.
    Herrero, J.R., Navarro, J.J.: Optimization of a statically partitioned hypermatrix sparse Cholesky factorization. In: Workshop on State-of-the-Art in Scientific Computing (PARA’04), LNCS, vol. 3732, pp. 798–807. Springer, Berlin (2004)Google Scholar
  17. 17.
    Herrero, J.R., Navarro, J.J.: Reducing overhead in sparse hypermatrix Cholesky factorization. In: IFIP TC5 Workshop on High Performance Computational Science and Engineering (HPCSE), World Computer Congress, pp. 143–154. Springer, Berlin (2004)Google Scholar
  18. 18.
    Irony, D., Shklarski, G., Toledo, S.: Parallel and fully recursive multifrontal sparse Cholesky. In: ICCS’02, LNCS, vol. 2330, pp. 335–344. Springer, Berlin (2002)Google Scholar
  19. 19.
    Karypis G. and Kumar V. (1999). A fast and high quality multilevel scheme for partitioning irregular graphs. SIAM J. Sci. Comput. 20(1): 359–392 zbMATHCrossRefMathSciNetGoogle Scholar
  20. 20.
    Koenker, R., Ng, P.: SparseM: A Sparse Matrix Package for R (2003). http://cran.r-project.org/ src/contrib/PACKAGES.html#SparseMGoogle Scholar
  21. 21.
    Lee, Y., Orlin, J.: GRIDGEN generator (1991). ftp://dimacs.rutgers.edu/pub/netflow/ generators/network/gridgenGoogle Scholar
  22. 22.
    Liu J.H.W. (1990). The role of elimination trees in sparse factorization. SIAM J. Matrix Anal. Appl. 11(1): 134–172 zbMATHCrossRefMathSciNetGoogle Scholar
  23. 23.
    Liu J.W., Ng E.G. and Peyton B.W. (1993). On finding supernodes for sparse matrix computations. SIAM J. Matrix Anal. Appl. 14(1): 242–252 zbMATHCrossRefMathSciNetGoogle Scholar
  24. 24.
    Navarro, J.J., Juan, A., Lang, T.: MOB forms: A class of Multilevel Block Algorithms for dense linear algebra operations. In: Proceedings of the 8th International Conference on Supercomputing, pp. 354–363. ACM Press, New York (1994)Google Scholar
  25. 25.
    NetLib: Linear programming problems. http://www.netlib.org/lp/Google Scholar
  26. 26.
    Ng E.G. and Peyton B.W. (1993). Block sparse Cholesky algorithms on advanced uniprocessor computers. SIAM J. Sci. Comput. 14(5): 1034–1056 zbMATHCrossRefMathSciNetGoogle Scholar
  27. 27.
    Noor A. and Voigt S. (1975). Hypermatrix scheme for the STAR–100 computer. Comp. Struct. 5: 287–296 CrossRefGoogle Scholar
  28. 28.
    PARASOL: (EU ESPRIT IV LTR Project No. 20160). http://www.parallab.uib.no/projects/ parasol/Google Scholar
  29. 29.
    Rothberg E. (1996). Performance of panel and block approaches to sparse Cholesky factorization on the iPSC/860 and Paragon multicomputers. SIAM J. Sci. Comput. 17(3): 699–713 zbMATHCrossRefMathSciNetGoogle Scholar
  30. 30.
    Rothberg E. and Gupta A. (1993). An evaluation of left-looking, right-looking and multifrontal approaches to sparse Cholesky factorization on hierarchical-memory machines. Int. J. High Speed Comput. 5(4): 537–593 CrossRefGoogle Scholar
  31. 31.
    Rothberg E. and Gupta A. (1994). An efficient block-oriented approach to parallel sparse Cholesky factorization. SIAM J. Sci. Comput. 15(6): 1413–1439 zbMATHCrossRefMathSciNetGoogle Scholar
  32. 32.
    Whaley, R.C., Dongarra, J.J.: Automatically tuned linear algebra software. In: Supercomputing ’98, pp. 211–217. IEEE Computer Society (1998)Google Scholar
  33. 33.
    Woo, S.C., Ohara, M., Torrie, E., Singh, J.P., Gupta, A.: The SPLASH-2 programs: characterization and methodological considerations. In: Proceedings of the 22nd Annual International Symposium on Computer Architecture, pp. 24–36. ACM Press, New York (1995). http://doi.acm.org/10.1145/223982.223990Google Scholar
  34. 34.
    Zhang Y. (1998). Solving large-scale linear programs by interior-point methods under the MATLAB environment. Optim. Methods Softw. 10(1): 1–31 zbMATHCrossRefMathSciNetGoogle Scholar

Copyright information

© Springer-Verlag 2007

Authors and Affiliations

  1. 1.Computer Architecture DepartmentUniversitat Politècnica de CatalunyaBarcelonaSpain

Personalised recommendations