Evaluations of HPF for practical scientific algorithms on T3E

  • Chris H. Q. Ding
2. Computational Science
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1401)


HPF coding examples of practical scientific algorithms are examined in detail, with the idea that on these simple but non-trivial examples, we can fairly well understand issues related to different data distributions, different parallel constructs, and different programming styles (static vs dynamic allocations). Coding examples include 2D stencils solution of PDEs, N-body problem, LU factorization, several vector/matrix library routines, 2D and 3D array redistribution.The performances of HPF codes are close to hand-written message passing MPI codes, for LU factorization, vector/matrix routines, array redistributions. But for less regular data/communication patterns, the stencils calculation and the N-body problem, HPF codes perform considerably less efficient, about 2-4 times slower. Scaling of HPF codes is not as good as MPI codes. Some of the HPF codes performances are highly inconsistent, i.e., minor change of code could results in factor of 10 change in performance. Many peculiarities of HPF coding will be discussed.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    High Performance Fortran Language Specification, version The High Performance Fortran Forum. The web site includes a list of comprehensive and useful links.Google Scholar
  2. [2]
    The High Performance Fortran Handbook, by C.H. Koelbel, D.B. Loveman, R.S. Schreiber, G.L. Steele, and M.E. Zosel, MIT Press, Cambridge, Massachusetts, 1994.Google Scholar
  3. [3]
    Solving Problems on Concurrent Processors, by G. Fox, M. Johnson, G. Lyzenga, S. Otto, J. Salmon, D. Walker. Prentice Hall, Engelwood Cliff, NJ, 1988.Google Scholar
  4. [4]
    Proceedings of First Conference of HPF Users Group, Santa Fe, New Mexico, February 1997.Google Scholar
  5. [5]
    A Portable 3D FFT Package for Distributed-Memory Parallel Architectures, H.Q. Ding, R.D. Ferraro, D.B. Gennery, in Proceedings of 7th SIAM Conference on Parallel Processing. p.70.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1998

Authors and Affiliations

  • Chris H. Q. Ding
    • 1
  1. 1.National Energy Research Scientific Computing Center Lawrence Berkeley National LaboratoryUniversity of CaliforniaBerkeleyUSA

Personalised recommendations