Parallel Processing on Shared Memory Multi-User Systems

  • V. R. Saunders


The present notes attempt to give a feeling for the techniques available to the programmer when using shared memory parallel processing systems, such as the CRAY-XMP,-YMP,-2, CONVEX C-2, IBM 3090, SEQUENT Balance and Symmetry, or the ENCORE Multimax series of systems. We hope to present:
  1. (a)

    A style of parallel programming (within the context of FORTRAN77) which is portable over this range of machines, although unfortunately the lack of language standardization precludes the portability of parallel syntax at the present time. It is intended that the style give rise to efficient use of such systems taking into account both the existence of shared memory and the fact that each computing node may be given over to multi-user operation; by way of contrast, in distributed memory systems, such as the Intel iPSC/2 hypercube or the Meiko transputer based series, it is normal for each node to be dedicated to a single user. We assert that whilst multi-user nodes should be inherently the more efficient and flexible, it is necessary to take account of the effects of the multi-user environment when evolving a programming style.

  2. (b)

    An explanation of much of the jargon of the subject

  3. (c)

    An indication of the range of applicability of parallel concepts in the calculation of molecular electronic structure by the usual methods of quantum chemistry.



Shared Memory Global Index Common Block Programming Style Molecular Electronic Structure 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Plenum Press, New York 1990

Authors and Affiliations

  • V. R. Saunders
    • 1
  1. 1.SERC Daresbury LaboratoryDaresbury, Warrington, CheshireUK

Personalised recommendations