Fast “Short” Messages on a Linux Cluster

  • M. Danelutto
  • A. Rampini
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 2110)


We discuss an experiment aimed at lowering the operating system related overheads when performing small size message communications on a Beowulf class Linux PC cluster. The experiment consists in adding a small number of new system calls to the Linux kernel allowing user code to send/receive messages to/from remote processes. The system calls have been implemented using the standard kernel module mechanism provided by Linux. Those new system calls allow small size messages to be exchanged between cluster nodes with times that are 10 to 15% smaller than those achieved using standard TCP/IP communications.


Destination Node Interconnection Network System Call Kernel Module Kernel Space 
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.


  1. 1. Top500 supercomputer sites.
  2. 2. The beowulf project.
  3. 3.
    B. Bacci, M. Danelutto, S. Pelagatti, and M. Vanneschi. SkIE: a heterogeneous environment for HPC applications. Parallel Computing, 25:1827–1852, December 1999.CrossRefGoogle Scholar
  4. 4.
    M. Danelutto. Task farm computations in java. In Buback, Afsarmanesh, Williams, and Hertzberger, editors, High Performance Computing and Networking, LNCS, No. 1823, pages 385–394. Springer Verlag, May 2000.CrossRefGoogle Scholar
  5. 5.
    M. Cole. Algorithmic Skeletons: Structured Management of Parallel Computations. Research Monographs in Parallel and Distributed Computing. Pitman, 1989.Google Scholar
  6. 6.
    Congduc Pham and Carsten Albrecth. Tuning message aggregation on high performance clusters for efficient parallel simulations. Parallel Processing Letters, 9(4):521–532, 1999.CrossRefGoogle Scholar
  7. 7.
    G. Ciaccio and G. Chiola. GAMMA and MPI/GAMMA on Gigabit Ethernet. In Proceedings of 7th EuroPVM-MPI, LNCS, No. 1908. Springer Verlag, September 2000. Balatonfured, Hungary.Google Scholar
  8. 8.
    M. Beck, H. Bohme, M. Dziadska, U. Kunitz, R. Magnus, and D. Verworner. LINUX Kernel Internals. Addison Wesley, 1998.Google Scholar
  9. 9.
    D.A. Rusling. The Linux Kernel. version 0.8-3,
  10. 10.
    M. Snir, S.W. Otto, and J. Dongarra. MPI-The complete reference. Volume 2: The MPI core. MIT Press, Cambridge, 1998.Google Scholar
  11. 11.
    Argonne National Laboratory. Mpich home page.

Copyright information

© Springer-Verlag Berlin Heidelberg 2001

Authors and Affiliations

  • M. Danelutto
    • 1
  • A. Rampini
    • 1
  1. 1.Dept. of Computer ScienceUniversity of PisaItaly

Personalised recommendations