Advertisement

Improving the PVM daemon network performance by direct network access

  • Ron Lavi
  • Amnon Barak
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1497)

Abstract

One approach to reduce the multi-layer overhead of TCP/IP is to bypass it, accessing the LAN directly from user level. Most existing implementations of such a shortcut rely on a polling mechanism, they either support a single-user environment, or do not coexist with TCP/IP. In order to improve the PVM daemon network performance, we developed a new Direct Network Access (DNA) library for PVM and the Myrinet LAN, that bypasses and coexists with TCP/IP. The DNA includes a non-polling receive mechanism, an efficient mutual exclusion mechanism and a shared buffer management scheme, thus enabling operation in a time sharing environment. The paper describes the DNA and its integration in the PVM daemon, and presents performance results, which amount to a latency reduction of 18% for PVM tasks.

Keywords

Packet Size Mutual Exclusion User Program Buffer Management Fast Message 
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.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    N.J. Boden, D. Cohen, R.E. Felderman, A.K. Kulawik, C.L. Seitz, J.N. Seizovic, and W-K. Su. Myrinet: A Gigabit-per-Second Local Area Network. IEEE Micro, 15(1):29–36, February 1995.CrossRefGoogle Scholar
  2. 2.
    A. Geist, A Beguelin, J. Dongarra, W Jiang, R. Manchek, and V. Sunderam. PVM: Parallel Virtual Machine, A Users ' Guide and Tutorial for Networked Parallel Computing. The MIT Press, 1994.Google Scholar
  3. 3.
    I. Gilderman. Performance of the Communication Layers in Gigabit LANs. Master's thesis, Institute of Computer Science, The Hebrew University of Jerusalem, 1997.Google Scholar
  4. 4.
    M. Lauria and A. Chien. MPI-FM: High Performance MPI on Workstation Clusters. Journal of Parallel and Distributed Computing, 40(1):4–18, January 1997.CrossRefGoogle Scholar
  5. 5.
    A. M. Mainwaring and D. E. Culler. Active Message Applications Programming Interface and Communication Subsystem Organization. Technical Report CSD-96-918, U. C. Berkeley, October 1996.Google Scholar
  6. 6.
    M. K. McKusick, K. Bostic, M. J. Karels, and J. S. Quarterman. The Design and Implementation of the 4.4BSD Operating System. Addison-Wesley Publishing Company, 1996.Google Scholar
  7. 7.
    S. Pakin, M. Lauria, and A. Chien. High Performance Messaging on Workstations: Illinois Fast Messages (FM) for Myrinet. In Supercomputing, December 1995.Google Scholar
  8. 8.
    H. Tezuka, A. Hori, and Y. Ishikawa. PM: A High-Performance Communication Library for Multi-user Paralllel Environments. Technical Report TR-96-015, Tsukuba Research Center, Real World Computing Partnership, November 1996.Google Scholar
  9. 9.
    T. von Eicken, A. Basu, V. Buch, and W. Vogels. U-Net: A User-Level Network Interface for Parallel and Distributed Computing. In Proc. of the 15th ACM Symposium on Operating Systems principles, December 1995.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1998

Authors and Affiliations

  • Ron Lavi
    • 1
  • Amnon Barak
    • 1
  1. 1.Institute of Computer ScienceThe Hebrew University of JerusalemJerusalemIsrael

Personalised recommendations