Integrating software distributed shared memory and message passing programming

H. J. Wong, A. P. Rendell

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Software Distributed Shared Memory (SDSM) systems provide programmers with a shared memory programming environment across distributed memory architectures. In contrast to the message passing programming environment, the SDSM can resolve data dependencies within the application without the programmer having to explicitly specify communication. However, this service is provided at a cost to performance. Thus it makes sense to use message passing directly when data dependencies are easy to solve using message passing. For example, it is not complicated to specify data transfer for large contiguous regions of memory. This paper outlines how the Danui SDSM library has been extended to include support for message passing. Four different message passing transfers are identified depending on whether the data being sent/received resides in private or globally shared buffers. Transfers between globally shared buffers are further categorized as symmetrical or asymmetrical depending on whether they correspond to the same region of shared memory. The implication of each transfer type on the memory consistency of the global address space is discussed. Central to the Danui SDSM extension is the use of information provided and implied by message passing operations. The overhead of the implementation is analyzed.

Original languageEnglish
Title of host publication2009 IEEE International Conference on Cluster Computing and Workshops, CLUSTER '09
DOIs
Publication statusPublished - 21 Dec 2009
Externally publishedYes
Event2009 IEEE International Conference on Cluster Computing and Workshops, CLUSTER '09 - New Orleans, LA, United States
Duration: 31 Aug 20094 Sep 2009

Publication series

NameProceedings - IEEE International Conference on Cluster Computing, ICCC
ISSN (Print)1552-5244

Conference

Conference2009 IEEE International Conference on Cluster Computing and Workshops, CLUSTER '09
CountryUnited States
CityNew Orleans, LA
Period31/08/094/09/09

Keywords

  • message passing
  • programming environments
  • data dependency

Fingerprint Dive into the research topics of 'Integrating software distributed shared memory and message passing programming'. Together they form a unique fingerprint.

  • Cite this

    Wong, H. J., & Rendell, A. P. (2009). Integrating software distributed shared memory and message passing programming. In 2009 IEEE International Conference on Cluster Computing and Workshops, CLUSTER '09 [5289176] (Proceedings - IEEE International Conference on Cluster Computing, ICCC). https://doi.org/10.1109/CLUSTR.2009.5289176