Message passing and shared address space parallelism on an SMP cluster
Journal Article
·
· Parallel Computing
OSTI ID:825127
- LBNL Library
Currently, message passing (MP) and shared address space (SAS) are the two leading parallel programming paradigms. MP has been standardized with MPI, and is the more common and mature approach; however, code development can be extremely difficult, especially for irregularly structured computations. SAS offers substantial ease of programming, but may suffer from performance limitations due to poor spatial locality and high protocol overhead. In this paper, we compare the performance of and the programming effort required for six applications under both programming models on a 32-processor PC-SMP cluster, a platform that is becoming increasingly attractive for high-end scientific computing. Our application suite consists of codes that typically do not exhibit scalable performance under shared-memory programming due to their high communication-to-computation ratios and/or complex communication patterns. Results indicate that SAS can achieve about half the parallel efficiency of MPI for mo st of our applications, while being competitive for the others. A hybrid MPI + SAS strategy shows only a small performance advantage over pure MPI in some cases. Finally, improved implementations of two MPI collective operations on PC-SMP clusters are presented.
- Research Organization:
- Ernest Orlando Lawrence Berkeley National Laboratory, Berkeley, CA (US)
- Sponsoring Organization:
- USDOE Director. Office of Science. Office of Computational and Technology Research; National Science Foundation Grant ESS-9806751; PE-CASE, Sloan Research Fellowship (US)
- DOE Contract Number:
- AC03-76SF00098
- OSTI ID:
- 825127
- Report Number(s):
- LBNL--53114
- Journal Information:
- Parallel Computing, Journal Name: Parallel Computing Vol. 29; ISSN PACOEJ; ISSN 0167-8191
- Country of Publication:
- United States
- Language:
- English
Similar Records
Message passing vs. shared address space on a cluster of SMPs
Study of parallel efficiency in message passing environments
Comparative Study of Message Passing and Shared Memory Parallel Programming Models in Neural Network Training
Conference
·
Sun Jan 21 23:00:00 EST 2001
·
OSTI ID:787126
Study of parallel efficiency in message passing environments
Conference
·
Thu Feb 29 23:00:00 EST 1996
·
OSTI ID:211366
Comparative Study of Message Passing and Shared Memory Parallel Programming Models in Neural Network Training
Conference
·
Mon Dec 13 23:00:00 EST 1999
·
OSTI ID:791052