Performance Evaluation of View-Oriented Parallel Programming

Sixth IEEE International Symposium on Cluster Computing and the Grid (CCGRID'06)

Purvis

et al. 2006

Self Cite

Section: Introductionmentioning

confidence: 99%

VODCA: View-Oriented, Distributed, Cluster-Based Approach to Parallel Computing

Huang

Sixth IEEE International Symposium on Cluster Computing and the Grid (CCGRID'06)

Purvis

et al. 2006

Self Cite

“…Much work has been done in the previous papers [3,5,4] to introduce optimizations in VOPP compared with traditional distributed shared memory systems. In the following section, we will describe our recent work on VOPP.…”

Section: View-oriented Parallel Programmingmentioning

confidence: 99%

“…The reason why there have to be more barriers in VOPP programs has been explained in [5]. Barriers have to be used to make sure the sequential consistency.…”

Section: Factors Contributing To the Performance Of Voppmentioning

confidence: 99%

Performance Evaluation of View-Oriented Parallel Programming on Cluster of Computers

Shang

Zhang

High Performance Computing and Communications

et al. 2007

Abstract. View-Oriented Parallel Programming(VOPP) is a novel programming style based on Distributed Shared Memory, which is friendly and easy for programmers to use. In this paper we compare VOPP with two other systems for parallel programming on clusters: LAM/MPI, a message passing system, and TreadMarks, a software distributed shared memory system. We present results for ten applications implemented and optimized using all the three systems. Experimental results demonstrate that VOPP is almost as efficient as Message Passing Interface when running on up to 32 processors, which means there is significant performance improvement compared with TreadMarks. The factors contributing to the performance of VOPP are discussed and analyzed. VOPP is still slower than MPI when the number of processes is large because of extra messages for separate synchronization and lack of bulk transfer mechanisms.

“…It has been integrated into Magpie [9]. -GE [20] A message passing implementation of the Gauss Elimination. It is an efficient algorithm for solving systems of linear equations.…”

Section: Benchmarks and Applicationsmentioning

confidence: 99%

Process Mapping for MPI Collective Communications

Zhang

Zhai

Lecture Notes in Computer Science

et al. 2009

Abstract.It is an important problem to map virtual parallel processes to physical processors (or cores) in an optimized way to get scalable performance due to non-uniform communication cost in modern parallel computers. Existing work uses profile-guided approaches to optimize mapping schemes to minimize the cost of point-to-point communications automatically. However, these approaches cannot deal with collective communications and may get sub-optimal mappings for applications with collective communications.In this paper, we propose an approach called OPP (Optimized Process Placement) to handle collective communications which transforms collective communications into a series of point-to-point communication operations according to the implementation of collective communications in communication libraries. Then we can use existing approaches to find optimized mapping schemes which are optimized for both point-to-point and collective communications.We evaluated the performance of our approach with micro-benchmarks which include all MPI collective communications, NAS Parallel Benchmark suite and three other applications. Experimental results show that the optimized process placement generated by our approach can achieve significant speedup.