A comparison of data-parallel collective communication performance and its application
- 22 November 2002
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 137-144
- https://doi.org/10.1109/hpc.1997.592137
Abstract
Collective communications such as broadcast and reduction are commonly used in data parallel programs. It is important to understand the performance of such primitive communications to characterize parallel systems and analyze the performance of parallel applications running on specific parallel systems. We measured the performance of collective communication operations on several multiprocessor systems. In this paper, we report experimental results for collective communication performance on distributed memory systems. We also describe the performance prediction of data parallel programs using the performance of the primitives.Keywords
This publication has 3 references indexed in Scilit:
- Communication overhead on the CM5: an experimental performance evaluationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- A performance model and metrics for fine grain parallel computing systems-finding optimal parallelismPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Performance prediction of parallel processing systemsPublished by Association for Computing Machinery (ACM) ,1993