论文信息 - On Benchmarking Collective MPI Operations

On Benchmarking Collective MPI Operations

This article concentrates on recent work on benchmarking collective operations with SKaMPI. The goal of the SKaMPI project is the creation of a database containing performance measurements of parallel computers in terms of MPI operations. These data support software developers in creating portable and fast programs. Existing algorithms for measuring the timing of collective operations are discussed and a new algorithm is presented, taking into account the differences of local clocks. Results of measurements on a Cray T3E/900 and an IBM RS 6000 SP are presented.

Werner Augustin | Ralf H. Reussner | Thomas Worsch

[1] Ralf H. Reussner,et al. SKaMPI: A Detailed, Accurate MPI Benchmark , 1998, PVM/MPI.

[2] Jack Dongarra,et al. MPI: The Complete Reference , 1996 .

[3] William Gropp,et al. Mpi---the complete reference: volume 1 , 1998 .

[4] Ralf H. Reussner,et al. Achieving Performance Portability with SKaMPI for High-Performance MPI Programs , 2001, International Conference on Computational Science.

[5] Bronis R. de Supinski,et al. Accurately measuring MPI broadcasts in a computational grid , 1999, Proceedings. The Eighth International Symposium on High Performance Distributed Computing (Cat. No.99TH8469).

[6] Rolf Rabenseifner. Die geregelte logische Uhr, eine globale Uhr für die tracebasierte Überwachung paralleler Anwendungen , 2000 .

[7] William Gropp,et al. Reproducible Measurements of MPI Performance Characteristics , 1999, PVM/MPI.