Runtime message uniquification for accurate communication analysis on incomplete MPI event traces
Open Access
- 15 September 2013
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 123-128
Abstract
Communication analysis of parallel applications based on event traces depends on correct matching of associated MPI send and receive events. Selective monitoring techniques, however, may result in incomplete MPI event traces and, in that case, current matching strategies fail. In this paper we introduce an additional unique identifier for each message to make MPI events distinguishable from others. Therefore, it is possible to identify missing MPI events and match all remaining MPI events correctly. An overhead study with a real-life application and a benchmark suite demonstrates the applicability and benefits of this approach.Keywords
Funding Information
- Seventh Framework Programme (287703)
This publication has 10 references indexed in Scilit:
- Strategies for Real-Time Event Reduction2013
- Score-P: A Joint Performance Measurement Run-Time Infrastructure for Periscope, Scalasca, TAU, and Vampir2012
- GTI: A Generic Tools Infrastructure for Event-Based Tools in Parallel Systems2012
- Enhanced Encoding Techniques for the Open Trace Format 2Procedia Computer Science, 2012
- Characterizing the Influence of System Noise on Large-Scale Applications by Simulation2010
- The Vampir Performance Analysis Tool-Set2008
- GROMACS 4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular SimulationJournal of Chemical Theory and Computation, 2008
- On the Performance of Transparent MPI Piggyback MessagesLecture Notes in Computer Science, 2008
- The Tau Parallel Performance SystemThe International Journal of High Performance Computing Applications, 2006
- Implementation and Evaluation of a Scalable Application-Level Checkpoint-Recovery Scheme for MPI Programs2005