Tracing Data Movements within MPI Collectives

Kevin A. Brown, Jens Domke, Satoshi Matsuoka
2014 Proceedings of the 21st European MPI Users' Group Meeting on - EuroMPI/ASIA '14  
We propose extending common performance measurement and visualization tools to identify network bottlenecks within MPI collectives. By creating additional trace points in the Peruse utility of Open MPI, we track low-level InfiniBand communication events and then visualize the communication profile in Boxfish for a more comprehensive analysis. The proposed tool-chain is non-intrusive and incurs less than 0.1% runtime overhead with the NPB FT benchmark.
doi:10.1145/2642769.2642789 dblp:conf/pvm/BrownDM14 fatcat:7jni3k5ydva7pnapfp74f36yyi