A Source Code Analyzer for Performance Prediction

Performance prediction is necessary and crucial in order to deal with multi-dimensional performance effects on par-allel systems. The increasing use of parallel supercomput-ers and cluster systems to solve large-scale scientific prob-lems has generated a need for tools that can predict scala-bility trends of applications written for these machines. In this paper, we describe a compiler tool to automate perfor-mance prediction for execution times of parallel programs by runtime formulas in closed form. For an arbitrary paral-lel MPI source program the tool generates a corresponding runtime function modeling the CPU execution time and the message passing overhead. The environment is proposed to support the development process and the performance en-gineering activities that accompany the whole software life cycle. The performance prediction tool is shown to be ef-fective in analyzing a representative application for varying problem sizes on several platforms using different numbers of processors.