The focus of this project is to develop tool support to provide the ability for scientific programmers to inquire about scalability problems and correlate this information back to source code. Furthermore, we believe that tools should be able to suggest and evaluate optimizing transformations to alleviate these problems. This would constitute a significant improvement over current performance analysis practice.
The key intellectual merit is in providing an automatic framework for detecting scalability problems and correlating them back to source code. We will experiment with our framework on the ASCI codes, which is intended to stress high-performance clusters.
The broader impact of this work is in three main areas. First, both PIs are working to create an interdisciplinary educational and research program. Second, students will be educated in high-performance computing. Finally, the proposed work allows for technology transfer to a wide arena of emerging fields, such as cluster computing as well as the established areas of SMPs and massively parallel computing. The developed framework and tools will be made generally available to the research community and high-performance computing labs.
"Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation."