This project will develop an understanding of common computational patterns in applications for scientific computing, and their efficient implementation on high performance computing systems. By implementing one or more benchmark codes in the Chapel language and performing a detailed evaluation of the performance, we will develop an understanding of how the programming model maps to the underlying hardware, and the implications of this mapping on computational performance.
- Demonstrate understanding of performance characteristics of high-performance scientific computing applications
- Perform rigorous performance evaluations of benchmark codes
- Demonstrate knowledge of previous work on parallel programming language implementation
- Effectively communicate research questions, methods and outcomes in oral, written and graphical forms to a broad audience