Overview
The PAOT group concentrates on issues relating to performance improvements. Over the years, the group developed algorithmic improvements for better utilization of architectural resources. This technology is embodied in several contexts, such as in stand-alone post-compilation tools and performance visualization tools. Our work has resulted in significant performance improvements to the executable binary files of many important applications and systems, measured on benchmarks such as SPEC and TPC-C.
The techniques we have been developing apply to a wide variety of target machines, both instruction level parallel (e.g., superscalar) processors, as well as other machine architectures. These techniques also pertain to areas such as parallel programming and performance debugging. To co-op with performance improvements, there is a need to better understand the origins of performance bottlenecks. We developed a visualization tool that offers a deeper understanding of the performance impact of programs compiled for the Power architecture.
Manager
Michael Vinov,
