Principal Investigator: John Mellor-Crummey, Rice University
This project aims to develop HPCToolkit, which is an integrated suite of tools for measurement and analysis of program performance on systems ranging from desktops to supercomputers. By using statistical sampling of timers and hardware performance counters, HPCToolkit collects accurate measurements of a program’s work, resource consumption, and inefficiency. It attributes metrics to the full calling context in which they occur. To adapt HPCToolkit on exascale platforms, this project will extend HPCToolkit with improved capabilities for measurement and analysis of computation, data movement, communication, and I/O at extreme scale. Enhancements planned include improved support for lightweight measurement of massive node-level parallelism using hardware counters, measuring OS activity, recording measurement data efficiently, analyzing performance data in parallel, and integrating code-centric, time-centric, data-centric, and resource-centric views for analysis.