Project Group: HPC Performance Almanac (HPC-PA)

The goal of this project group is to build a develop a trusted database of performance data for common operations in HPC applications (e.g., numerical operations, communication, I/O). The performance characterization data for each of these operations will be determined with empirical measurement in a continuous integration pipeline. To make the data easily accessible for researchers and practitioners, all data will be made available through a web application that supports:

• visualization of data with plots
• selection and configuration of data
• download/embedding of visualizations, structured data, and raw data
• download of benchmarking code

All code and data will be released under an open source/data license.

For realizing this project, we are looking for students with a broad selection of skills and interests, in particular:

• webapp frontend and backend development
• data visualization and analysis 
• continuous integration toolchain for performance measurements
• parallel computing and code optimization
• scientific writing

You find more details in the slides that have been presented at the project group presentation event,