HPC Application Energy Efficiency (2022 - Present)

Jul 8, 2022 · 2 min read

Analytics and AI Methods at Scale Group (AAIMS), Oak Ridge National Laboratory, USA
Research Staff, July 2022 ~ Present

Overview

This initiative increases energy awareness by providing tools and frameworks for scientific users.

Analyzed and reported energy projections given the system hardware and system software capabilities
Developed a prototype on-demand job-level performance and energy monitoring tool for user-driven analysis.
Collaborated with the vendor (AMD) to enhance energy monitoring and workflow integration.
Led a division-wide collaboration to engage end users and bridge system operations with energy efficiency.
Served as a subject matter expert, driving collaboration through presentations and publications shaping energy efficiency in leadership computing.

Python based BYOM (bring your own monitoring) concept HPC monitoring tools for users
AMD OmniStat, AMD MI250x power & energy sensors and tools
Python based UI backends into user driven HPC power & energy data viewer

Last updated on Jul 8, 2022