PERFACCT

GPU-Computing Energy Monitoring

Continued development of the open-source EMA framework for energy measurement on HPC systems – including a plugin architecture, CLI utilities, and Docker-based development infrastructure.

Client/Company/Industry

PERFACCT GmbH

Duration

ongoing

Product

Open-Source Software

Expertise

Software Development

Goal

EMA is an open-source framework for measuring and monitoring energy consumption on HPC systems. The goal of the project was to extend and stabilise the framework: new core APIs, a dynamic plugin loading mechanism, and supporting tooling were to make EMA more robust and ready for broader deployment on real cluster systems.

Tasks

Extension and improvement of the framework's core APIs
Implementation of a dynamic plugin loading mechanism
Design and development of CLI utilities and supporting tools
Development of core plugins using NVML, MQTT, and RAPL (Linux powercap)
Extension and maintenance of the CMake build system
Implementation and extension of unit tests for core plugins
Development of a Docker image set for the core framework and associated tools

Challenges

Implementing a dynamic plugin loading mechanism required deep familiarity with the C runtime environment and careful interface design — plugins needed to load and unload stably without destabilising the core framework.

Integrating heterogeneous energy sources — GPU consumption via NVML, CPU packages via RAPL, and network communication via MQTT — into a unified plugin architecture placed high demands on abstraction and API consistency.

Programming Languages

C, Shell

Technologies

NVML, MQTT, Docker, CMake, Linux powercap

Architecture overview of the EMA framework showing the plugin system and energy measurement sources.

Takeaway

EMA was developed into an extensible, plugin-based framework that standardises energy measurement across modern HPC systems. The containerised development infrastructure and expanded test coverage provide a solid foundation for continued open-source development.

Similar Projects

RIM2D - Highly Efficient 2D Hydraulic Simulation of Fluvial, Pluvial, and Urban Flooding

Hydrodynamic Simulation Web Application Geodata GPU Computing

RIM2D is an existing, highly efficient 2D hydraulic simulation model for fluvial, pluvial, and urban flooding. As part of a strategic partnership, we supported the extension of the research code with a web application and a cloud-based GPU simulation environment, enabling its transition into a market-ready product.

See project

Data Lake for Geoscience Research Data Management

Datalake Research Data Management Geodata Cloud-Native Open-Source

We developed an Open-Source S3-based data lake solution for the centralized ingestion, categorization, and searchability of data. The goal was to automate and improve manual data management through an integrated architecture with workflow orchestration, data cataloging, and access control.

See project

All Projects