Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

METRION: A Framework for Accurate Software Energy Measurement

Benjamin Weigell, Simon Hornung, Bernhard Bauer

TL;DR

<3-5 sentence high-level summary> The paper tackles the challenge of accurately attributing energy consumption of software applications to underlying hardware components in complex CPU/memory architectures. It introduces the Energy Attribution Model (EAM) and the METRION framework, a platform-agnostic system that uses hardware performance counters and on-device sensors to distribute CPU and DRAM energy to per-thread activity, accounting for multi-socket NUMA topology, SMT, and frequency scaling. The authors implement an initial Linux/Intel extension and validate the approach across three workloads, achieving CPU energy attribution with about 4.2% MAPE for CPU-focused workloads and 16.1% MAPE for DRAM-focused workloads, while noting higher DRAM attribution errors in some cases and a notable runtime overhead. The work lays a foundation for cross-platform, fine-grained energy optimization and provides a path toward more accurate energy-aware software engineering and infrastructure management.

Abstract

The Information and Communication Technology sector accounted for approximately 1.4% of global greenhouse gas emissions and 4% of the world's electricity consumption in 2020, with both expected to rise. To reduce this environmental impact, optimization strategies are employed to reduce energy consumption at the IT infrastructure and application levels. However, effective optimization requires, firstly, the identification of major energy consumers and, secondly, the ability to quantify whether an optimization has achieved the intended energy savings. Accurate determination of application-level energy consumption is thus essential. Therefore, we introduce an energy attribution model that quantifies the energy consumption of applications on CPU and DRAM at the thread level, considering the influence of Simultaneous Multithreading, frequency scaling, multi-socket architectures, and Non-Uniform Memory Access. To ensure cross-platform applicability, we integrate the proposed model into an extensible framework, METRION, including a platform-independent data model and an initial implementation for Linux systems using Intel CPUs. We evaluate METRION across three different workloads and demonstrate that the energy attribution model can accurately capture the CPU energy consumption of applications targeting solely the CPU with a Mean Absolute Percentage Error of 4.2%, and the DRAM energy consumption of applications targeting DRAM with an 16.1% error.

METRION: A Framework for Accurate Software Energy Measurement

TL;DR

<3-5 sentence high-level summary> The paper tackles the challenge of accurately attributing energy consumption of software applications to underlying hardware components in complex CPU/memory architectures. It introduces the Energy Attribution Model (EAM) and the METRION framework, a platform-agnostic system that uses hardware performance counters and on-device sensors to distribute CPU and DRAM energy to per-thread activity, accounting for multi-socket NUMA topology, SMT, and frequency scaling. The authors implement an initial Linux/Intel extension and validate the approach across three workloads, achieving CPU energy attribution with about 4.2% MAPE for CPU-focused workloads and 16.1% MAPE for DRAM-focused workloads, while noting higher DRAM attribution errors in some cases and a notable runtime overhead. The work lays a foundation for cross-platform, fine-grained energy optimization and provides a path toward more accurate energy-aware software engineering and infrastructure management.

Abstract

The Information and Communication Technology sector accounted for approximately 1.4% of global greenhouse gas emissions and 4% of the world's electricity consumption in 2020, with both expected to rise. To reduce this environmental impact, optimization strategies are employed to reduce energy consumption at the IT infrastructure and application levels. However, effective optimization requires, firstly, the identification of major energy consumers and, secondly, the ability to quantify whether an optimization has achieved the intended energy savings. Accurate determination of application-level energy consumption is thus essential. Therefore, we introduce an energy attribution model that quantifies the energy consumption of applications on CPU and DRAM at the thread level, considering the influence of Simultaneous Multithreading, frequency scaling, multi-socket architectures, and Non-Uniform Memory Access. To ensure cross-platform applicability, we integrate the proposed model into an extensible framework, METRION, including a platform-independent data model and an initial implementation for Linux systems using Intel CPUs. We evaluate METRION across three different workloads and demonstrate that the energy attribution model can accurately capture the CPU energy consumption of applications targeting solely the CPU with a Mean Absolute Percentage Error of 4.2%, and the DRAM energy consumption of applications targeting DRAM with an 16.1% error.

Paper Structure

This paper contains 27 sections, 11 equations, 4 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Energy Attribution Model
  4. Design Considerations
  5. Idle and Active Power Consumption
  6. Multiple-Socket Architectures
  7. Thread-Level Execution and SMT
  8. Frequency Effects
  9. Model Foundations
  10. Components and their Active and Idle Energy
  11. Dividing Applications into Threads
  12. The Energy Attribution Model
  13. Active Energy Attribution
  14. CPU Work Function
  15. DRAM Work Function
  16. ...and 12 more sections

Figures (4)

  • Figure 1: Concurrent thread execution and execution interval splitting of two threads
  • Figure 2: The Platform Independent Data Model
  • Figure 3: Metrion Architecture and Dataflow
  • Figure 4: CPU package energy, DRAM energy, and runtime results, each determined across three workloads: CPU-, Memory-, and Combined-workload.