About this page

Energy-efficient software draws on a wide body of work: web and software standards, open measurement tools, guidance for particular platforms, and research on where the energy goes and how much can be saved. This page collects the resources we find most useful and that have informed GreenCode's work. GreenCode's own papers and deliverables are on the publications page, and the case for acting on software's energy use is set out in the Climate Case for GreenCode.

GreenCode project materials

The project's own printed material, free to download and share.

GreenCode project profile leaflet

ITEA's two-page profile of GreenCode: the challenge, the proposed solution, the expected results and the partners across seven countries.

PDF · 2 pages · 1.1 MB

GreenCode poster, ITEA PO Days 2026

The A0 project poster made for the ITEA PO Days 2026 in Istanbul: summary, key results, the energy optimisation process and the consortium.

PDF · 1 page · 1.8 MB

Standards and guidelines

• W3C Web Sustainability Guidelines: 93 guidelines across user experience design, web development, hosting and infrastructure, and business strategy.
• ISO/IEC 21031:2024, Software Carbon Intensity: the international standard for calculating the rate of carbon emissions of a software system.
• Software Carbon Intensity specification and SCI Guide: the Green Software Foundation's methodology, SCI = ((E × I) + M) per R, and practical techniques for reducing a score.
• Green Software Patterns catalogue: cloud, web and AI patterns for reducing carbon, from demand shaping to scaling to zero.
• Green Software Foundation learning material: the principles of green software, including energy efficiency, energy proportionality and carbon awareness.
• CISQ Automated Source Code Green Measure: code-level weaknesses that waste resources, detectable by static analysis.
• WCAG 2.2: accessibility practice, much of which also lightens pages and extends the life of older devices.
• ITU Toolkit on Environmental Sustainability for the ICT Sector and ITU-T Study Group 5: data centre metrics (PUE, WUE, CUE), server utilisation and greenhouse gas scopes for ICT.
• ETSI EN 303 804, SNIA Emerald and The Green Grid: energy efficiency measurement for data storage and data centres.

Measuring energy and carbon

• Sustainable Web Design model and CO2.js: the Green Web Foundation's model for estimating emissions from data transfer, and the library that implements it. This site uses it to show the footprint of each page view.
• Greencheck and carbon.txt: checking whether a host runs on renewable energy, and publishing the evidence in a machine-readable form.
• Grid-aware websites: adapting what a service does to the carbon intensity of the electricity grid at the time.
• HTTP Archive Web Almanac: page weight, resource breakdown and emissions across the web, the baseline behind web carbon ratings.
• CodeCarbon: energy and emissions of Python and machine learning workloads, using regional grid intensity.
• Scaphandre and Kepler: per-process, per-container and per-pod power monitoring for servers and Kubernetes.
• Creedengo (formerly ecoCode): static analysis rules that flag energy code smells in Java, Python, PHP, JavaScript, C# and Rust.
• Eco-CI and GreenFrame: energy measurement in CI pipelines, and carbon budgets for web applications enforced in CI.
• Slurm energy accounting, LIKWID and the Green500: energy per job, per component and per flop in high-performance computing.
• cloc: counting lines of code, a starting point for sizing an optimisation.

Green AI

• AI environmental impact report (Green Web Foundation): what developers can and cannot control about the footprint of AI.
• How much energy does ChatGPT use? (Epoch AI): per-query energy estimates for current models.
• Small changes can reduce AI energy use by 90% (UNESCO): model choice and prompt efficiency.
• AI Energy Score (Hugging Face): comparable energy ratings for models across common tasks.
• Carbon emissions and large neural network training (Patterson et al., 2021): how data centre, processor and model architecture combine.
• Energy and policy considerations for deep learning in NLP (Strubell et al., 2019): the paper that put the cost of training language models on the agenda.
• Towards sustainable NLP (Poddar et al., 2025): benchmarking the inference energy of large language models.
• DynamoLLM (Stojkovic et al., 2025): LLM inference clusters designed for performance and energy efficiency.
• The rise of small language models (Zhang et al., 2025).

Guidance for particular platforms

• Web and streaming: the Greening of Streaming industry body and its low-energy streaming work; the IEA's data centres and networks statistics; Netflix's climate commitment and its per-hour streaming emissions methodology.
• Games and interactive applications: the Sustainable Games Alliance environmental standard and its energy-saving resources for Unreal, Unity and Godot.
• Mobile: Apple's Energy Efficiency Guide for iOS apps; Android's Doze and App Standby; and research on dark mode and OLED power.
• Embedded and IoT: low-power modes in the Espressif and Nordic Semiconductor documentation, Zephyr power management and FreeRTOS tickless idle.
• Mainframes: IBM Enterprise COBOL compiler and performance tuning, and partition-level power monitoring through the HMC API.
• Cloud and infrastructure: the Azure Well-Architected sustainability guidance, Google Cloud sustainability and OpenStack Watcher; Docker build best practices and KEDA for right-sizing and scaling to zero; Open Policy Agent, Checkov and Sentinel for policy as code.
• Architecture: the environmental impact of common architecture patterns (Kong) and Microsoft's sustainable software blog.
• Databases and storage: Environmental footprints of query processing (Bachras and Jacobsen, PVLDB 2025) and A case for ecological efficiency in database server lifecycles (Bodner et al., CIDR 2025).
• Research software: the Software Sustainability Institute's Green DiSC programme, and the GREENER principles for environmentally sustainable computational science (Lannelongue et al., 2023).

Research on software and energy

• The real climate and transformative impact of ICT (Freitag et al., Patterns, 2021): a critique of estimates, trends and regulations, and of rebound effects.
• Black software: the energy unsustainability of software systems in the 21st century (Manner, Oxford Open Energy, 2022).
• Energy efficiency across programming languages (Pereira et al., 2017): how energy, time and memory relate across 27 languages.
• The case for energy-proportional computing (Barroso and Hölzle, 2007).
• Green IT and green software (Verdecchia et al., IEEE Software, 2021) and Understanding green software development (Ardito et al., 2015).
• Software development lifecycle for energy efficiency (Georgiou et al., ACM Computing Surveys, 2019).
• Green Software Measurement Model (Guldner et al., 2024) and Criteria for sustainable software products (Guldner et al., 2021).
• Incorporating energy efficiency measurement into CI/CD (Kruglov et al., 2021).
• RAPL in action (Khan et al., 2018) and Measuring energy and power with PAPI (Weaver et al., 2012): measuring power on real hardware.
• Network energy use not directly proportional to data volume (Mytton et al., 2024) and Electricity intensity of internet data transmission (Aslan et al., 2017).
• The greenhouse gas footprint and abatement potential of ICT (Bieser et al., 2023).
• Mining energy-greedy API usage patterns in Android apps (Linares-Vásquez et al., 2014).
• EcoFaaS: serverless environments designed for energy efficiency (Stojkovic et al., 2024).
• Data centers on wheels: emissions from computing onboard autonomous vehicles (Sudhakar et al., 2023).
• The quality of source code generated by AI engines (Tosi, 2024).
• Refactoring and energy: Tales from the Code #2 (2022), refactoring techniques across object-oriented languages (2019), refactoring in embedded computing (2016), refactoring for portable devices (2022) and perspectives on refactoring planning and practice (2015).

Reports and data

• Electricity 2024 (International Energy Agency): data centre electricity use and the forecast to 2026.
• The role of artificial intelligence in the European Green Deal (European Parliament study).
• How to stop data centres from gobbling up the world's electricity (Nature, 2018).
• Lean ICT: towards digital sobriety (The Shift Project, 2019).
• On global electricity usage of communication technology: trends to 2030 (Andrae and Edler, 2015).
• Green software insights (CAST): the state of green software in enterprise codebases.
• Coding on Copilot (GitClear, 2024): what AI coding assistants are doing to code quality.
• Electricity grid resilience: climate change is expected to have far-reaching effects (US Government Accountability Office, 2021).

The wider picture

• Show Your Stripes: Ed Hawkins' warming stripes, a year-by-year picture of how temperatures have changed where you live and across the world.
• UN Sustainable Development Goals: the goals the case for greener software supports, from climate action to affordable and clean energy.