[Paper Review] Frugal Computing -- On the need for low-carbon and sustainable computing and the path towards zero-carbon computing

Wim Vanderbauwhede|arXiv (Cornell University)|Mar 12, 2023

ICT in Developing Communities4 citations

TL;DR

The paper proposes 'frugal computing'—a paradigm shift to drastically reduce energy and material use in computing by extending device lifespans and optimizing efficiency across all layers of the computing stack. It argues that without radical improvements in energy efficiency and hardware longevity, computing emissions will exceed 1.5°C climate targets by 2040, making frugal computing essential for sustainable digital transformation.

ABSTRACT

The current emissions from computing are almost 4% of the world total. This is already more than emissions from the airline industry and are projected to rise steeply over the next two decades. By 2040 emissions from computing alone will account for more than half of the emissions budget to keep global warming below 1.5$^\circ$C. Consequently, this growth in computing emissions is unsustainable. The emissions from production of computing devices exceed the emissions from operating them, so even if devices are more energy efficient producing more of them will make the emissions problem worse. Therefore we must extend the useful life of our computing devices. As a society we need to start treating computational resources as finite and precious, to be utilised only when necessary, and as effectively as possible. We need frugal computing: achieving our aims with less energy and material.

Motivation & Objective

• Address the unsustainable rise in computing emissions, which are projected to exceed 50% of the 1.5°C global carbon budget by 2040.
• Counter the misconception that energy efficiency alone can offset growing demand, emphasizing that production emissions now exceed operational emissions.
• Reframe computational resources as finite and valuable, urging society to treat them as scarce rather than abundant.
• Propose a systemic shift toward 'frugal computing'—achieving more with less energy and material through optimized design and extended device lifetimes.
• Drive a transformation in computing science, industry, and policy to prioritize energy efficiency and sustainability as core design principles.

Proposed method

• Define computational resources as encompassing all energy and material inputs across the entire stack—from end devices to networks and data centers.
• Advocate for extending device lifespans as a primary strategy, arguing that longer hardware lifecycles reduce total emissions more effectively than incremental efficiency gains.
• Introduce 'frugal computing' as a design philosophy focused on minimizing energy and material use across the full lifecycle of computing systems.
• Propose integrating energy and carbon cost awareness into all layers of computing systems, including compilers, programming languages, and operating systems.
• Emphasize the role of data-driven optimization using system telemetry to enable degradation-aware operation, predictive maintenance, and energy-aware scheduling.
• Highlight the importance of human-computer interaction (HCI) in nudging users toward energy-efficient behaviors and increasing device longevity.

Experimental results

Research questions

• RQ1 How can computing systems be redesigned to minimize total energy and material use over their entire lifecycle, especially given that production emissions now exceed operational emissions?
• RQ2 What systemic changes in software engineering, hardware design, and business models are required to extend the useful life of computing devices beyond current short lifespans?
• RQ3 To what extent can energy efficiency improvements in software, compilers, and system architecture compensate for the projected growth in computational demand?
• RQ4 How can user behavior be influenced through interface design and feedback mechanisms to reduce the carbon footprint of digital activities?
• RQ5 What role can formal methods, such as energy-aware programming languages and compilers, play in minimizing energy consumption across all system components?

Key findings

• Computing currently accounts for nearly 4% of global carbon emissions—more than the aviation industry—and is projected to exceed 50% of the 1.5°C carbon budget by 2040 if current trends continue.
• The production of computing devices generates more emissions than their operation, making device longevity a more impactful lever for emissions reduction than incremental energy efficiency.
• Moore’s Law has effectively ended due to physical scaling limits, slowing performance-per-watt improvements and making energy efficiency gains insufficient to offset rising demand.
• Market dynamics alone will not prevent energy demand from rising; without proactive systems-level optimization, computing energy use will grow unchecked.
• Extending device lifespans through better software support, repair infrastructure, and long-term part availability is critical to reducing emissions and achieving zero-carbon computing.
• A vision of zero-carbon computing is achievable through sustained improvements in energy efficiency, longer device lifecycles, and systems-aware design—where performance increases without proportional energy cost.