Frugal Computing -- On the need for low-carbon and sustainable computing and the path towards zero-carbon computing (2303.06642v1)

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.

An Examination of Frugal Computing and Low-Carbon Strategies in Computing Sciences

The paper "Frugal Computing: On the need for low-carbon and sustainable computing and the path towards zero-carbon computing" by Wim Vanderbauwhede presents a compelling argument for a paradigm shift in computational practices, focusing on sustainability and the reduction of carbon emissions. The paper is anchored on a pressing issue: computing emissions already exceed those from the airline industry and are estimated to drastically increase. By 2040, emissions from computing could consume more than half of the emissions budget required to limit global warming to below 1.5°C.

Vanderbauwhede identifies the unsustainable nature of current trends in computing, where the production emissions of computing devices surpass those from their operation. This crucial insight implies that increased energy efficiency alone will not suffice if device production continues at current rates. Therefore, there is a vital need to increase the lifespan of computing devices and approach computational resources as finite entities.

Finite Computational Resources

The paper discusses the historical reliance on Moore's Law, which predicted the exponential increase of computing power with consistent decreases in relative cost and size over time. Today's technological constraints and climate imperatives halt this trend, compelling a shift towards "frugal computing." This strategy involves judicious utilization of computational resources, optimized for energy efficiency with better hardware-software integration, and a focus on extending device life.

The economic and practical implications are significant. Historically, the hardware-software development economics assumed performance doubles with each generation without incremental costs. This assumption no longer holds true, requiring a reassessment of strategies in both research and practice.

The Scale of the Emissions Challenge

The paper forecasts a dramatic rise in energy consumption and carbon emissions from computing. Based on data from institutions like the Semiconductor Industry Association and the International Energy Agency, the paper underscores that emissions from computational resources could grow to comprise about 30% of the total electricity consumption by 2040. The research discusses emissions both from operation and production, which collectively could dominate the acceptable global emissions budget if unchecked.

Supporting data points presented include statistics on device sales and energy consumption forecasts for various device categories. The work stresses that extending device lifespans and reducing operational energy use are essential strategies. From 3,000 TWh of current global energy consumption by communication and computing technologies, projections show a potential rise—demonstrating a critical call for action.

Towards a Zero-Carbon Computing Future

Vanderbauwhede envisions a future where computational demands are sustainably met, emphasizing the need for radical changes in how computational technologies are developed and deployed. Efforts to progress towards zero-carbon computing may involve:

• Increasing device longevity through infrastructure for repairs, availability of replacement parts, and training.
• Adopting new economic incentives and policies to adjust consumer and manufacturer behaviors.
• Developing next-generation technologies that prioritize energy efficiency over frequent hardware upgrades.

Research Challenges and Directions

The initiative for sustainable computing faces numerous research challenges across multiple domains:

1. Cloud and Data Centers: Addressing energy consumption through dynamic voltage frequency scaling (DVFS), optimized scheduling, and improved energy-efficient software.
2. Networking: Incorporating energy consumption as a quality-of-service metric, thereby optimizing network infrastructures for reduced energy use.
3. IoT and Mobile Devices: Extending device lifespans and emphasizing energy-efficient design practices through better software engineering and maintenance strategies.
4. Human-Computer Interaction: Encouraging user awareness and behaviors that reduce energy consumption, thereby integrating user-centric designs.

Research directions include energy-aware programming paradigms, algorithm optimization for minimal energy usage, and sustainable systems design. These directions emphasize the necessity for comprehensive, data-driven approaches to resource allocation and enduring systems.

Conclusion

The paper advocates for a transformative approach in computing science to address climate change, emphasizing radical optimization of both energy efficiency and device longevity. The call for frugal computing invites computing scientists and engineers to rethink existing paradigms, exploring solutions that are viable within the confines of current capability development. The proposal challenges the field to rethink foundational assumptions and embrace the imperative of sustainable practices within the next two decades.