EnergyVis: Interactively Tracking and Exploring Energy Consumption for ML Models

Shaikh, Omar; Saad-Falcon, Jon; Wright, Austin P.; Das, Nilaksh; Freitas, Scott; Asensio, Omar Isaac; Chau, Duen Horng

doi:10.1145/3411763.3451780

Cited by 14 publications

(13 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, recent work has found that their results vary significantly and are not fully representative of the true emissions incurred by training [3]. Perhaps most similar to our work, EnergyVis [41] is an interactive tool for visualizing and comparing energy consumption of ML models as a function of hardware and physical location (U.S. state), given metadata about a model's energy use per epoch. Other studies have gone beyond simply tracking the emissions from training models, aiming to quantify the emissions resulting from manufacturing computing hardware [15], the broader impacts of sustainable AI [49], and the methodologies used to assess those impacts [21,26].…”

Section: Related Workmentioning

confidence: 61%

“…Improving the carbon transparency of research and practice. Despite the existence of tools such as Code Carbon [38] and EvergyVis [41], both carbon estimation and reporting in ML publications and technical reports remain a relatively rare phenomenon. Conferences such as NeurIPS and NAACL have recently added emissions reporting as an optional part of the submission process; however, more encouragement will be necessary for this to become commonplace.…”

Section: Future Directionsmentioning

confidence: 99%

See 1 more Smart Citation

Measuring the Carbon Intensity of AI in Cloud Instances

Dodge

Prewitt

Combes

et al. 2022

2022 ACM Conference on Fairness, Accountability, and Transparency

View full text Add to dashboard Cite

The advent of cloud computing has provided people around the world with unprecedented access to computational power and enabled rapid growth in technologies such as machine learning, the computational demands of which incur a high energy cost and a commensurate carbon footprint. As a result, recent scholarship has called for better estimates of the greenhouse gas impact of AI: data scientists today do not have easy or reliable access to measurements of this information, which precludes development of actionable tactics. We argue that cloud providers presenting information about software carbon intensity to users is a fundamental stepping stone towards minimizing emissions.In this paper, we provide a framework for measuring software carbon intensity, and propose to measure operational carbon emissions by using location-based and time-specific marginal emissions data per energy unit. We provide measurements of operational software carbon intensity for a set of modern models covering natural language processing and computer vision applications, and a wide range of model sizes, including pretraining of a 6.1 billion parameter language model. We then evaluate a suite of approaches for reducing emissions on the Microsoft Azure cloud compute platform: using cloud instances in different geographic regions, using cloud instances at different times of day, and dynamically pausing cloud instances when the marginal carbon intensity is above a certain threshold. We confirm previous results that the geographic region of the data center plays a significant role in the carbon intensity for a given cloud instance, and find that choosing an appropriate region can have the largest operational emissions reduction impact. We also present new results showing that the time of day has meaningful impact on operational software carbon intensity.Finally, we conclude with recommendations for how machine learning practitioners can use software carbon intensity information to reduce environmental impact.

show abstract

Section: Related Workmentioning

confidence: 61%

Section: Future Directionsmentioning

confidence: 99%

Measuring the Carbon Intensity of AI in Cloud Instances

Dodge

Prewitt

Combes

et al. 2022

2022 ACM Conference on Fairness, Accountability, and Transparency

View full text Add to dashboard Cite

show abstract

“…When DiSalvo et al [22] analysed the SHCI landscape in 2010, they classifed 45% of it as persuasive technology, for which "the standard approach is to design systems that attempt to convince users to behave in a more sustainable way". This is not echoed in our corpus, which contains only one study that uses an openly persuasive approach [77]; a few other studies have an indirect persuasive element as they encourage refection [15], support learning [19,20] or raise awareness [76]. What we see instead is a substantial amount of speculative design (e.g.…”

Section: Speculation Instead Of Prescriptionmentioning

confidence: 84%

“…What we see in our corpus is that energy has remained a key theme (e.g. [60,76,81,89,90]), although the research approaches and contexts have changed, as described above. A subcluster of energy-related research explores the carbon footprint and, more generally, the (un)sustainability of digital technology.…”

Section: Diverse Explorations Beyond Resource Consumptionmentioning

confidence: 99%

Have We Taken On Too Much?: A Critical Review of the Sustainable HCI Landscape

Bremer

Knowles

Friday

2022

CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

By CHI 2022, ffteen years will have passed since the emergence of Sustainable HCI (SHCI), which now constitutes an important subfeld of HCI. In this paper, we draw on two SHCI corpora to ask: Has SHCI progressed? How has the feld responded to prominent critiques? Have we identifed and adopted constructive strategies for impacting environmental unsustainability? We further show the wide array of competencies SHCI researchers have been called to develop, and how this has been refected in subsequent work. Our analysis identifes signifcant shifts in the SHCI landscape, toward research that is diverse and holistic, but also away from eforts to address the urgent climate crisis. We posit that SHCI has tended to take on far more than it could reasonably expect to deliver, and propose 'Green Policy informatics' as a pathway that enables SHCI to leverage a more traditional HCI skillset in addressing climate change. CCS CONCEPTS• Social and professional topics → Sustainability; • Humancentered computing → Human computer interaction (HCI).

show abstract

“…In [47], the authors propose a framework to investigate carbon emissions of end-to-end automatic speech recognition (ASR). EnergyVis [59] is a more general tool, which is capable of tracking energy consumption for various kinds of machine learning models and provides an interactive view to compare the consumption across different locations. Unfortunately, it is limited to the USA at the moment.…”

Section: Practical Suggestions: How To Quantify Environmental Impact?mentioning

confidence: 99%

Towards Green Automated Machine Learning: Status Quo and Future Directions

Tornede¹,

Tornede²,

Hanselle³

et al. 2021

Preprint

View full text Add to dashboard Cite

Automated machine learning (AutoML) strives for the automatic configuration of machine learning algorithms and their composition into an overall (software) solution -a machine learning pipeline -tailored to the learning task (dataset) at hand. Over the last decade, AutoML has become a hot research topic with hundreds of contributions. While AutoML offers many prospects, it is also known to be quite resource-intensive, which is one of its major points of criticism. The primary cause for a high resource consumption is that many approaches rely on the (costly) evaluation of many ML pipelines while searching for good candidates. This problem is amplified in the context of research on AutoML methods, due to large scale experiments conducted with many datasets and approaches, each of them being run with several repetitions to rule out random effects. In the spirit of recent work on Green AI, this paper is written in an attempt to raise the awareness of AutoML researchers for the problem and to elaborate on possible remedies. To this end, we identify four categories of actions the community may take towards more sustainable research on AutoML, namely approach design, benchmarking, research incentives, and transparency.

show abstract

EnergyVis: Interactively Tracking and Exploring Energy Consumption for ML Models

Cited by 14 publications

References 16 publications

Measuring the Carbon Intensity of AI in Cloud Instances

Measuring the Carbon Intensity of AI in Cloud Instances

Have We Taken On Too Much?: A Critical Review of the Sustainable HCI Landscape

Towards Green Automated Machine Learning: Status Quo and Future Directions

Contact Info

Product

Resources

About