A survey of 300+ HPC users reveals that fewer than 30% are aware of their energy consumption. This research introduces Energy-Based and Carbon-Based Accounting models that tie computing costs to environmental impact, demonstrating through real experiments and simulations that such incentives can dramatically reduce energy use while maintaining scientific productivity.

We’re excited to shine a spotlight on the recently refreshed hardware discovery browser on Chameleon! Launched a little while ago, this upgrade makes it easier and faster to pinpoint the exact hardware you need for your experiments, connecting you directly to real-time scheduling and reservations.

As you most likely know, Chameleon offers dedicated, reservable resources. That means your work gets exclusive access to the nodes you choose but navigating hardware specs and reservation calendars together can sometimes be a bit of frustration.

The new resource discovery experience helps you link “what hardware am I getting?” with “when can I actually run?” into a more streamlined workflow than previously.

This past month brings 2025 to a close and also exciting updates, including a year in review, Trovi improvements, and a CHI@Edge artifact, along with announcements about upcoming webinars (next up: How to Organize Artifact Evaluations (AEs) with Shared Infrastructure on Jan. 13th), our call for presentations for our sixth user meeting at NCAR’s campus in Boulder, and highlights from our newly appointed reproducibility ambassadors.

Modern networks are growing increasingly complex, and managing them effectively requires sophisticated policy configurations that can adapt in real-time. Languages that enable direct programmability of network data planes, like P4, offer unprecedented control over packet processing, but come at a steep learning cost. This raises a crucial question: can we automate the generation of these complex network policies while ensuring they're correct, efficient, and adaptable? NetPrompt is a novel framework that harnesses the power of Large Language Models (LLMs) to automatically generate and optimize P4 scripts for Software-Defined Networks (SDN). Read more to learn how the research team implement NetPrompt using Chameleon.

As we approach the holiday season, we’d like to thank you once again for being a part of our wonderful Chameleon community!

Please keep in mind the following adjustments in our schedule over the next few weeks:

• The Chameleon Help Desk will be closed from Wednesday, December 24 until January 5th.

• As tradition, between December 19 to the 23rd, we’ll permit you to make 14-day leases for nodes and VMs that typically have 7-day limits.

It's the end of the year, so that means it's time for our annual Tickets of the Year blog! This blog covers some of our most common help desk tickets and provides solutions, so it can hopefully be a helpful reference if you run into any of these problems. If you need more help, the best places to look would be the docs, the FAQ, or submitting a new ticket to the help desk.

The Sixth Chameleon User Meeting will be held April 15-16, 2026 at the NCAR Mesa Lab in Boulder, Colorado. This year's meeting focuses on computer science research and education in Artificial Intelligence (AI) and Machine Learning (ML). The objective of the meeting is to create a community discussion on AI research, approaches to education, and most importantly what platform you need – or may need in the future – to solve the hard problems and train the workforce of the future. We invite researchers and educators using Chameleon to submit presentation proposals sharing their experiences. Come to network with scientists working on similar topics, share tips on how to muster resources and data for hard-to-get experiments, and find materials and digital artifacts to teach AI classes!

As in previous years, we will reimburse travel expenses of up to $1,500 for the presenting authors of the top 10 selected abstracts (one author per abstract).

Prof. Fraida Fund's (NYU) recent webinar presented her extensive open source materials that she developed and used to teach a large (190+ students) graduate-level course on machine learning operations (MLOps) -- not just how to develop an ML model and code in isolation, but also how to design and implement a full-fledged system to develop and deploy ML models in various environments. The webinar was a great success: we had 100+ registrations, close to 50 participants at the webinar, and extensive Q&A following the presentation. We are excited to announce that the recording of her presentation, along with materials and slides, are available on the FOUNT website.

Watch the presentation and view the materials she shared from her webinar. Let Marc Richardson (mtrichardson@uchicago.edu) know if you'd be interested in hosting similar webinars like hers in the future.

This month brings exciting updates including enhanced hardware browser integration, CHI@Edge stability improvements, and expanded support for distributed training on AMD GPUs, along with announcements about upcoming webinars and our sixth user meeting at NCAR’s campus in Boulder.

Trovi helps you package and share computational artifacts that run on Chameleon, from Jupyter notebooks to complete experimental workflows. Whether you're importing code from GitHub, organizing artifacts into Collections, collaborating with co-authors, or publishing work to get a citable DOI, these tips will help you make the most of Trovi for your research and teaching.