This month we’ve again been focusing on improvements to the Chameleon backend, but we still are excited to show our CHI@Edge peripheral guide and new CHI-in-a-box features. We also are preparing for an OpenStack upgrade on July 1st.
In this blog post, we explore groundbreaking research on optimizing production ML inference systems to achieve high accuracy while minimizing costs. A collaboration between researchers from multiple institutions has resulted in the development of three adaptive systems - InfAdapter, IPA, and Sponge - that tackle the accuracy-cost trade-off in complex, real-world ML scenarios. Learn how these solutions, implemented on the Chameleon testbed, are pushing the boundaries of cost-effective ML inference and enabling more accessible and scalable ML deployment.
Discover the impact you can make in scientific research as a Senior Research Software Engineer on the Nimbus team. Join us to innovate and transform science on the Chameleon platform!
Recognizing the need for easier modification of running containers, we’ve introduced a straightforward method for SSHing into CHI@Edge containers, enhancing your development and experimentation processes. Our new tutorial on Trovi helps you create an SSH-ready base container, extending the CHI@Edge “hello world” example to enable remote access. This allows you to troubleshoot, configure, and update containers as if they were real servers. Access the tutorial via the CHI@Edge dashboard, and start developing with enhanced flexibility and ease.
Many folks in our community (now over 10,000 strong!) use our testbed for projects with student interns, REU sites, and summer courses. In Summer 2022, we started a series of introductory tutorials to help students quickly learn the ropes and hit the ground running on their summer projects. While Chameleon has extensive documentation, webinars, and blogs demonstrating our features and latest updates, these tutorials add extra structure, interaction, and context for those eager to harness the power of Chameleon.
This month, we’ve been mostly focusing on improving the Chameleon backend, and don't have any exciting new features for you. We still found some time for a few things that hopefully will make experimentation easier, including an upcoming reproducibility workshop, an updated Chameleon tutorial video, and two new Trovi artifacts: for orchestrating instances and using SSH on CHI@Edge.
"Connecting SLICES-RI and Chameleon: An Approach towards Portable, Reproducible Experiments" explores the creation of replicable scientific experiments through 'pos', a novel tool and methodology. Integrated with Chameleon and other public research infrastructure, 'pos' allows researchers to replicate experiments reliably across shared infrastructures. This collaboration aligns with the SLICES-RI initiative to enhance research portability. Authors Henning Stubbe, Sebastian Gallenmüller, and Georg Carle also share insights into maintaining adaptability and variety in long-term research projects, crucial for advancing experimental reproducibility and collaboration.
In 2021, we introduced CHI@Edge, a cutting-edge testbed tailored to the dynamic needs of edge computing and IoT research. Our initial offerings included a robust selection of devices like Raspberry Pi 4s, Nvidia Jetson Nanos, and Jetson Xaviers, tailored for high-performance yet low-power edge computing tasks. Today, we're thrilled to announce the transition of CHI@Edge from a widely embraced preview to its official full production phase.
As the digital age accelerates, the importance of reproducibility in computational science has never been more critical. This year, we are thrilled to announce a workshop focusing on the theme of "Practical Reproducibility" in Computer Science. Scheduled for November 18, 2024, at Terminus 330 in Atlanta, this full-day workshop promises to be a milestone event in advancing the science of reproducibility.