Director’s Letter

Michael E. Papka directs the ALCF, a DOE Leadership Computing Facility providing computational resources, expertise, and training to the research community in support of major computational science campaigns.

Michael E. Papka

From the Office of the Director
Division Director
Michael Papka

The fully assembled Aurora is now generating data that has the potential to revolutionize many scientific fields. Having surpassed the exascale barrier this year and ranked among the fastest machines in the world for traditional computing tasks, Aurora is also unmatched as the world’s fastest artificial intelligence (AI) system for open science.

Aurora Early Science projects range from using machine learning to discover singlet fission materials to tackling biochemical challenges like biofuels to understanding extreme-scale cosmological hydrodynamics. All of these projects have seen significant gains on Aurora, with equally far-reaching expectations for discoveries in fields as diverse as materials science, drug discovery, and neutrino physics.

In other facility updates, with the retirement of Theta and ThetaGPU, ALCF decoupled and reconfigured ThetaGPU to become a new system named Sophia, which went into production in July with a focus on running JupyterHub instances and large language models. ALCF’s GPU-accelerated system, Polaris, continues to support projects from INCITE, ALCC, and other programs and serves as a development platform for merging ALCF resources with large-scale experimental facilities. Lastly, the ALCF’s AI Testbed of next-generation AI accelerator technologies continues to enable a range of exploratory research campaigns.

With its powerful computing resources and strong expertise in using AI in science, ALCF continues to grow its scientific user community. Building on a decade of hosting training workshops and various outreach activities to teach young researchers how to use Department of Energy (DOE) supercomputers, ALCF’s training program in the AI space has taken off—attracting over 600 participants in the past three years alone. This year, ALCF’s new Lighthouse Initiative to build a national network of university partnerships kicked off a pilot in collaboration with the University of Chicago and the University of Illinois Chicago, with plans to expand to more universities next year.

Along with Aurora and Polaris, the ALCF AI Testbed is enabling the facility to support major U.S. initiatives to expand the nation’s AI capabilities for pioneering science. These efforts include the DOE’s Frontiers in Artificial Intelligence for Science, Security and Technology (FASST) and the National Science Foundation-led National Artificial Intelligence Research Resource (NAIRR) Pilot launched this year in collaboration with DOE and other partners.

Meanwhile, Argonne’s Nexus effort is firmly underway to integrate supercomputers and experimental facilities on the scale of the DIII-D National Fusion Facility and CERN’s Large Hadron Collider. Nexus builds on established collaborations between ALCF and other DOE computing facilities to help shape DOE program plans for a future Integrated Research Infrastructure (IRI) and to coordinate and align Argonne’s IRI research and collaborations—a significant focus being ALCF’s coupling with the newly upgraded Advanced Photon Source (APS)—with DOE’s efforts in this area. The new tools deployed between ALCF and APS, along with the latest capabilities for on-demand computing and managing complex workflows developed by ALCF and its partners in other large experimental facilities, support the overall IRI program goals.