Category I: Expanse 2 - Continuing to Serve the Long Tail of Science and AI Innovation with New Technologies

The San Diego Supercomputer Center (SDSC) at the University of California, San Diego (UCSD) acquires and deploys Expanse2, a multi-Petaflop/s supercomputer system consisting of the latest processors and accelerators with an advanced Ethernet-based network to support data intensive workloads. This system responds to national priorities in research and development, with the goal of enhancing national competitiveness. It supports key research areas, including electronic chip design automation, advancing drug discovery with Artificial Intelligence (AI), application of AI in digital agriculture, study of instabilities and turbulence in fusion, and open inference models. As a national resource, Expanse2 replaces Expanse and provides advanced cyberinfrastructure (CI) to continue support for the Long Tail of Science, which reflects breadth in science, researchers, and their institutions. It will also enable Artificial Intelligence and Machine Learning (AI/ML) research, as well as data-intensive analysis at ever-increasing scales and complexity to process vast and growing data sets. The project centers on delivering capabilities that increase the capacity and performance for users of batch-oriented and science gateway computing, and that enable new research in AI applications and data analytics in all areas of science and engineering, which increasingly depend on heterogeneous, distributed, and integrated CI. Expanse2 includes innovations in system software, AI/ML frameworks, operations, and support that extend its capabilities far beyond the limits of the physical system. Expanse2 is an integral part of the national CI by offering user friendly allocation and scheduling policies, supporting rapid application development and data analysis via Jupyter notebooks, providing extensive AI/ML frameworks and libraries, connecting with federated and extensible data ecosystems, and providing user support focused on delivering the tools, training, and support needed to increase productivity and decrease time-to-science results. Expanse2 features an innovative storage system that includes an all-NVMe file system ideal for AI/ML workloads, integrated with a large capacity storage system, which provides fault tolerance in terms of system availability and data integrity. Publicly accessible network interfaces on compute nodes facilitate integration with distributed data and compute sources, such as public cloud, Open Science Grid services (OSDF and OSPool), and campus systems. Expanse2 is deployed in SDSC's energy-efficient data center and connected to multiple high performance research and education networks at 400 Gbps, serving thousands of users who require high performance yet modest-scale resources. Allocation and usage policies, honed from years of experience at SDSC, are tailored to achieve fast turnaround and responsiveness. A team of experts in computational science, AI/ML, data-intensive computing, scientific workflows, and large-scale systems operations supports Expanse2 at the highest levels of utilization, reliability, and usability. The strong training and education components of the Expanse2 project are focused on domestic workforce development from universities, teaching colleges, and high school students. The system’s advanced compute and GPU architecture with high-performance storage are ideal for use by industry to foster public-private partnerships, especially startups from AI, advanced manufacturing, pharmaceutical, and medical device companies. Knowledge gained through the project leads to improvements in algorithms, software, and systems management tools, as well as a better understanding of how integrated CI can be configured to support emerging research challenges. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2614012 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Michael Norman | Institution: University of California-San Diego, LA JOLLA, CA | Award Amount: $10,000,000 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2614012 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2614012.html