The San Diego Supercomputer Center (SDSC) located at UC San Diego has answered the National Science Foundation’s call for a cyber infrastructure ecosystem that meets the needs of today’s data-intensive science. Offered as a prototype national research platform (PNR), the innovative all-in-one system (computing resources, research and education networks, advanced computing devices and other instruments) is a test bed for pilots scientists as diverse as the platform itself to accelerate science and enable transformative discoveries.
For this one-of-a-kind resource, NSF provided SDSC $ 5 million over five years, with matching funding for the operation of the systems. The award will support hardware and deployment at three facilities: on the east coast at the Massachusetts Green High Performance Computing Center (MGHPCC) in Mount Holyoke, MA; in the Midwest at the University of Nebraska-Lincoln (UNL) and on the West Coast at SDSC, as well as five data caches in the Internet2 backbone of the network.
“Supercomputing continues to be essential for researchers in an ever-growing number of scientific fields at institutions across the United States,” said UC San Diego Chancellor Pradeep K. Khosla. “The SDSC is nationally recognized for its innovations in cyber infrastructure and in building high performance computer systems specifically designed for scientific research. The NSF sought a national innovator to help improve the crucial resources needed to further enhance America’s prowess in scientific discovery. “
According to Frank Würthwein, Acting Director of the SDSC and Principal Investigator (PI) on the NRP project, the funding will allow him and the co-PIs of the SDSC, UNL and MGHPCC to work with the research community. to explore the experimental architecture of the NRP.
“The NRP provides resources and capacities for various sciences, as well as the expertise of systems specialists and the user community,” said Würthwein. “It is an open system designed for growth and inclusion, a way for academic institutions to join a national system and, through their participation, to expand and enrich the HPC ecosystem. “
Key components of the NRP architecture include research and education networks, compute accelerators, heterogeneous computing resources (e.g. advanced computing close to the location of the instrument), a network content delivery to provide easy access to all data, anytime, anywhere, and a software stack that enables centralized system management, while remaining open to growth. The innovative technical features of the HPC subsystem include a blend of Field Programmable Gate Array (FPGA) chips, graphics processing units (GPUs) with memory and storage in a fully integrated ultra-low latency fabric from GigaIO.
“The selection of our dynamic fabric FabreX for the composable architecture of this new research platform aligns with our mission to provide the agility and infrastructure performance necessary for next-generation scientific discovery,” said GigaIO CEO Alan Benjamin.
The wider impact of the PNR focuses on four themes: 1) empowering under-represented and / or under-represented researchers by making them “co-owners” of the PNR through the “Bring Your Own Resource (BYOR) / Bring Your Own” program. Device (BYOD) ”; 2) the well-being of society, focusing on health and disaster response; 3) STEM education, workforce development and awareness; and 4) improving industrial competitiveness.
“Any campus in the continental United States can be accessed from some of the NRP hardware within about 500 miles,” said Thomas De Fanti, professor emeritus of computer science at the University of Illinois in Chicago and researcher at the Qualcomm Institute at UC San Diego.
UNL will lead NRP’s infrastructure operations using Kubernetes to manage remote systems at the three main sites and Internet2, as well as BYOR sites.
“Kubernetes is a new interface and a new model of operations for many computer scientists. We are excited to expand the community’s experience with Kubernetes while providing cyber infrastructure resources to scientists, ”said Derek Weitzel, assistant research professor of computer science at the Holland Computing Center at UNL.
The system deployment will take place at SDSC on the UC San Diego campus, where several science pilots from astrophysics, molecular dynamics, health sciences and more will participate. For example, National Center for Microscopy and Imaging Research (NCMIR) Director Mark Ellisman will examine how NRP can enhance NCMIR’s work with NSF and other agencies to create multi-scale wiring models of the brain and components. molecules that make brain cells. the most energy efficient information mediators known.
“Success on this [project] will allow greater alignment with other key NSF initiatives, such as NeuroNex2, ”noted Ellisman, referring to the technology-based and team-based neuroscience project, for which he is the group leader.
The PNR will also support research on the science of forest fires. Ilkay Altintas, head of data science at SDSC, for example, works with researchers to create tools and techniques like WIFIRE for firefighters. The PNR will provide urgent IT resources in support of fire perimeter forecasts that can be used in the field by fire managers to identify where to most effectively deploy firefighting resources to achieve containment of fire. ‘fire.
Other examples of PNR’s versatility include harnessing UNL’s expertise in livestock management and using shared instruments for precision animal management, and UNL professor Robert Twomey, digital artist, researcher and educator at the Johnny Carson Center for Emerging Media Arts, who will explore the poetic intersection of human and machine perception.
The PNR project is structured in two distinct phases: the benchmarking phase, which includes the evaluation and expansion of the PNR for scientific and technical research (years one to three), and the grants phase, which involves widespread adoption of the PNR resource (years four to five).
“We aspire that our science pilots not only use the infrastructure, but intellectually influence each other, forming networks of mutual support that we activate through managed Slack channels in addition to more standard user support. We see this in itself as a test bed for scalable user support for an HPC system, ”said Co-PI Mahidhar Tatineni, SDSC.
The PNR will be deployed in the first half of 2022.