Project Proposed: This project, acquiring of a computational cluster to replace a six-year-old system, allows access to fast mid-sized parallel computation to dozens of researchers and serves as the institution's primary resource for parallel computation. The system is structured for a variety of uses. Standard MPI computation is carried out with a tightly coupled cluster of quad core processors, while 'fat nodes' with 256 GB of RAM as well as local high speed disk storage service jobs that require large shared memory. Researchers actively developing codes can take advantage of the unique performance characteristics of GPU (graphics processing) nodes. The system will have several NVidia Tesla nodes. The users of the system are drawn from all five departments of the College of Engineering (Chemical Engineering, Computer Science, Electrical & Computer Engineering, Materials, and Mechanical Engineering), seven departments of the Division of Mathematical, Life and Physical Sciences (Chemistry & Biochemistry, Earth Science, Ecology Evolution & Marine Biology, Mathematics, Molecular Cellular & Developmental Biology, Physics, and Psychology), as well as the departments of Economics, Geography, and Media Arts & Technology from the Division of Humanities and Social Sciences. In addition, the system supports research in eight campus centers: Allosphere Research Facility, the California NanoSystems Institute, the Center for Polymers and Organic Solids, the Institute for Crustal Studies, the Kavli Institute for Theoretical Physics, the Materials Research Laboratory, the National Center for Ecological Analysis & Synthesis, and the Neuroscience Research Institute. The new system will be housed in the same location as the previous one where the same successful administrative and maintenance procedures used for the past six years will be applied. The proposed cluster will be accessible via the UC Grid, a web portal interface that makes high performance computing resources easy to use from desktop machines (PCs or Macs). The acquired system will come with a three-year warranty. Prior experience has shown that only a small number of nodes are expected to malfunction during the useful lifetime (> 3 years) of the cluster. Broader Impacts: The research enabled by the campus-wide facility, interdisciplinary and collaborative in nature, is available to the broad research community. The large majority of users, roughly 75%, consists of postdocs, graduate students, and undergraduates (5%), allowing this award to accomplish NSF's longstanding goal of integrating research and education. Outreach to K-12 takes place via a new initiative, The School for Scientific Thought (SST), an extension to the Let's Explore Physical Sciences (LEAPS) Program. Under the SST Program UCSB science and engineering graduate students design and teach a course for an audience of high school students on Saturdays. In addition, many of the faculty associated with this proposal participate in the UC Leadership Excellence through the Advanced Degrees program that has increased the number of underrepresented students in science and engineering at UCSB.
NSF award CNS-0960316 was used to purchase a large high performance computing (HPC) cluster for use in scientific research and education on the campus of the University of California, Santa Barbara (UCSB). Over 300 faculty, students and staff have used and continue to use the cluster. These users span the departments of Chemical Engineering, Chemistry & Biochemistry, Computer Science, Earth Science, Ecology Evolution & Marine Biology, Economics, Electrical & Computer Engineering, Geography, Materials, Mathematics, Mechanical Engineering, Molecular Cellular & Developmental Biology, Physics and Psychology as well as the following campus research institutes: the California NanoSystems Institute, the Center for Polymers and Organic Solids, the Institute for Crustal Studies, the Kavli Institiute for Theoretical Physics, the Materials Research Laboratory, the National Center for Ecological Analysis and Synthesis and the Neuroscience Research Institute. Over 50 peer reviewed scientific publications have resulted from computational studies carried out on this HPC cluster and it is anticipated that approximately 50 more will be published before the machines are retired. In addition to scientific research, the cluster serves an important role in education. Approximately 75% of the cluster users are undergraduate students, graduate students and postdoctoral scholars. In many cases, this cluster is the primary tool that these students make use of in their thesis work and related studies. The cluster has also been used in advanced graduate level coursework in molecular modeling. In the Spring term of 2012, the participants on this award (both the principal investigators and IT staff that maintain the cluster) hosted the "Southern California Simulations in Science Conference" to expose students (both undergraduates and graduates) to the wide breadth of research carried out on the HPC cluster and also the opportunities to pursue computational science in an industrial setting after graduation from college / graduate school. This conference was attended not only by UCSB students, but also from students in the greater Southern California and Central Coast of California regions. UCSB has also hosted two "school for scientific thought" programs for local high school students based upon the research carried out on the cluster. These programs introduce the next generation of scientists and engineers to modern scientific problems currently being tackled with scientific computing.
