SBU Celebrates Upgraded $1.6M Seawulf Computational Cluster
Stony Brook’s SeaWulf computational cluster has been upgraded with a new high-performance computing (HPC) system that introduces a significant improvement in memory bandwidth, resulting in applications running 2-4x faster.
The SeaWulf $1.6 million HPC system uses Hewlett Packard Enterprise (HPE) ProLiant DL360 Gen11 servers and Intel Xeon CPU Max series processors, which became available for the first time earlier this year. Stony Brook is the first academic institution in the United States to set up this new HPC solution that uses the Intel Xeon CPU Max series with high-bandwidth memory.
The upgrade to the SeaWulf was celebrated with a reception in the Institute for Advanced Computational Science (IACS) building on November 7. Provost Carl Lejuez reflected on the impact on the campus community related to the campus values detailed in the strategic plan: community, excellence, equity, collaboration and innovation.
“As we grow over the next few years, we are going to take bigger swings and we’re going to have even bigger accomplishments than the amazing year that we’ve had. But we’re also going to build that foundation, the way in which we ensure that we are treating people in a way that is going to help them be as effective as possible,” said Lejuez. SeaWulf, he explained, provides the foundation for campus researchers and experts to become the best they can be.
“This brings to Stony Brook a computational prowess that we do not have enough of,” said Vice President for Information Technology and Chief Information Officer Simeon Ananou. “I look forward to working with all of you as we make use of SeaWulf.”
IACS Director Robert Harrison noted the upgrade of SeaWulf is the first academic deployment worldwide of the Intel Sapphire Rapids processors with high-bandwidth memory (HBM) and Infiniband NDR networking. “Those technical specs boil down to this — moving data from the memory to the processor or between processors is much much faster,” he said. “In turn, it’s going to be much easier for Stony Brook University researchers to access the peak computer speed, with some users experiencing 3+ speedup over our previous fastest computers.”
SeaWulf is co-managed by the IACS and the Division of Information Technology (DoIT). IACS has long provided supercomputing capacity available to faculty and students. Most recently, IACS installed a supercomputer called Ookami in 2020.
“The National Science Foundation (NSF)-funded Ookami computer testbed for the Japanese supercomputer processor Fujitsu A64FX made the new system possible — it gave us the experience and confidence that high-bandwidth memory (HBM) could have a transformational impact, as well as the credibility to make a competitive proposal to NSF. The upgraded SeaWulf will be much easier to use than Ookami since SeaWulf is using a standard Intel processor with performance characteristics familiar to Stony Brook researchers,” said Harrison.
Currently, 43 departments are using SeaWulf, with nearly 2,500 overall users and 436 total projects. And it is not just scientists and mathematicians using SeaWulf.
Jason Jones, professor in the Department of Sociology and the Institute for Advanced Computational Science, studies the language individuals use to describe themselves. “I use Seawulf to conduct this research “at scale” — meaning with data from millions of individuals across dozens of nations,” said Jones. “Without the Seawulf cluster, it would be impossible for me to perform natural language processing analysis at this scale. The cluster makes it easy and fast to ask and precisely answer sociological questions about self-concepts,”
Professor Heather Lynch in the Department of Ecology and Evolution uses satellite imagery collected from drone to count seals, penguins and birds in Antarctica. She works with datasets in the petabytes, and the SeaWulf has revolutionized the speed in which the data is analyzed. Katherine Gallagher, a postdoctoral researcher in Lynch’s lab, noted, “SeaWulf has been extremely helpful in speeding up analysis that is not possible on other clusters due to software differences and takes weeks on local machines.”
The HPE ProLiant DL360 Gen11 servers also improve cost savings and reduce the data center footprint for the university with a closed-loop liquid cooling capability. The cooling solution, which requires fewer racks to maximize space efficiency, provides alternative coolant solutions that do not require additional plumbing by removing heat from various system components and transferring it to coolant tubes to cool down the system.
Funding for the Seawulf was provided by the National Science Foundation (NSF Major Research Instrumentation award 2215987), matching funds from Empire State Development’s Division of Science, Technology and Innovation (NYSTAR) program (contract C210148), plus crucial funding from Stony Brook’s President, Provost, Vice President for Research, CIO, and the chair of the Department of Materials Science and Chemical Engineering. Additional funding was provided by the Deans of the College of Arts and Sciences (CAS), the College of Engineering and Applied Sciences (CEAS), the School of Marine and Atmospheric Sciences (SoMAS), and the IACS, without whose leadership, vision, and financial support this would not have been possible.
— Beth Squire