$2.5 Million Computing Cluster Will Allow University to Better Serve the Research Community
The Hewlett Packard Enterprise Data Science Institute at the University of Houston has partnered with the UH Cullen College of Engineering to add a third supercomputer to its stable of high-performance computers, dramatically expanding the computational power available to researchers at the University of Houston and across the UH System.
The new cluster, dubbed “Carya,” is a $2.5 million supercomputer from Hewlett Packard Enterprise, a purchase made possible in part with a Governor’s University Research Initiative award to Andrea Prosperetti, Distinguished Professor of mechanical engineering at UH and a member of the National Academy of Engineering.
Carya arrived on campus this summer and will officially go live Sept. 1. Faculty involved in computationally intensive disciplines will continue testing over the next few months, and the HPE Data Science Institute will schedule training sessions to introduce new users to the cutting-edge technology.
Institute Director Claudia Neuhauser said the new computing cluster is an important step in allowing the HPE Data Science Institute to better meet the needs of researchers.
“Carya is more powerful than the current supercomputing clusters we have at the institute, and its addition will allow the researchers we work with to run more analyses, more simulations and to do so more quickly,” said Neuhauser, who also is associate vice president/vice chancellor for research and technology transfer. “It is able to handle more data, and more complex data.”
The institute launched a collaboration with HPE in 2018, which included a $10 million gift from HPE to the University. Located in the Durga and Sushila Agrawal Engineering Research Building, it offers high-performance computing services to any researcher with UH or the UH System.
“We at the Cullen College of Engineering are very excited to acquire this state-of-the-art HPC cluster in partnership with the HPE Data Science Institute,” said Joseph Tedesco, Elizabeth D. Rockwell Dean of the Cullen College of Engineering. “The cluster will provide critical support to our faculty as we build a culture of innovation, cross-disciplinary collaborations and transformative use of data science and high performance computing in education and research.”
Carya – the name was taken from the Latin name of the pecan tree, the state tree of Texas – joins two other supercomputing clusters at the institute: Opuntia, the Latin name of the prickly pear cactus, and Sabine, named for the East Texas river. Martin Huarte-Espinosa, associate director for the HPE Data Science Institute, said the name was chosen after talks with collaborators across the University.
Carya is a distinct upgrade over the institute’s earlier supercomputers. Neuhauser said it provides twice as many computer processing cores as Sabine, previously the institute’s most powerful computer, as well as an improved communication network. In addition, the 64 graphic processing units are state-of-the-art V100 32GB models, a significant improvement over the Sabine’s P100 and first-generation V100 models.
Yashashree Kulkarni, Bill D. Cook Professor of mechanical engineering and director of research computing for the Cullen College, said Carya will boost multi-disciplinary collaborations across the college and university, as well as with industry.
“Data science and high performance computing are poised to be the drivers of research and innovation in all aspects of engineering,” she said. Kulkarni’s group works in high performance computing, performing large-scale simulations to design materials with ultra-high strength.
In addition to serving researchers from the Cullen College of Engineering, Huarte-Espinosa said the Research Computing Data Core at HPE DSI serves faculty from a variety of other disciplines, from business and the humanities to the physical sciences and computer science.
The UH Information Technology High Performance Computing group, led by Keith Crabb, manager for the research data centers, supports all research computing at the University, including the clusters in the HPE Data Science Institute.