MRI: Acquisition of a high-performance computing cluster to enhance undergraduate research and education across the sciences at Oberlin College

点击翻译按钮获取中文摘要

Non-Technical Abstract:An award is made to Oberlin College to support acquisition of a high-performance computational (HPC) cluster. As part of its educational mission, Oberlin College emphasizes undergraduate research and hence endeavors to provide the facilities and institutional support to sustain meaningful, student-centered faculty research. Numerous science faculty have externally funded research programs, an increasing number of which depend on high-performance computing. The cluster will be used by more than a dozen Oberlin faculty in departments throughout the natural and social sciences to support a diverse array of research, ranging from modeling black hole collisions to analyzing the effects of organic molecules on air quality, and from determining the protein characteristics of the earliest life on Earth to reconstructing the evolution of flowering plants using massive amounts of DNA sequence data. The award will also have a dramatic impact on student research and training. More than 100 Oberlin undergraduates per year will perform analyses on the computing cluster through mentored research and class-associated projects in at least eight courses, in the Departments of Biology, Chemistry, Computer Science, Physics, and Psychology.Technical Abstract:The award will fund the acquisition of a greatly improved high-performance computational (HPC) cluster that will fill a pressing need for computational support of research, research training, and teaching at Oberlin College. To facilitate the diverse computational needs of Oberlin science faculty, a cluster that maximizes parallelization while also offering multiple nodes with relatively large amounts of RAM for protein folding analyses, de novo assembly of next-generation sequencing data, and large-scale phylogenetic analyses will be installed. The cluster will include at least 38 Intel Xeon E5 compute nodes with a total of 544 cores, with RAM varying across these nodes from 32 GB/node to 512 GB/node. The cluster will also include at least two Nvidia Tesla K20 GPUs and at least two Intel Xeon Phi 5110P coprocessors to dramatically increase processing speeds for jobs that are embarrassingly parallel, and will have at least 244 TB of long-term storage space. The HPC cluster will thus have sufficient computational power and long-term archival storage to advance the increasingly large genomics, chemistry, and physics data sets that several Oberlin faculty now use in their research, and will have the flexibility to be utilized by the increasing number of Oberlin faculty who incorporate HPC work in their research, including in astrophysics, genomics, phylogenetics, proteomics, metabolomics, biophysics, atmospheric chemistry, behavioral psychology, cognitive neuroscience, and computer science.