Replacement of the Social Science Computing Cooperative’s HPC Cluster with a Multi-purpose SLURM Cluster
Social science research relies increasingly on computationally intensive methods. The field has seen a dramatic expansion in available data as well as the size of these data sets. The availability of powerful computational tools has pushed the research frontier to expect richer and more complicated models. Empirical inference methods are expanding to include machine learning and other methods suited for large data sets.
A common challenge that arises in applied fields of macroeconomics, international trade, industrial organization, labor and public economics involves the solution of dynamic problems where agents make choices subject to a current state and these choices affect the transition paths into future states.
This project will create a new computing cluster, managed by Simple Linux Utility for Resource Management (SLURM). SLURMis a popular tool at large research computing facilities, including UW–Madison’s Center for High Throughput Computing, due to its versatility. SLURM can manage multi-node jobs like those run on the SSCC’s current HPC cluster as well as single-node jobs and jobs that require GPUs.
This project puts campus in a good position to meet the computational needs of data science. It also effectively positions members of the campus community to engage in cutting-edge social genomics research, which requires large amounts of computational power, memory and data storage. Much of it is covered by HIPAA and thus requires secure environments like SSCC’s Silo.
Eric Grodsky, professor of sociology
Andrew Arnold, director of the Social Science Computing Cooperative
Dan Bongert, Linux system administrator of the Social Science Computing Cooperative
Russell Dimond, associate director of the Social Science Computing Cooperative
Ryan Horrisberger, technical services manager of the Social Science Computing Cooperative
David Thompson, Linux system administrator of the Social Science Computing Cooperative
Social Science Computing Cooperative