CAGIS announces the development of a supercomputing cluster

CHARLOTTE NC MARCH 13, 2017. Dr. Wenwu Tang, CAGIS Executive Director unveiled a new, high-performance (throughput) computing cluster/grid, (tentative name: CAGIS-Condor), to support the Center’s CyberGIS initiative. Currently, the CAGIS-Condor cluster has 114 CPUs and 488G memory. Significantly, this cluster is Windows-based–i.e., an ArcGIS capable “supercomputer” for parallel geocomputation!

The CAGIS-Condor cluster functions as a mini-supercomputing platform based on cutting-edge cyberinfrastructure technologies–desktop grid computing enabled by HTCondor (https://research.cs.wisc.edu/htcondor/). HTCondor is the core middleware that empowers DOE Open Science Grid—a flagship worldwide cyberinfrastructure (https://www.opensciencegrid.org/). This development keeps pace with other universities that are currently building desktop grid-based supercomputing platforms, such as the BoilerGrid at Purdue University, which has 36,524 CPU).

The CAGIS-Condor cluster is based on the networking of desktop machines (without any extra hardware cost). This allows researchers to better use computing resources only when they are idle (so-called opportunistic computing resources; e.g., lab/library machines during night or weekend). While the number of CPUs on the cluster is 114 now, the CAGIS-Condor will continue to grow.

Dr. Tang invites Departmental colleagues and their student to use the CAGIS-Condor Cluster on request.

Dr. Tang acknowledges the work of to Dr. Xiang Zhao (visiting scholar), who lead this CAGIS-Condor initiative. Special thanks is also given to Jiyang Shi, Minrui Zheng, Yu Lan, and Charles Andrews, as well as the Department Geography and Earth Sciences, CLAS, and MSEAP program for their continual and strong support.