High Performance Computing

Medusa

High Performance Compute Cluster: Medusa is a High Performance Computing (HPC) Cluster for researchers at Georgetown University. All faculty, staff, and collaborators are eligible for access to Medusa as Principal Investigators of research projects, please refer to hpc.georgetown.edu for additional info about HPC Medusa. Medusa infrastructure is continually upgraded based on research needs. 

Established Medusa (HPC) costs:

  1.  Initial Allocation Fee: Per Principal Investigator - $100 per fiscal year
  2. Initial Service Units per Principal Investigator - 100,000 SUs per fiscal year
  3. Over 100,000 Service Units - $22 per 10,000 SUs.

For an account please contact us directly by email at 
UIS-RT@georgetown.edu 
or contact the 
Account Management team

Current Cluster Status Report 

New, lower service rates for FY015 

GU Scientists in the following departments use our HPC cluster:

Biochemistry, Biology, Chemistry, Computer Science, Linguistics, Economics, Mathematics, Medicine, Physics, Innovation Center for Biomedical Informatics.

 

 

GU Use Cases:

Kertez The research in the Kertesz group relies heavily on the computational resources that are provided and maintained by UIS. Our computational chemistry research provides new insights into how molecules are held together and how these interactions can be affected by varying the chemical composition of the molecules involved. While the focus is on basic science, i.e. understanding of interactions and developing design principles based on these insights for new molecules with desired properties, we collaborate with experimentalists for the applications of these principles. An area where we are pushing the limits of computational resources involves high end quantum mechanical computations that accurately include tiny contributions of up to and beyond a billion computed electronic configurations. While each such configuration has only a nearly negligible effect, in total they are responsible for the molecules to coming together, making aggregates and performing their function, for example allowing charge carriers to carry electronic signals from molecule to molecule. In this endeavor, computers at UIS play a critical role, as does the training provided by UIS to our graduate students and postdoctoral scientists.

Additional Information about Medusa:

For the purpose of accounting, system usages are calculated on SSUs, computing core hours of all queues with each Queue Factor (QF):  SU's = Sum of CPU_Core_Hours x QF in all queues.

This is a substantial upgrade from our previous cluster. Currently, Medusa has three different classes of compute nodes: (1) primary compute nodes of the latest Intel Xeon processors; (2) large-memory and many-core nodes of AMD Opteron processors for OpenMP or large-memory projects; (3) supplementary nodes of Intel Xeon processors for development or general computing.

The modular system is also designed to be expandable on the fly for additional computing nodes and or storage nodes as required by research project initiatives. This means that PI's can request funds for nodes via grants and help us grow and maintain this valuable Institutional resource