EHPCC: Home

Welcome

The EHPCC exists to satisfy the research need for a local, high-throughput, computational environment that is state-of-the-art, professionally managed, reliable, and scalable. Current research questions require powerful computational capabilities to generate and analyze overwhelming amounts of data. The EHPCC provides support for high throughput investigations such as:

• neural simulation
• genomic comparison (e.g. SNP discovery, gene finding)
• biological sequence analysis
• statistical research
• algorithm research and development
• numerical analysis

Compute clusters are valuable tools in the determination and evaluation of research strategies and represent an advance in computing technology by greatly enhancing the performance of software applications which can quicken the pace of discovery. A primary goal of the EHPCC is to provide easy access to cluster resources by way of a job management product which eliminates the need for a strong ,"nuts and bolts" knowledge of the underlying infrastructure. The EHPCC represents a collaboration between the Emory College, the Information Technology Division , and BimCore (School of Medicine). The EHPCC is to be viewed as a complement to existing campus cluster resources such as the Cherry Emerson Center for Scientific Computation which offers expert consultation in targered areas such as computational chemistry.

History

An awareness of the need for a high throughput computing resource emerged early in 2004 when it was noted that within Emory College there were several separate funding requests for cluster technology. (Similar needs were expressed within the Medical School community also). In response, faculty and representatives of the College administration felt it prudent to develop a general approach to maximize institutional funds available for this sort of technology and to establish support relationships with ITD and BimCore to insure a stable operating environment for a shared Emory compute cluster. This collaborative model represents a sound basis for conveniently adding future cluster resources and users with minimal financial impact to particular departments and associated local support staff. High throughput clusters should be considered as complex scientific instruments and ,as such, are far more than mere aggregates of processors, cables, and network switches. It is expected that participating researchers will assist in the preparation of grants to refresh the underlying technology to insure it is always current and capable of addressing computationally intensive research problems.

Training

We will soon be offering a cluster orientation class to provide information on how to use the cluster which includes discussions on common terminology, available software, using the load management software, effective benchmarking, and results management. Watch this space for more details.