The ARI High Performance Compute Cluster was set up to address the increasing need for parallel computing at the Institute. Research areas that will be utilizing the cluster include the computational bioinformatics and bioimaging, traffic visualization and power systems simulations. At present, the Computational Bioinformatics and Bioimaging Lab (CBIL) at ARI is using the cluster to reduce computational cost in bioinformatics applications projects, namely the Robust Biomarker Discovery and the Generalization Performance Assessment of a Diagnostic Predictor in the caBIG™-cancer biomedical informatics grid sponsored by the National Cancer Institute (NCI) of the National Institutes of Health (NIH). In the near future, the The Spatial Data Management Lab will begin using the cluster as its computational backbone in its Advanced Interactive Traffic Visualization System (AITVS) project, an incident detection and traffic modeling project in the Northern Virginia transportation corridor. 

ARI's HPCC is a sixteen node cluster consisting of the following hardware and software:

Head Node:               HP Proliant DL385 G1 server, 4GB RAM, two dual core 2.4GHZ AMD64 prosessors
Domain Controller:     HP Proliant DL385 G1 server, 4GB RAM, two dual core 2.4GHZ AMD64 processors
Compute Nodes:        Sixteen HP Proliant DL145 G2 servers, 1GB RAM, dual core AMD64 processors
Storage Server:          HP StorageWorks Modular Smart Array 20 (1TB to be installed)

Operating System:       Windows 2003 64-bit Release2
Cluster System:           Windows Compute Cluster Server 2003*
Applications:                Matlab 2006b, MS Visual Studio 2005, SQL2005

* ARI has been designated by Microsoft as one the "Early Adopters" of its Windows Compute Cluster Server 2003 technology.

ARI's HPCC has been in use by the Computational Bioinformatics and Bioimaging Lab (CBIL) in their bioinformatics applications projects, namely the Robust Biomarker Discovery and the Generalization Performance Assessment of a Diagnostic Predictor in the caBIG™-cancer biomedical informatics grids. The group has posted the performance gains obtained by using the cluster in a series of reports below. Moreover, the group presented its result at Microsoft eScience Workshop which was held at at The Johns Hopkins University during Oct 13-15, 2006.

Performance Measure of Microsoft Compute Cluster Server 2003 I
Performance Measure of Microsoft Compute Cluster Server 2003 II
Performance Measure of Microsoft Compute Cluster Server 2003 III
Performance Measure of Microsoft Compute Cluster Server 2003 IV

Cluster-based High Performance Computing (HPC) Applications to Bioinformatics,
presented at theMicrosoft eScience Workshop for 2006.