Scientific discovery today depends as never before upon ease of access to data, associated sophisticated tools and application, to enable research, education. Researchers who once worked in local, isolated laboratories now collaborate routinely and on a global scale. Specialized instruments that were spread across multiple locations can now fit into a single
lab connected via cyberinfrastructure resources. Set within this evolving cyberinfrastructure, networks have become the primary artery connecting scientists to each other and to the data so critical to their work. Going forward, such networks arc likely to evolve to become “data mediums” where data can be positioned to serve an ever-changing tool for scientific discovery.
The cluster and visualization instrumentation enhances existing research activities currently funded at the University of Arkansas at Pine Bluff, to combine a shared facility that invigorates a wide range of modeling and analysis project across multiple disciplines, enabling new research and education activities. Also, the new cluster provides training opportunities to a majority minority student population by exposing a new generation of young scientists to numerous research facets in HPC and the exciting area of high-performance computing.
The resource is managed by the Department of Mathematical and Computer Sciences, in collaboration with the Departments of Agriculture, Biology (Biotechnology Research group), Physics and Chemistry (Nanosciences Research group), and Mathematics and Computer Sciences with Intel Xeon Sandy Bridge processors. In addition, research groups are leveraging existing infrastructure, which includes a distinctive web based front-end job submission, remote visualization tool developed by Advanced clustering known as eQueue, which enables a research to submit jobs without specialized HPC training or knowledge. Also, the resource offers the city of Pine Bluff, and its population unique opportunities to engage in tangible data-intensive computing activities, while gaining exposure to the complex discipline of data-oriented scientific discovery.
1x Head node
9x GPU compute nodes
1x High Memory GPU compute node
1x Video wall
9x GPU/Compute Nodes (node01 - node09)
Dual CPU 6-core Intel Xeon
320GB of RAM
Nvidia Tesla K20 5GB GDDR5.
Capable of 30.52 Tflops
1x Head (Apollo)
32TB of RAM
16x 2TB drives RAID 6
Runs SGE scheduler, DHCP and Web server