BMI and collaborators awarded $3 mil research infrastructure grant
Thursday 09/09/2004
BMI and two other research entities have been award a 5-year $1.5 million research infrastructure grant, from the NSF CISE RI program. With strong, cost-sharing support from OSU and the Ohio Board of Regents, total funding for the infrastructure is $3 million.
BMI, the OSU Department of Computer Science and Engineering (CSE), and The Ohio Supercomputing Center (OSC) will set up a networked cluster testbed, consisting of a compute cluster, a memory cluster, inter-cluster networking, a wireless testbed with hand-held and laptop clients, and a lab with visualization/haptic devices.
The proposed networked cluster configuration follows:
- The compute cluster will consist of 64 dual-processor nodes interconnected with InfiniBand, supporting 10 Gbps intra-cluster bandwidth. Some nodes will be connected with Quadrics QsNet to benefit from the extremely low latency of that technology (2 microseconds).
- The memory cluster will be composed of 50 dual-processor nodes with commodity disks, memory modules, and Gigibit Ethernet interconnect to support 1.0 Terabyte of physical memory and 48 Terabytes of disk storage.
- Gigabit Ethernet—dual fibers of 10.0 Gbps each—will provide high-speed connectivity between the clusters, as well as other existing computing, mass storage, and visualization clusters.
- 20 wireless PDAs and laptops and a number of state-of-the-art graphics adapters, haptic devices, and head-mount displays will complete the hardware set up.
The goals of this new high-speed cluster network are to advance on-going research and enable new projects, to improve graduate education in CSE and BMI by providing a state of the art testbed, and to promote new collaborations between institutions within and outside of OSU. The new infrastructure will be used by the 18 involved investigators and their graduate students (around 60), in addition to new faculty and their students.
The infrastructure will enable more productive collaborative projects that focus on challenging, data intensive, data driven, interactive applications, and it will provide a crashable and reconfigurable resource for systems researchers. It will also greatly increase each institution's capabilities for research in cutting edge visualization techiques and devices, such as surgical simulators that require haptic and headmount displays. The wireless testbed will enable us to design, develop, and evaluate adaptive algorithms and end-to-end Quality-of-Service solutions for end-clients with limited computing power and network connectivity.
[edit this news item]