BU OIT

Scientific Computing and Visualization

OIT's Scientific Computing and Visualization (SCV) group provides a number of facilities and services to support research and education in areas requiring extensive computation, advanced computer graphics or scientific visualization.

At the core of SCV's computational facilities are a number of advanced, multiprocessor computing systems. The most powerful of these is an IBM BlueGene/L supercomputer. This system comprises 1024 compute nodes interconnected by several high-performance internal networks. Each compute node contains two PowerPC processors plus 512MB (million bytes) of main memory. The entire system offers a peak performance of 5.6 TFLOPS (trillion floating point operations per second). The BlueGene system is most effectively used for computations that can be done in parallel using dozens or hundreds of processors concurrently. The second facility is a collection of IBM pSeries shared memory multiprocessor systems with 8 to 32 processors each and a total of 208 GB (billion bytes) of main memory. This facility contains 156 Power4 processors, each running at either 1.1 or 1.3 GHz, resulting in an aggregate peak performance of 0.75 TFLOPS. The third facility is a Linux cluster with 52 dual-processor Intel-based nodes and 14 quad core AMD-based nodes. The Intel nodes are interconnected with Myrinet while the AMD nodes use Infiniband. Data storage for these systems is provided by a 7 terabyte disk storage array and a robotic tape storage system. These systems offer a wide range of programming languages, parallelizing compilers, mathematical and scientific libraries, graphics and visualization software, and discipline-specific application packages.

SCV's Computer Graphics Laboratory and Laboratory for Immersive Virtual Environments (LIVE) provide advanced graphics facilities for scientific visualization and virtual reality immersion. The centerpiece of this facility is SCV's Deep Vision Display Wall. The Display Wall is a very large (8' x 15'), high-resolution (18 million pixel), stereoscopic display driven by 24 Linux computers, interconnected with the Linux compute cluster mentioned above, and 24 commodity projectors. The Computer Graphics Laboratory also provides a number of Windows and Linux workstations for graphics, visualization, data preparation, and analysis.

The Access Grid Conference Facility, located in room B17 at 111 Cummington Street, is another resource run by the SCV group. This facility is one of many nodes on the Access Grid (AG). The AG consists of a suite of hardware, software, and tools to facilitate group collaborative experiences over the Internet. At the Access Grid Conference Facility, you can participate in lectures, seminars, tutorials, and conferences on topics related to high-performance computing, featuring speakers from across the country and around the world. For more information, including a calendar of events, please visit scv.bu.edu/accessgrid.

SCV staff are available to assist the University community in the effective utilization of these resources. Staff consultants can help in the use of the various programming tools and packages, the development of parallel algorithms and programs, the development of custom computer graphics tools, the application of graphics algorithms, the conversion of data into effective graphical representations, and other areas. In addition, SCV maintains on-line training materials at scv.bu.edu/tutorials.

Accounts on the scientific computing facilities are limited to faculty, research staff and students working on computationally intensive research projects or in courses approved for these facilities. Web forms for applying for accounts can be found at scv.bu.edu/scf. Resource allocation and policies are coordinated with the Center for Computational Science (website ccs.bu.edu). The Center, located at 3 Cummington Street, also coordinates interdisciplinary research and instruction programs in computational science. Accounts for the computer graphics and virtual environment laboratories are available by request for any research or instruction requiring these capabilities.

More information on the services and facilities provided by the Scientific Computing and Visualization group can be found on their website at scv.bu.edu.