Boston University | Center for Computational Science
HomeNews and EventsResearchEducationPeopleSeminarsFacilitiesContact Us

Computational Science at Boston University

October 29, 2007

Boston University has supported computational science for over 25 years by providing computing and communications facilities to the University community as a resource for research and education much as it provides libraries, rather than on a cost recovery basis. In the early days, this meant that researchers were given access to what would have otherwise been unused cycles on the academic mainframe. For more than fifteen years, we have provided separate, dedicated supercomputing facilities to our community under the same general philosophy, while some other institutions require that such computer time be explicitly paid for out of research funds. This approach has enabled our researchers to do things they could not readily do elsewhere and has led to numerous awards from external agencies for their computational research and for the enhancement of our facilities.

In 1990, the University established the Center for Computational Science (CCS), under the direction of Prof. Claudio Rebbi of the Physics Department, as an interdisciplinary research center with the mission of fostering and coordinating the development of computational science programs and facilities across the institution. The establishment of CCS marked the formal recognition by the University that computation was taking its place alongside theory and experiment as a fundamental mode of scientific research. CCS works closely with the Scientific Computing and Visualization Group (SCV) of the Office of Information Technology, which is responsible for the operation of the supercomputers and related facilities. CCS and SCV work together to manage the process of allocating resources and planning the ongoing development of the facilities.

The University’s early recognition of the importance of support for computational science has led to national recognition for our pioneering research in supercomputing applications and our pioneering educational programs in high performance computing. The University’s investment in computational facilities has enhanced recruitment of faculty and students and has generated several million dollars in NSF funding for major computing equipment as well as millions more in vendor support. While we do not have a comprehensive list of the many research projects over the years that have included significant computational work, we have kept track of the funding that has been awarded directly to CCS, in many cases in support of our high performance computing and communication facilities, and funding that has been awarded to research proposals submitted jointly by CCS and various academic departments. These awards have totaled more than $60 million since 1992.

One notable aspect of our computational science program is that we make our resources available to instructional programs as well as to research projects. It is likely that this can only be done where there are local resources, since the national centers are fundamentally intended to provide cycles for research. We have received a number of awards, listed below, for the establishment of instructional programs at both the undergraduate and graduate level, including most recently a $3.2 million, five-year IGERT award involving nine University departments.

From its inception, CCS reached out to colleagues at other institutions, establishing us early on as something of a regional facility as well as a University resource. Since 1995, this extramural involvement has expanded with the support of the NSF, first through the Metacenter Affiliates program and then as part of the restructured national supercomputing center program. As one of five midlevel centers partnering with NCSA, we achieved a significant degree of national visibility and involvement. Awards that have supported this and other national collaborations are also listed below.

Today, Boston University has one of the largest supercomputing installations among all universities in the nation. In addition to supercomputing resources, we also provide advanced graphics facilities for visualization of research results and have maintained a leading position in high-performance regional and national networking for many years. We were the first institution in New England to connect to Internet2, and in recent years have been working in partnership with Harvard and MIT to build and enhance regional network connectivity for ourselves and the other leading research institutions in New England through establishment of the Northern Crossroads (the NoX, see www.nox.org). We also continue to participate in various national grid computing projects using the Internet2 Abilene network.


Current High End Computational Resources

In the spring of 2005, we installed an IBM BlueGene supercomputer. At that time, our machine ranked 59th on the TOP500 list of the fastest computers in the world. Boston University was the fifth site in the US and eighth site in the world to install a Blue Gene system. (Today our BlueGene still appears in the Top500 list, ranked 374 in the world in the June 2007 release.) We also operate a large complex of IBM Power4-based shared memory systems running the AIX version of Unix and a smaller cluster of Intel-based machines connected by a high-speed optical network. Together, these facilities provide a peak capacity of approximately 6.8 Teraflops (trillion floating point operations per second). Specifically, we currently provide:

? IBM BlueGene, 1024 dual processors, 5.7 Tflops peak
? IBM P690, 112 processors, 580 Gflops peak
? IBM P655, 72 processors, 370 Gflops peak
? Linux Cluster, 52 Pentium two-processor nodes for computation with 2.2 gigabit/sec Myrinet interconnect, 130 Gflops peak.
? IBM ESS Model 800 Storage Server with 1.15 Terabytes of storage.
? Deep Vision Display Wall, a scalable, high resolution stereographic display wall powered by a 24 node IBM Intellistation.
? Access Grid Conference Facility – a node on the Internet2 high-performance network using experimental suite of hardware and software to facilitate group collaboration around the world.
? 1Gb/s connection to NoX 2.4 Gb/s Abilene-Internet 2 port

Summary of Major Awards

Awards for Instrumentation

Architectural Enhancement for Connection Machine Environment – NSF – 1992 -- $200,000.

Acquisition of a New Generation Connection Machine for Multidisciplinary Research and Training in High Performance Computing – NSF – 1994 -- $2,000,000.

Creation of a High Bandwidth Regional Infrastructure with National Connectivity – NSF – 1997 -- $350,000.

Acquisition of a Power4-based IBM SP and PC-based Scalable Display wall for Multidisciplinary Computational Science Research – NSF – 2001 -- $938,315.

IBM Sponsored University Research (SUR) Equipment Grant – IBM – 2001 – equipment list value $400,000.

IBM Sponsored University Research (SUR) Equipment Grant – IBM – 2002 – equipment list value $800,000.

MRI – Acquisition of a BlueGene/L Supercomputer for Computational Science Research – NSF – 2005 -- $800,000.


Awards for Interdisciplinary Computational Science Education

Undergraduate Curriculum in Massively Parallel Computing – NSF – 1992 -- $397,697 (Chemistry, Computer Science, Electrical and Computer Engineering, Physics).

Integrating High Performance Computing into Research: Molecular Dynamics Simulation in Chemistry, Physics and Engineering – NSF – 1994 -- $562,500 (Chemistry, Electrical and Computer Engineering, Physics).

IGERT: Graduate Research Training in Bioinformatics – NSF – 1999 -- $2,694,389 (Biology, Biomedical Engineering, Chemistry).

IGERT: Multidisciplinary Integration of High Performance Computing in Ph.D. Programs in Science and Engineering – NSF – 2002 – $3,224,043 (Aerospace and Mechanical Engineering, Astronomy, Chemistry, Computer Science, Cognitive and Neural Systems, Electrical and Computer Engineering, Manufacturing Engineering, Mathematics and Statistics, Physics).


Awards for Regional and National Collaborations

MARINER: Metacenter Affiliated Resource In the New England Region – NSF – 1995 – $821,441.

Partnerships for Advanced Computational Infrastructure (PACI): Regional Partners – NSF – 1997 -- $2,433,018.

Partnerships for Advanced Computational Infrastructure (PACI): Education, Outreach and Training (EOT) – NSF – 1997 -- $1,183,879.

Partnerships for Advanced Computational Infrastructure (PACI): Access Grid – NSF – 2003 -- $150,000.

Collaborative Research: EPIC: Engaging People in Cyberinfrastructure – NSF – 2005 -- $1,500,000.

Selected Research Awards

National Computational Infrastructure for Lattice Gauge Theory – DOE – 2001 -- $772,989.

ITR/AP: An International Virtual Data Grid Laboratory for Data Intensive Science – NSF – 2001 -- $1,195,000.

ITR – (ASE) – (sim): Optimized Dirac Inverter for Quantum Field Theory – NSF – 2004 -- $400,000.

ITR: Collaborative Research: (ASE+ECS) – (int+dmc+soc) – Environmental Computing for Global Grids – NSF – 2004 -- $375,000.

The U.S. ATLAS Research Program: Empowering U.S. Universities for Discoveries at the Energy Frontier – NSF – 2005 -- $1,450,000.

National Computational Infrastructure for Lattice Gauge Theory – DOE – 2006 -- $866,741.

Collaborative Research: Multigrid QCD at the Petascale – NSF – 2008 – $743,095.

 

 


copyright © 2006, Center for Computational Science | Boston University , MA, 02215