SGI today announced at the J. Craig Venter Genomes, Medicine and the Environment Conference that Purdue University purchased a selection of SGI® high-performance compute and storage technology from the SGI Scientific Workflow Solutions portfolio. Purdue University's Rosen Center for Advanced Computing (RCAC) will test genomics and other codes and explore cost-saving computing strategies using the SGI® Altix® 450 system configured with groundbreaking SGI® RASC(TM) (Reconfigurable Application-Specific Computing) RC100 Blades for Field Programmable Gate Arrays (FPGAs). A newly acquired SGI® Altix® 4700 system with a 30TB SGI® InfiniteStorage array will add powerful SGI shared-memory processing architecture to the Center's large array of resources. The SGI 4700 system will be used for climatology, weather modeling, nanotechnology and quantum chemistry calculations. RCAC, on the West Lafayette, Ind. campus, is the research computing activity of the university, providing advanced computing resources and services to support the computationally intensive research of Purdue faculty and staff.
"The primary reason we purchased Altix is that it gives us balance in the resources we offer, said Bruce Loftis, PhD, interim executive director of the Rosen Center for Advanced Computing, Purdue University. "The new, large shared memory Altix 4700 machine is important for a specific set of our users. We have large clusters, but this gives us a different resource that some of our research groups will find very efficient for their computationally intense calculations. The shared memory is important for the chemists, and it is also important for a number of areas that need a large shared-memory system to do their analyses and simulations, including quantum chemistry codes and climate research. Linux was an important part of our decision because our current resources are Linux-based. We don't want to have to learn another operating system."
The shared-memory processing of the SGI Altix 4700 system is expected to be of great interest to State of Indiana state climatologists, who have been using RCAC's cluster systems. One economic development project targets improved crop forecasting and planning in rural areas.
"The farmers in our state have to make a lot of decisions based on best guesses on the weather, and our state climatologist estimates that if we could improve their decision-making process by giving them better information, running these statistical models, then it could benefit the state by as much as $250 million," said Dr. Loftis. "This weather information will also be very important for disaster planning and emergency response before, during and after weather events in the state, a benefit to both rural and urban areas. Another thing that we're very concerned about in an agricultural state like Indiana is the water run-off, such as where things go once they're applied to the soil, and so mapping of our water resources and our water flow is very important -- and a computationally intensive project as well."
SGI RASC FPGA-based Reconfigurable Computing
The SGI Altix 450 system with two SGI RASC RC100 Blades enables four FPGAs to tap into an Altix system's shared memory and address space at full memory bandwidth. The FPGAs in the system can be reconfigured via software to accelerate different applications at different points in time, providing up to 100x performance improvement. At Purdue, the FPGAs will initially be working in concert with four dual-core Intel Itanium 2 processors as a separate development environment focused on code porting, production and operations.
David Braun, Research Programmer, Rosen Center for Advanced Computing, Purdue University, has chosen Mitrionics' Mitrion-C and Mitrion Virtual Processor to accelerate applications from multiple disciplines, starting with bioinformatics. Braun is especially interested in codes in nanotechnology, image processing, and high energy physics and is currently recruiting researchers who are interested in exploring the possibilities of FPGAs with SGI RASC technology.
"We're also going to be looking at RASC from an operation and production standpoint," said Braun. "We run an operations facility and we are launching an FPGA initiative that is more attuned to a production environment, and we're going to be looking at reducing the cost of operation. We already know that, out of the box, an FPGA draws one-quarter to one-third the power of a CPU. Let's say a problem comes in that needs to have 4 CPUs to do the job correctly, but we can figure out a way to cut that to 2 CPUs and 2 FPGAs doing the equal amount of work then the cost of running that code has gone down. I'm looking at RASC FPGA-based computing to take existing science and make it scalable to the masses in an economical way."
Part of making science scalable to the masses is the university's involvement in science portals and grid computing. For example, Purdue University has a very strong Nanotechnology program, and the SGI Altix 4700 system will be a resource associated with a "nano-hub," a science gateway that lets users from all over the world gain access via the Internet to nanotechnology simulation models, databases and other resources that they would not be able to otherwise access. The SGI Altix 4700 system will be joining the RCAC's other high performance resources on the Northwest Indiana Computational Grid (NWICG).
Funding for the NWICG, including the SGI technology purchase, is through the U.S. Department of Energy, with the strong support of Indiana Congressman Peter Visclosky. The main purpose of the Grid -- a partnership with Notre Dame University, Purdue University at West Lafayette, and Purdue University at Calumet -- is to examine economic development in the northwest corner of Indiana. The NWICG is a member of the Open Science Grid (OSG), a large grid headed out of the Fermi Lab, and the SGI Altix 4700 system will be the first shared memory machine connected to the OSG.
Through SGI's exclusive education reseller, James River Technical, Inc. (JRTI), Purdue University's RCAC purchased an SGI Altix 4700 with 128 Intel® Itanium® processors and 512GB memory, a 30TB SGI InfiniteStorage 4000 storage system which will be used for scratch storage and connected to the university's larger storage systems and archive, and an 8-core Altix 450 with 2 SGI RASC RC100 Blade systems. The order was placed in late September and is scheduled for November installation. The Center also plans on a strong educational component for the SGI Altix 450 with RASC including a Graduate/Undergraduate class that uses the technology, development of a seminar series to discuss issues and a training class for users.
"Purdue University's Rosen Center for Advanced Computing conducts its own research and development to enhance the capabilities of their resources and we anticipate their implementation of an SGI Scientific Workflow Solution with new shared-memory, RASC and storage technologies will provide new opportunities to conduct differentiated research and generate breakthrough results," said Michael Brown, sciences segment manager at SGI. "Adding the large shared memory processing capability of the Altix 4700 to their arsenal of scientific computing equipment will enable researchers and students to study a wide range of problems in more detail than ever before, and the use of RASC technology will rapidly advance discovery in Biosciences and other fields."