ACADEMIA
MSU Achieves 40 Percent Productivity Improvement With SGI Cluster Solution
- Written by: Writer
- Category: ACADEMIA
MSU's HPC Center Leverages 1,024-Core SGI Altix Cluster to Enable Breakthrough Research in a Host of Research Disciplines: To accelerate research as diverse as creating high-speed evolution of digital organisms and studying cosmological distances using exploding white dwarfs, Michigan State University's High Performance Computing Center (HPCC), a long-time SGI customer, has added to their existing technology from SGI. With the installation of a 1,024-core SGI Altix XE1300 cluster the HPCC can now offer more power and processors to researchers than most individual university labs. In use since September, the SGI Altix cluster has delivered field-proven performance for a number of very high-level studies and development. Charles Ofria, an assistant professor in MSU's computer science and engineering department, has several projects that his students are moving to the SGI system. Ofria, who also teaches courses in the Ecology, Evolutionary Biology and Behavior Program, is the Director of the Digital Evolution Laboratory and the primary creator of Avida open-source software. Developed at CalTech in a UNIX environment that included previous generation SGI systems, Avida is now ported to the Linux OS. Avida software is an auto-adaptive genetic system designed primarily for use as a platform in Artificial Life research, allowing scientists to track the birth, life and death of digital organisms. Not a simulation of evolution, but a code that actually evolves, Avida contains all the core parts of the Darwinian process: digital organisms replicate, mutate and compete with one another. Because digital organisms can breed thousands of times faster than living organisms, the faster the computer, the faster scientists can study mutations and complex genetic traits over millions of generations. Ofria's preliminary tests of Avida on the SGI Altix cluster versus his department's 200-node Beowulf cluster show the impressive speedup he was hoping for. "Our benchmarks were run on individual nodes and showed that the Altix cluster is giving us about a 40 percent speedup over our Beowulf cluster," said Ofria. "For some of the software codes and projects that we're currently developing, the speed of the Altix is going to be very useful for the close interconnection of the nodes, and because we can have projects running on multiple nodes and interacting." Ofria and his students will also tap the SGI cluster to continue developing Think Tank, a computer code that focuses on getting groups of mobile robots to work together - virtually at first, but practical applications will follow. Robotic sensors might go deep in the ocean, do ecosystem monitoring in a forest, be placed throughout a building to track movement, or regulate environments, such as climate control. Other MSU researchers lining up to use the new SGI system include Dr. Edward Brown, assistant professor of physics and astronomy. He will be using the SGI Altix system for a new project to study white dwarfs - stars that have burned up their fuel and died. Dr. Brown will use several hundred processors for the computer simulations of exploding white dwarfs, which can be used to measure the phenomenon of the accelerated expansion of the universe. Evolutionary biologist Barry Williams, assistant professor, Microbiology & Molecular Genomics, and Zoology, will use the SGI system to identify which forces of nature, such as natural selection or changes in population size, cause changes in DNA to accumulate over time, as well as those forces that prevent new mutations from arising. Many other MSU Departments including Nuclear Physics, Mathematics and Chemistry will also be harnessing the power of the SGI system which was purchased through James River Technical, Inc, (JRTI), SGI's exclusive higher education reseller. "This installation at MSU is quite significant for JRT and SGI," notes Tom Mountcastle, President of James River Technical, Inc. "SGI's technology has been a cornerstone in the contemporary development of the HPCC at MSU, first with the Altix 3700 and the TP 9300 and 9500 storage, and now with the Altix XE1300 Cluster. Working with Bill Punch, HPCC Director, and the researchers and principal investigators there have allowed us to put forth solutions that will support innovative, cross-disciplinary research for years to come." MSU purchased a SGI Altix XE1300 cluster with 1,024 cores housed in just two standard equipment racks, with a total footprint of 16 square feet of floor space. The equivalent computing power in a standard desktop form factor would require over 35 racks, or 300 square feet of floor space, and require more power and cooling. "Michigan State is one of the top research universities in the country, and their use of the SGI Altix 1300 cluster will speed data search and results in all the sciences from astrophysics to zoology," said Deepak Thakkar, Higher Education and Research Solutions Manager, SGI. "As an increasing number of national and international universities and genomic centers have discovered, the ultra-dense Altix XE clusters, which are fully factory integrated, deliver top data center density and reduced infrastructure costs without sacrificing performance or functionality."