Scientists Look Back 13 Billion Years to Learn How Milky Way Was Born

Aided by an SGI Altix system from Silicon Graphics, astronomy students and researchers at the University of Washington are trying to unravel the mystery of how an ocean of dark matter came to form the galaxies that populate the universe today. The astronomists are using an SGI Altix system powered by 24 Intel Itanium 2 processors to specifically model the life of our Milky Way from its inception billions of years ago until today. To accomplish this, researchers develop different "recipes" for the formation of stars on the Altix supercluster, and then test their Milky Way models against much larger models that include billions of galaxies. These cosmological tests take place on shared supercomputer systems with thousands of processors. "By understanding how stars form in this cosmological context, we can determine how hydrogen gas became this massive heavenly body that, in effect, lights up," said Tom Quinn, astronomy professor at the University of Washington. "We can learn how galaxies are born." The university installed the Altix system in August of 2003. Since then, researchers have been able to more quickly and accurately adjust their models, enabling more productive daily work. With each cosmological test, the team learns more about the process of star formation and galactic evolution. "The Altix system gives us a powerful, dedicated resource to run and adjust our models on a day to day basis," Quinn said. "We can't do this at a national supercomputer center. It must be worked on and tweaked constantly, so offloading isn't an option." With the astronomical models generating massive data sets in excess of 10GB, the University of Washington's Department of Astronomy knew it required a system with a low-latency, high-bandwidth architecture capable of addressing huge amounts of data and moving it quickly into and out of the system. The Altix system, outfitted with 48GB of local memory, was an ideal fit. The Altix family leverages the built-in SGI(R) NUMAlink(TM) interconnect fabric, which allows global addressing of all memory in the system and delivers data across nodes up to 200 times faster than conventional interconnects. For the first time, more complex data sets and complete workflows can be driven entirely out of memory, enabling productivity breakthroughs that traditional Linux(R) clusters or repurposed e-commerce UNIX(R) servers can't tackle. Altix systems feature a fully supported, standard 64-bit Linux operating system and advanced development environment specifically optimized for technical applications. SGI Altix has consistently set numerous records for sheer performance, and for its ability to efficiently run manufacturing, engineering and scientific applications across hundreds of processors in a Linux operating environment. The Altix architecture also enables easy installation and administration. "One of the advantages of going with Altix over a more Beowulf-type configuration was that the Altix installation was trivial compared to a cluster of PCs," said Quinn. "Programming is much easier on Altix, and it fits in nicely with our network of Linux desktop systems, which have migrated from Sun machines." With Altix, Quinn and his team can spend more time on their science, and less on system management. "The advantage of Altix is that we don't have to worry about it," he said. "I collaborate with computer scientists because I have a need for doing state-of the-art computational work. Altix gives us a powerful platform for doing just that." "The thirst for knowledge about the universe is unquenchable, and understanding its origins is one of the most essential pursuits of science," said Bill Mannel, server product line management, SGI. "The groundbreaking work currently underway at the University of Washington is illustrative of what is possible by combining infinite intellectual curiosity, tireless scientific discipline, and the world's most scalable Linux system architecture."