Project Columbia Aims to Accelerate Scientific Studies and Rebirth of Space Missions
NASA has chosen SGI Altix as the foundation of Project Columbia, an extensive collaboration with SGI and Intel Corporation that is expected to fuel scientific breakthroughs in space exploration, global warming research, and aerospace engineering. With Project Columbia, NASA plans to integrate a total of twenty 512-processor SGI Altix systems with a 500-terabyte SGI InfiniteStorage solution to create the Space Exploration Simulator, which will be among the world’s largest Linux OS-based supercomputers. Powered by a total of 10,240 Intel Itanium 2 processors, the Space Exploration Simulator will equip NASA scientists with one of the most sophisticated and capable supercomputers in history.
“For over 20 years NASA and SGI have worked very closely and very successfully to deliver a series of technological firsts,” said G. Scott Hubbard, NASA Ames Director. “Project Columbia will build on the knowledge we gained from this close relationship to underpin the most ambitious missions in NASA’s history. As we set our sights to the Vision for Space Exploration, we look forward to making the journey with SGI and Intel.”
“Now more than ever, NASA’s efforts are crucial to understanding our place in the universe and our destiny on this planet,” said Bob Bishop, chairman and CEO of SGI. “We are truly honored to have Altix chosen as the compute platform for NASA’s next big step forward.”
“The exploration of the universe remains one of man’s greatest challenges,” said Intel Chief Executive Officer Craig Barrett. “The ‘Project Columbia’ supercomputer designed by NASA, SGI and Intel will enable the world’s brightest designers and scientists to look a little deeper and reach a little farther in their understanding of, and achievements in, space.”
By boosting its computing capacity ten-fold through Project Columbia, the NASA Advanced Supercomputing Facility (NAS) will be able to more effectively handle such critical projects as simulating future space missions, projecting the impact of human activity on weather patterns, and designing safe and efficient space exploration vehicles and aircraft. The present collaboration builds upon the highly successful 8-year partnership that last year developed the world’s first 512-processor Linux server – based on standard, “off-the-shelf” microprocessor and open source technology — the SGI Altix at NASA Ames Research Center named ‘Kalpana’ after Columbia astronaut and Ames alumna Kalpana Chawla.
Housed at Ames Research Center in the heart of California’s Silicon Valley, the Space Exploration Simulator will rely heavily on the open standards-based Altix system’s unique globally addressable memory architecture, which allows complex data sets to be driven entirely out of memory, enabling productivity breakthroughs that conventional commodity clusters or proprietary UNIX servers can’t achieve.
“As NASA pushes the boundaries of space exploration through the solar system and beyond, simulation will play a role of increasing importance,” said Addison Snell, research director, High-Performance Computing, IDC. “NASA scientists using the computer system from SGI with Intel processors will expand our understanding of the complexities of manned and unmanned space exploration. The installation of the Space Exploration Simulator is a significant achievement for NASA, SGI, and Intel, and beyond that, it is a major step forward for the United States.”
SGI has already delivered the first three of the new Altix systems to NASA Ames, with the entire 10,240 Itanium 2 processor-based supercomputing array to be completed over the next several months. In response to a May task force recommendation by Office of Science and Technology Policy, NASA will make a portion of the Space Exploration Simulator broadly available to ensure that the entire U.S. science and engineering community has access to the nation’s most advanced supercomputer architectures.
The Altix architecture also will provide easy installation and administration of the Space Exploration Simulator. “Aside from introducing new supercomputing capabilities and performance levels,” added John Parks, Deputy Chief of the NASA Advanced Supercomputing (NAS) Division, “we expect the Space Exploration Simulator to initiate a new era in high-performance computing — one in which operating a 10,000 processor-supercomputer may be as straightforward as managing twenty workstations.” The SGI Altix family leverages the built-in SGI NUMAlink 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. Altix systems feature a fully supported, LBS (Linux Standard Based) 64-bit Linux operating system and advanced development environment specifically optimized for technical applications. SGI Altix has consistently set records for sheer performance – including breaking the 1 terabyte/second barrier on the STREAMS Triad benchmark, an industry standard measurement of memory bandwidth — and for its ability to efficiently run manufacturing, engineering and scientific applications across hundreds of processors in a Linux operating environment.
NASA and SGI are also working to expand the Altix system architecture to support globally addressable memory across 2,048 processors in a single system. Such an effort will enable NASA to push the envelope of scientific discovery because a supercomputer with a single pool of shared memory will run many scientific applications significantly faster than a cluster of small systems whose memory is fragmented and distributed across an interconnect fabric.
Users accessing the new system will need to share current and historic data, and are expected to generate new data at a rate of multiple terabytes per day. In order to keep the system’s massive compute capacity efficiently utilized, it will draw data from a 500 Terabyte SGI InfiniteStorage storage area network (SAN) and leverage a further 800 terabytes of existing data managed by SGI’s InfiniteStorage data lifecycle management solution — Data Migration Facility (DMF). To get the best possible performance and utilization out of the SAN, SGI InfiniteStorage shared filesystem, CXFS will provide all applications on the SAN instant access to all data without the cost, complexity and wait time of traditional data sharing methods. To store data that needs to be retained over longer periods of time, the disk-based SAN will be augmented with the DMF-based data lifecycle management environment which delivers faster performance than archive alternatives, but which costs significantly less than disk.