Using InfiniBand Interconnect, Supercomputer Achieves High Performance

Voltaire, the leader in interconnect solutions for high performance grid computing, announced that the company’s InfiniBand interconnect solutions are powering the world’s largest supercomputer for NASA. The Agency’s new Columbia supercomputer, located at Ames Research Center, Moffett Field, CA, which achieved sustained performance of 42.7 trillion calculations per second (teraflops), has surpassed the performance of every other supercomputer operating today. This marks the first time that the powerful combination of InfiniBand technology and industry standard systems has created one of the world’s largest supercomputer. NASA’s Columbia supercomputer, a collaborative effort between several companies including Silicon Graphics (NYSE: SGI), Intel and Voltaire promises to revolutionize the rate of scientific discovery at NASA by reducing the computation time required to solve some of the world’s most complex scientific problems. Delivering the interconnect performance for the 10,240 CPU clustered supercomputer is the Voltaire ISR 9288, a 288 port, 4X (10 Gbps) InfiniBand switch – the largest InfiniBand switch available today. The system consists of 20 SGI® Altix® systems, each powered by 512 Intel® Itanium® 2 processors. All of the twenty 512 processor nodes are connected with a Voltaire ISR 9288 InfiniBand switch, which features multi-rail support. The switch’s multi-rail capabilities enable each node to deliver massive throughput making a significant contribution to the overall performance of the cluster. Columbia was deployed in fewer than 120 days, of which, only three weeks were needed to cable and bring up the InfiniBand fabric. Columbia’s record results were achieved running the LINPACK benchmark on 8,192 of the supercomputer’s 10,240 processors. Columbia also achieved an 88 percent efficiency rating on the LINPACK benchmark, the highest efficiency rating ever attained in a LINPACK test on large systems. “NASA, Intel, SGI, and Voltaire have built the fastest, most efficient cluster in the world – and in only 120 days,” said Walt Brooks, division chief, Advanced Supercomputing Division, NASA. “The scalability and ease of use of Voltaire’s InfiniBand interconnect solution and their dedicated support through the deployment process were key factors in achieving world class performance and efficiency for the Columbia supercomputer.” “We are extremely pleased to see the efforts of NASA, SGI and Voltaire produce such remarkable results – for both the advancement of science and computer technology – in a such a short amount of time,” said Dave Parry, senior vice president and general manager, Server and Platform Group, SGI. “Working closely with Voltaire we look forward to delivering more ‘Columbia-class’ results to our mutual customers.” “Columbia uses InfiniBand in a real world, mission-critical environment to help solve some of the world’s most complicated scientific problems. And now, this can be done faster and more efficiently than ever before,” said Ronnie Kenneth, CEO and chairman, Voltaire. “The scalability and performance of our InfiniBand interconnect solutions have enabled NASA to build the world’s most powerful supercomputer using industry-standard platforms. This is truly a breakthrough achievement.”