INDUSTRY
IBM Unleashes New Breed of Hybrid Supercomputers With System Cluster 1350
Top Research University and Global Pharmaceutical Company Enjoy Early Success With IBM-Based Supercomputers: IBM announced today a new line-up of next-generation processor systems and blades, high speed connectivity and coprocessor acceleration technology for the IBM System Cluster 1350.
With its improved ability to handle a broad range of application environments and capability to scale up to 1,024 nodes, the System Cluster 1350 is a primary choice for building supercomputing solutions for a variety of industries including financial services, industrial, petroleum and life sciences. "The IBM System Cluster 1350 provides clients with an arsenal of technologies to tackle the challenges of high performance computing environments, while reducing the complexity and risk of configuring, deploying and managing Linux clusters," said Wendy McGee, director, IBM Cluster Solutions. "We've collaborated with leading-edge processor, accelerator and networking companies to provide clients unprecedented levels of speed, choice and flexibility with this offering." One higher education client, Indiana University (IU), recently flipped the switch on the 23rd most powerful supercomputer in the world, "Big Red," to support research into new discoveries in the life sciences, astronomy, informatics, computational physics and the humanities. IU's new Cluster 1350 can perform more than 20.4 trillion numerical operations per second. At the dedication ceremony for Big Red in August, IU Provost Michael A. McRobbie said, "The first supercomputer IU acquired after my arrival in 1997 had a peak processing power of roughly 25 billion mathematical operations per second. Just one of the 512 individual computing blades in Big Red is capable of more computation than that entire supercomputer in 1997. In fact, Big Red is capable of nearly 1,000 times more computation." IU Associate Vice President for Information Technology Craig Stewart said, "A critical element of Big Red's value to the local and national scientific communities is the integration of this computational power in IU's local cyberinfrastructure, which provides excellent support for data-centric research, as well as the availability of Big Red to the national research community via the TeraGrid." Big Red will support a variety of research applications, but will be targeted particularly toward support for the life sciences. (For more information on how US researchers can obtain allocations on Big Red, see its Web site.) Biopharmaceutical company Telik has put its System Cluster 1350 to work to find new cures for cancer. According to Telik, it has experienced a 5x improvement in processing power utilizing the cluster technology for its complex research. "We're dedicated to discovering, developing and bringing new treatments to market as fast as possible. IBM's System Cluster 1350 has allowed us to complete calculations 50 times faster than our previous solutions, and since our compute jobs can run for several months, the Linux operating system offers reliability that is critical in our drug development," said Robert T. Lum, Ph.D., Vice President, Preclinical Development, Telik. Product Details and Availability IBM's new System Cluster 1350 includes clustering technologies that provide clients unprecedented levels of speed, choice and flexibility to create hybrid supercomputers in a fully integrated, factory-built and tested solution. Products available in the new IBM System Cluster 1350 portfolio today include: -- AMD Opteron-based systems, including IBM System x3455, System x3655, System x3755, BladeCenter LS21 and BladeCenter LS41.
-- Intel Xeon-based systems, including IBM System x3550, System x3650 and BladeCenter HS21
-- IBM Power-based systems, including IBM System p5 505, System p5 505Q, System p5 510, System p5 510Q, System p5 550 and System p5 550Q and BladeCenter JS21
-- IBM Cell Broadband Engine-based systems, including IBM BladeCenter QS20 IBM's new System Cluster 1350 offers unmatched networking options, including Ethernet, Infiniband, and Myrinet interconnects that enable information flow among the systems. IBM is working with leading network providers to bring the Autobahn to the System Cluster 1350. Some of these new choices include: -- HyperTransport (HTx): a high bandwidth, low-latency connection that can be leveraged with AMD Opteron-based systems. System Cluster 1350 is one of the first in the industry to support the Qlogic/PathScale HTx adapters for the new servers. This capability, combined with the HyperTransport technology and other key technical features of these servers, makes them ideal for memory-intensive High Performance Computing and Business Performance Computing applications that have previously been constrained by processor-memory and I/O bandwidth performance. -- Cisco high performance Ethernet and Infiniband networking and BladeCenter products, including the native 10Gbps BladeCenter H Infiniband switch blade, offer enhanced manageability, choice based on HPC application requirements and proven interoperability with IBM 1350 products. -- Cluster 1350 broadens the Ethernet portfolio with the inclusion of switches from Force10 that allow for flexible designs in cluster systems. -- Myri-10G (10Gbps) adapters and switches from Myricom for high speed and low latency networks. In Ethernet mode, the Myri-10G adapters can be used with commercial 10 Gbps Ethernet switches and perform at near wire rate speeds using TCP/IP and UDP/IP protocols. -- Clients can now double the bandwidth of their communications from 10 Gbps to 20 Gbps with the new Voltaire Infiniband products using Infiniband 4x Double Data Rate adapters and switches. A new innovation -- the Voltaire InfiniBand Pass-through Module for Blades can eliminate the need for intermediate switches and provides direct connections to large core switches that can be long distances from a BladeCenter without sacrificing performance. IBM System Cluster 1350 will also offer ClearSpeed Advance accelerator boards, which expands hybrid cluster architecture options. The PCI-X adaptors are designed to improve the performance of many numerically intensive workloads by routing math library routines to the ClearSpeed accelerator board. This acceleration process is transparent to the end user and the application, except for improved speed of calculation. IBM System Cluster 1350 pricing will vary depending upon specific configuration. For more information about IBM, go to: its Web site