|
Easing the move to supercomputers
IBM and Microsoft team up to boost productivity in high-performance computing environments
Email the article  Download pdf
 When commercial organizations and research institutes move from desktop computers or workstations to high-performance computing (HPC) clusters, they anticipate big performance gains. They’re also often concerned about return on investment (ROI), accommodating multiple operating systems, and keeping up staff productivity in a new, potentially unfamiliar computing environment. IBM and Microsoft are partnering to bring solutions to the HPC community that address these concerns.
IBM® System x® and IBM BladeCenter® servers with Windows HPC Server 2008 provide a cost-effective high-performance computing platform that improves the productivity of administrators, developers and users of HPC systems. Windows HPC Server 2008 enables broader adoption of HPC by simplifying deployment, administration and management as well as easing HPC application development. As a result, more organizations can benefit from the increased performance provided by HPC systems.
Ensuring big performance gains was critical for the Research Informatics Core at Nationwide Children’s Hospital in Ohio. The Core staff collects thousands of tissue samples from cancer patients and digitizes them for study, but using desktop computers was slowing the process. To speed research results, the Core had an opportunity to move from PCs to the IBM HPC cluster-based Ohio Supercomputer Center (OSC). There was just one problem: OSC used a Linux® operating system, while the hospital’s tools were designed for Microsoft® Windows®.
Familiar Windows environment helps boost productivity
Since other OSC users had also been asking for Windows, OSC welcomed the opportunity to expand its environment. IBM and Microsoft worked together to help the hospital implement the Windows HPC Server 2008 operating system on a 16-node cluster with 64 cores. OSC integrated the system with its existing Linux environment to allow users to access image files stored in the Linux-based storage shares.
Along with faster compute times, the hospital has found that its choice of operating system has made a difference in ease of use. “The system provides a smooth transition from the desktop to the supercomputer for our researchers,” says Tom Barr, biomedical imaging manager for the Research Informatics Core. “They can make greater strides because they’re working faster in an environment that they’re already comfortable using every day.”
Multiple operating systems attract new researchers
Return on investment was a key objective when King Fahd University of Petroleum & Minerals (KFUPM) in Saudi Arabia commissioned its new IBMbased 1,024-core HPC cluster. Moving from workstations to HPC systems would enable researchers at KFUPM to use more realistic test cases for seismic studies and advanced fluid dynamics work. But to ensure rapid ROI, the cluster needed to attract additional researchers—and that meant multiple operating systems to support diverse applications.
KFUPM implemented Windows HPC Server 2008 and Linux in a dual-boot environment with the help of IBM and Microsoft. Researchers are using the new cluster to solve complex problems more quickly, and the university is pleased that the cluster is attracting researchers from around the world. “We’re giving our researchers a premium high-end computing environment where they now can run larger and more realistic test cases, and importantly, we are also focusing on getting an enhanced ROI through increased usage,” says Dr. Sadiq M. Sait, chief information officer at KFUPM.
Commercial organizations and research institutes can meet their objectives with HPC solutions from IBM and Microsoft, whether it’s performance, ROI, productivity—or all of the above. IBM high-performance cluster solutions with Windows HPC Server 2008 allow researchers, system administrators and application developers to take advantage of their everyday familiarity with the Windows environment while accelerating time to solution by using high-performance computing systems.
Additional information is available at
|
 |
|
