IDC says HPC hardware revenue 2007 alone generated by workgroup and departmental platforms was nearly US$5.5 billion, just more than half of the $10 billion total. The prices on platforms in those segments range from $100,000 and below (workgroup) to $100,000 to $250,000 (departmental).
Microsoft's recent hardware-software partnership with Cray on the CX1 "personal" supercomputer aimed at financial services, aerospace, automotive, academia, and life sciences and priced at $25,000 is testament to Microsoft's plan -- as is the $475 per node price of HPC Server 2008.
That's not to say Microsoft won't make a run for the top. Earlier this year, a Windows Server 2008 HPC cluster built by the National Center for Supercomputing Applications garnered a No. 23 ranking on the list of the world's top 500 largest supercomputers, achieving 68.5 teraflops and 77.7% efficiency on 9,472 cores.
But experts say Microsoft's sweet spot will be much lower down the list.
"The Microsoft strategy is aiming hardest at verticals where Windows is strong on the desktop and then extending that Windows environment upward," says Steve Conway, research vice president for technical computing at IDC. "It includes applications such as Excel and tools like Visual Studio so people can unify their desktop and server workflow."
Microsoft also plans to integrate HPC Server with its System Center tools for application-level monitoring and rapid provisioning by releasing an HPC Management Pack for System Center Operations Manager by year-end, according to Ryan Waite, product unit manager for HPC Server 2008.
The company is aligning HPC Server 2008 with Visual Studio Team Services, and F#, a development language, designed to help write new applications and rewrite old ones for parallel computing environments.
"We are looking at the holistic system," says Vince Mendillo, director of HPC in the server and tools division at Microsoft.
Familiarity is the big theme. Windows HPC Server 2008 is built on the 64-bit edition of Windows Server 2008.
The platform combines into a single package the operating system with a message passing interface and a job scheduler built by Microsoft.
The server software, built to scale to thousands of cores, also includes a high-speed NetworkDirect RDMA, Microsoft's new remote direct memory access interface, and cluster interoperability through standards such as the High Performance Computing Basic Profile specification produced by the Open Grid Forum. The server features high-speed networking, cluster management tools, advanced failover capabilities and support for third-party clustered file systems.
"HPC is no longer a niche either in terms of hardware platform or in terms of pervasiveness," Illuminata's Haff says. "For the most part, it is using volume hardware and is being applied to all kinds of problems in all kinds of companies and organizations."
Sign up for Computerworld eNewsletters.