Addressing the Supercomputing 2005 conference in Seattle, Gates spoke to a crowd that makes scant use of the Microsoft operating system in high-performance systems. Gates sought to cross that divide by stressing the common problems of what he called mass computing and supercomputing.
"Many of these challenges that we face in software -- connecting machines together, having parallel algorithms that allow many compute systems to work on a problem and combine their results together -- these problems are very similar to the problems that exist in high-end supercomputing," said Gates.
"Its exciting to think that we can get the best brains from supercomputing and from mass computing and bring those together and make great progress in the decade ahead," he said.
Windows barely registers a pulse in high-end computing. On the latest Top500 supercomputing list, released this week, the breakdown by operating systems has Linux running nearly 75 percent of the top 500 systems; Unix has 20 percent, and Mac OS X is listed at 1 percent. Windows isn't noted at all.
"It is astonishing how they missed out on this altogether," said John Abbott, an analyst at The 451 Group in New York, referring to Microsoft's limited role in high-end technical computing. It's important for Microsoft to be recognized in the high-end market "because that's where half these technologies, like grid and clustering, are all being proven," he said.
Gates said he believes that as chips reach gigahertz speed limits, the need for "parallelism" becomes more important. The environment he sketched out imagines desktop supercomputers linked to more powerful clusters in a heterogeneous environment.
"Microsoft wants to play a role here -- to be a participant and work with partners to see how our software fits in these solutions," said Gates. "These solutions will often be extremely heterogeneous." Making certain that all these systems work together "is just one element on how software can do a better job," he added.
Gates also said that Microsoft is reaching out to more supercomputing centers to understand "what should we be doing with our software, how can it connect up to the other software that they have in a better way?"
Martin Gasthuber, a researcher at DESY (Deutsche Elektronen-Synchrotron), a high-energy physics laboratory for basic science research in Hamburg, Germany, was unimpressed. He said Gates' talk was about marketing Microsoft products. "For me, it's not a vision; it's the next step he wants to do -- which is coherent with the next generation of products he has in mind," said Gasthuber, a researcher at the facility who is working on storage issues.
But Ted Dodds, CIO for the University of British Columbia in Vancouver, said Gates offered "a very accurate description of current trends in technology and likely future possibilities."
Windows clusters are almost inevitable in many respects, said Dodds. This notion "of mass computing and very technical specializing computing teaming together, I think, is pretty evident," especially as researchers turn to low-cost commodity clusters to solve problems, he said.
"Without going into an open-source model, which I would never expect to see Microsoft necessarily do, they would develop products that will interoperate seamlessly" with the dominant open-source platforms, said Dodds.
William Kramer, general chairman of the Supercomputing conference and head of high-performance computing at the National Energy Research Scientific Computing Center at Lawrence Berkeley Laboratory, said Gates' appearance is an indication of the growing awareness of supercomputing's importance. "The output of [high-performance computing] activities are no longer hidden behind a curtain, if you will," he said.
Supercomputing is "being scaled down so more people can make use of these very complicated tools -- and I think that's one of the indications of Microsoft's interest here," Kramer said.