It's official: the age of supercomputing has ended. At any rate, the word "supercomputing" has been excised from the title of the annual trade shows, sponsored by the IEEE and ACM, that have been known for almost ten years as "Supercomputing '(final two digits of year). The next event, to be held in San Jose next November, has been redesignated "SC '97." Like Lewis Carroll's Cheshire Cat, supercomputing has faded steadily away until only the smile, nose, and whiskers remain. This comment is not a criticism. The ephors of this annual event have simply joined many others who have attempted, with little success, to define the identity of this activity that now stumbles along under the clumsy, ambiguous title "High Performance Computing." Or HPC, a doubly forgettable acronym because it stands for a term whose recognition value among the general citizenry is close to zero. I do not exempt myself. I am among the multitude of people who have wasted dozens or hundreds of hours that could have been devoted to some useful activity, like making chicken soup, in the quixotic search for a replacement term for "supercomputing." The loss is a real one. An enormous range of ordinary people had some idea, however vague, what "supercomputing" meant. No-caf, lo-cal alternatives like "SC" and "HPC" lack this authority. REASONS FOR A CHANGE OF NAME This is not a trivial issue. In these days of rapid change, passing technofancies, and information overload, a rose with the wrong name is just another thorn -- or forgotten immediately. After all, how can businessmen, ordinary consumers, and taxpayers be expected to pay money for something they can't comprehend? More important, will investors and grant-givers hand over money to support further R&D on something whose only identity is an arbitrary clump of capital letters? It must be admitted that there are exceptions. A technology and culture with a rather silly name, the World Wide Web, has gained almost universal recognition. So much so that the acronym WWW is almost as well-known. (Is alliteration the answer?) WIDER BOUNDARIES The need extends beyond the gossamer realm of image creation. For a long time, the scope of HPC, as practiced by its druids, has grown far beyond the boundaries of computing, super or otherwise. An explanatory note in the call for participation mailed out by the organizers of SC'97 makes this point comprehensively: "This change [from Supercomputing '00] reflects our growing attention to networking, distributed computing, data-intensive applications, and other emerging technologies that push the frontiers of communications and computing." In fact, these frontiers have been expanding visibly from the very start of the Supercomputing '00 events in 1988. This title change is a somewhat belated acknowledgement of trends that began some time ago and have simply accelerated during the 1990s. So a truly suitable name must adapt to expansion of the technological horizon. A BROADER MARKET HPC, or whatever it is, is also expanding into additional markets. Another crucial change with far-reaching consequences has taken place over roughly the same time span, although it began to gather solid momentum more recently. This is the growing use by commercial customers of machinery developed originally for the scientific/engineering market . (A trend acknowledged delicately by the SC'97 statement with the words "data-intensive applications.") For the vendors of HPC hardware, this expansion into a much wealthier realm has been a central preoccupation. Passionate dreams about commercial prospects were undoubtedly major motivations for the decisions by IBM -- and now Sun Microsystems -- to produce HPC systems. This transition has raised problems that are cultural as well as nomenclatural. Those accustomed to the scientific and engineering origins of supercomputing often retain a visceral disdain for commercial computing. (Perhaps it goes back to the days when supercomputing was one of several brave new alternatives to the stuffy, glass-enclosed world of mainframes.) On the other hand, business customers insist that their computers be as reliable and predictable as the roof over their building, telephone service, and their plumbing. They are accustomed to paying lots of money for computational facilities, but they also want to get their money's worth. Unfortunately, for a large proportion of business users, terms like "supercomputer" and acronyms like "MPP" conjure up unpleasant expectations of creaky operating systems and frequent unplanned outages. Vendors are aware of these sensitivities. Consequently, they have resorted to a variety of euphemisms and omissions in order to disguise any connection between their glorious new products and the problematic world of supercomputing. As the SC'97 prospectus implies, the scientific/engineering and commercial sects are both jousting with the same monster. The two hemispheres of high-performance computing may differ in their insistence upon, and definition of, high performance. Yet they both must cope with the most pressing challenge of the day. This is not teraflops performance (which is of interest to only a few specialized researchers) but making sense out of databases with terabytes and petabytes of data (which is of increasing concern to large and even medium-sized businesses as well as research labs.) A new name might, somehow, imply this shared concern. "HEY, ALEXANDER, WE NEED YOU OVER HERE." So there it is. The Alexander who succeeds in cutting this Gordian knot should come up with a new term that encompasses today's technological spread of high-performance computing. In addition, a new name should be comprehensible, acceptable -- and, for that matter, sound good -- to both the scientific/engineering and commercial/data-base congregations. I give up. There must be readers of HPCwire with fresh -- and workable -- ideas. Hits & Hisses is open to suggestions. ---------------------------- Norris Parker Smith is a journalist who specializes in HPC and high bandwidth communications. Reader comments are welcome.
SUPERCOMPUTING IS DEAD; THE NEW KING NEEDS A NAME
February 7, 1997