Disturbing the 'microprocessor ecology'
By Peter N. Glaskowsky, Photograph by David Toerge - September 1, 2001
In the real world, there are millions of distinct ecological niches and millions of species of plants and animals to exploit them. Most niches are small, and are filled by small creatures. There are only a few ecological niches that allow large animals or large numbers of animals, such as herbivores on America's Great Plains, or fish in the ocean.
Nature can be a useful analogy for the microprocessor market. Most of the processors in the world are adapted for specific embedded applications, such as handheld PDAs or network control. Only a few niches exist for large processors. Of these niches, the one for servers and workstations supports the biggest, most complex CPUs.
These systems, sold exclusively to professional customers who need them for high-value applications such as inventory management and product development, deliver a return on investment that justifies the use of the most expensive processors in the world. For example, a single Pentium III Xeon processor from Intel Corp., or one of Sun Microsystems' UltraSPARC III chips, can cost more than $4,000—nearly three times as much as a complete midrange PC.
Processor architects, and executives of major system OEMs, rightly regard the server/workstation market as a special challenge. As Frank Sinatra sang of New York City, "if you can make it there, you can make it anywhere." Over the years, the desire to create the fastest processor or system on the planet has clouded the minds of a great many people. As a result, the server/workstation processor niche generally has been overcrowded.
Intel's Itanium is the first new entrant in this market in years, though Merced—the first Itanium processor—is not very competitive. It poses no immediate threat to anyone in the server and workstation market. If Intel only had one or two billion dollars to invest in the Itanium effort, there would be no long-term threat either. Instead, the company's investment easily will exceed $5 billion in Itanium's first 10 years, starting with the announcement of the Intel/HP partnership in 1994.
Such large amounts of money can disturb the ecological balance all by themselves. Compaq Computer Corp. recently decided it would rather ride Itanium than be eaten by it. The company agreed to cease further development of its proprietary Alpha architecture, acquired in 1998 from Digital Equipment Corp. (see "When Alpha is Omega," page 37). Compaq sold Intel a license to the intellectual property behind Alpha processors and compilers, and will transfer hundreds of engineers to Intel to incorporate Alpha technology into Itanium chips and software.
Some of this technology likely will find a home inside future generations of Itanium. Compaq told me to expect elements of the now-defunct Alpha 21464 architecture in "Itanium 2" processors in 2004, but Intel has been less forthcoming about its plans. It's also likely that other aspects of the 21464 design—those not relevant to Itanium's explicitly parallel architecture—will never see the light of day. Intel surely will not want Compaq to make Alpha technology available to Sun, for example.
Just as the extinction of an animal species reduces biological diversity, the loss of Alpha diminishes our industry's technological diversity. Alpha's demise frees up resources that can be redirected to the greater benefit of Intel, Compaq, and computer buyers, so we shouldn't be too sad. And please, there's no need to call Greenpeace: If there's a whale to save, it's Merced, not Alpha.
Peter N. Glaskowsky is a senior editor of the Microprocessor Report, a Cahners publication. For more information on topics covered in this column, visithttp://www.chipadvisor.com. Send Peter e-mail email@example.com.