In search of Alpha

SAN FRANCISCO — Intel Corp.’s Developer Forum may be an opportunity for the chipmaker to outline its product roadmap and design strategy, but executives refuse to explain its plans for the Alpha intellectual property it acquired from Compaq

Computer Corp. last year.

Spokespeople for the firm have repeatedly rebuffed inquiries about the Alpha by saying that it does not discuss unannounced products, but IDF has largely been an opportunity to talk about chips that have yet to hit the market.

In a briefing on its high-end server strategy, for example, Intel enterprise processor marketing director Lisa Hambrick was bombarded with variations on the same question: To what extent will the Alpha, which was highly regarded in academic circles as a strong performer, influence the future direction of Intel’s 64-bit Itanium chip?

“”Technology from the Alpha will be introduced for products being worked on past 2003,”” she said each time. “”That’s all we’re going to say.””

Hambrick confirmed that any Alpha-related enhancements will likely be incorporated into Montecito, the successor to Intel’s second generation Itanium processor, McKinley. Montecito is not expected until 2004. (Next year’s chip is code-named Madison.) Though details about the Alpha are being kept under wraps, Intel has actually offered a substantial look at other aspects of Montecito. Like the Xeon processors released this week, for example, it will include the hyperthreading technology which essentially allows a single chip to operate like a dual processor. It will also mark Intel’s move to a new manufacturing process. McKinley is fabricated at .18 microns and Madison is expected to be made on a .13 micron process. With Montecio, however, Intel will move to a process of .9 microns.

Intel has been interested in the Alpha for a long time. Prior to the merger of Compaq and Digital Equipment Corp. in 1998, Intel was in the process of setting up an agreement whereby Intel would supply foundry services for the Alpha. With the rights it licensed last year, Intel also gained about 100 engineers from the Alpha team at Compaq, which also agreed to phase out its Alpha supercomputers in favour of Itanium-based servers. In Canada, Alphas are used in large compute-intensive environments at the University of Toronto, among other areas.

Despite industry acclaim for its features, Alpha never outsold comparable products from Sun Microsystems Inc. or IBM Corp. In an interview with Computing Canada, Intel CEO Craig Barrett said the chipmaker could give the Alpha what Compaq and Digital could not. “”I think what we learned from the Alpha is that volume economics win and standards win,”” he said. “”Alpha — although there were some attempts to take it to the open market — was really a Digital and then Compaq product. We can do a lot of work with application vendors and then make it as available and as affordable as possible.””

Nathan Brookwood, an analyst with research firm Insight 64, said a conversation with Intel chief technology officer Pat Gelsinger convinced him Alpha will be an important part of Itanium’s future. “”He definitely hinted that Montecito was likely to be the first processor where we will begin to see some of the Alpha EV8 thinking with regard to simultaneous multithreading,”” he said. “”I think they may even be doing the development on that back East in their microprocessor development centre that they started when they got all those Alpha engineers.””

Hambrick, however, said there are limits to the extent that the technology will be applied to Intel processors. Alpha instructions, for example, will not run on Itanium; these are already set. “”It’s not necessarily applying the Alpha technology itself,”” she said. “”It’s really having access to all that engineering talent and applying their expertise in developing compilers.””


Share on LinkedIn Share with Google+