Promising efforts follow HP's technology slide

Promising efforts follow HP's technology slide
PALO ALTO, Calif. — The commoditization of the computer industry has blunted Hewlett-Packard Co.'s technology edge. But the organization is still nurturing a number of promising point projects.

R&D spending as a percentage of sales has declined steadily at Hewlett-Packard, from 10 percent several years ago to 5.4 percent in 2000. The figure for proposed merger partner Compaq Computer Corp. was 3.5 percent last year, and the merged company would have a still-lower spend based on statements citing plans to trim $425 million from the combined company's R&D budget. In addition, the roughly 900-person staff at HP Labs (down from about 1,400 in 1999) will no doubt be diminished by expected post-merger layoffs of 15,000 people.

Just as significantly, HP's design orientation has weakened. Chief executive Carly Fiorina zeroed in on that fact herself early in her tenure, crafting a new logo with the word "invent" inscribed below the familiar letters HP as a message to staff as much as to the outside world.

Fiorina realized HP had lost touch with its roots as "a company of inventors [that] over time became one of merchandisers and distributors, and she said we are going to change that," said Boris Elisman, general manager of HP's embedded and emerging-business organization, a grab bag of new technologies pulled out of HP Labs and turned into a business unit under Fiorina.

A former engineering manager from HP's RISC/Unix group described the problem starkly, pointing to decisions to quickly resell or OEM storage products from EMC and Hitachi before settling in to develop storage products internally. "Over time HP shifted away from being a products company, but people like me want to build products," he said.

In an effort to fire up the product-development engines, Elisman's small group is incubating a number of interesting embedded-systems projects culled from HP Labs. Perhaps the most significant is a move to commercialize magnetic RAM technology developed by HP over the last eight years. The company hopes to announce within a month a commercial deal now being hammered out with a partner it hopes will bring its MRAM technology to market as early as 2003, possibly beating IBM and Motorola.

Josh Fisher, who helped define the Wide Word processor that became the basis for the IA-64 microarchitecture HP co-developed with Intel, has designed the LX very long instruction word (VLIW) chip, which STMicroelectronics has licensed. Fisher is now working on a mobile version of the existing LX multimedia processor, which is currently aimed at set-top boxes.

The Pico approach

HP's embedded and emerging-business group has also taken HP Labs' Program In, Code Out (Pico) software to pilot customers for a trial run, to see whether it is ready for commercial use. Pico can analyze trade-offs in optimizing a hardware accelerator block on a processor and spit out VHDL code based on a user's choice of design trade-offs.

Elisman's team is also incubating a project on atomic-resolution storage that might be five years away from seeing the market. And the group owns the Chai Java Virtual Machine, which is being used in HP printers and has been licensed to e-mail handset maker Research In Motion.

But Elisman's team of 140 persons, who have been working for him just since July, has limited resources and no defined position in the merged company. "My guess is we will not be able to sustain development in all the areas we are now targeting, but will have to pick two or even one," Elisman said.

— Anthony Cataldo contributed to this story.