HP Labs, like HP itself, is splitting into two separate organizations. Because of the split, Hewlett-Packard plans to add more employees to cover the areas of research common to the two new parts of the company, director Dick Lampman said in an interview.
The organizational change is just one of the transitions HP Labs has had to address in its long history as the mother lode of new ideas for HP's products. The labs and their 1,373 employees are grappling with issues such as the limits of current chipmaking technology, computers that build themselves, and, perhaps less glamorous, how to keep the company making money.
"We'll be expanding staff a bit," Lampman said, pointing to telecommunications, Internet networks, physical sciences, and materials science, all of which are areas of study which can impact the building of advanced new HP products. Lampman doesn't expect the split to be too difficult, compared to the ongoing issue of how to keep HP invigorated with new technologies so it can stay ahead of competitors.
The labs, based at HP's headquarters in Palo Alto, California, receive about 8 percent of the company's $3.5 billion in research and development funding, according to Rich Marconi, manager of technical communications at the HP Labs. Most of the company's R&D money stays within the company's product groups.
The labs, while independent from the product divisions, still must keep them supplied with next-generation products. "We don't have to do what those people want, and they don't have to take what we send them," Marconi said. But if the product divisions aren't kept happy, "We will hear from [chief executive] Lew Platt," Marconi said.
About 60 to 70 percent of the labs' funding goes toward work that's directly related to HP's existing business, and the remainder at new opportunities, Marconi said.
HP labs: More than a think tank
Although lab employees aren't paid to sit around "thinking great thoughts" as in the classic university research model, there's still room for people motivated by something more profound than just money, Lampman said. "Technology is my lifeblood. This is the place you can pursue that."
There's a fine line between order and chaos at the labs, Lampman said. Managers have to keep projects on track, but employees have to be cut a little slack for a few months for some "thinking on the side," Lampman said. "You don't see many great ideas coming out of committees."
The work at the labs is at an intermediate level between the product research within other parts of HP and the basic research at universities and national labs. Such high-tech research, which is often far ahead of actual shipping products, can take some unusual directions.
For example, HP Labs explored technology that was relevant to the now-dormant vision of a 500-channel television world. But the effort wasn't in vain: When the Internet grew in popularity, HP was able to dust off that particular technology and repackage it as its Web Quality of Service product in a matter of just a few months.
Sometimes HP will sway basic research by funding research grants, Marconi said. For example, HP contributes money to the Trimaran project, which involves New York University and the University of Illinois Urbana-Champaign to develop advanced compiler software. Compilers, the special programs that translate computer programs written by people into the machine language understood by computer chips, are a critical part of HP's work with its next-generation computer chip technology.
Projects kept under wraps
HP doesn't say much about what projects are under way the HP labs, but some work sees the light of day.
For example, there's the Teramac computer, which is made of defective chips assembled with error-prone methods, but that can function well anyway. The computer is an illustration of a way to build defect-tolerant computers, an important concept as chip circuitry is reduced to ever-smaller sizes.
HP also is investigating software that could make it vastly easier to design a new chip. Companies could exactly tune an architecture to meet the right requirements of size, power consumption, and price, Lampman said. Imagine, for instance, a 37-bit chip that processes 37 zeroes and ones at a time, with a 17-bit arithmetic engine, both precisely suited to the task at hand. By contrast, most chips today in that range process data in 32-bit or 64-bit chunks, which in some ways is like being able to buy a size 5 shoe and a size 10 shoe but no size 8 1/2.
One HP Labs technology that has seen the light of day is the light-emitting diode, (LED) which has grown from the tiny red displays of HP's first electronic calculator decades ago to an ultrabright, multicolored version that appear in traffic signals and brake lights. Using a cluster of LEDs instead of a single light bulb in traffic signals results in sevenfold power savings, not to mention the fact that they don't burn out, Marconi said.
Perhaps a better-known product of HP Labs, though, is its Explicitly Parallel Instruction Computing (EPIC) chip architecture, which will make its first appearance in 2000 as the first 64-bit chip from Intel. HP began work on EPIC in the late 1980s and entered a partnership with Intel in 1994 to make sure there would be a manufacturer with enough sales to offset the expense of building a new factory to make the chips.
The EPIC story illustrates the friction that sometimes exists between HP Labs and the rest of the company. HP Labs was putting its weight behind EPIC at the same time the rest of HP was trying convince the outside world to buy the existing generation of chips, based on the reduced instruction set computing (RISC) architecture, Lampman said.
Since those battles first erupted, HP appears to have embarked on a strategy that still emphasizes the performance of its own chips, while touting the ability to upgrade those systems to the next-generation chips, referred to by the generic name of IA-64 because Intel will manufacture the chips.
Already, though, even while EPIC's evangelism effort is under full steam and its IA-64 chips not even shipping, HP Labs is moving on. "We've been thinking about when it's going to be obsolete," Lampman said.