Monday, April 28, 2008

HP researches way to make Labs pay

Hewlett-Packard is one of the eldest statesmen in the computer community, a fact which bred the HP 3000 community's success with MPE/iX and the PA-RISC hardware. There was a time when HP took regular risks with basic research, the kind which does not always pay off in products. Computing was once driven by basic research to make leaps in technical ability.

Those risks are now rare among the major vendors of the computer community, but HP seems willing to steer its science toward enterprise computing more than it has in its recent past. Tomorrow the vendor announces a revival of HP Labs, the legendary research arm that created marvels such as cutting-edge ink technology and the chip designs which launched the HP 3000's current generation, as well as the latest HP Integrity servers.

HP Labs is one of few basic research groups still standing on the 21st Century computer landscape. IBM still operates the Almaden Research Center. Xerox's PARC center closed many years ago. HP Labs celebrated its 40th anniversary in 2006. Its Bristol, England arm extended HP's prowess in storage devices back when the HP 3000 was peaking at its largest installed base. Precision Architecture Reduced Instruction Set Computing — PA-RISC — grew up in the HP Labs.

The Labs need to be more product-oriented to survive the current bottom-line management scrutiny at HP. Projects which move to products are important to HP's migrating HP 3000 customers. Hewlett-Packard once relied on innovation first and standards second to capture and keep its customers. Some migrating customers who choose HP need assurance that the vendor will do more than the best possible integration of components for Industry Standard Servers. ISS provides the growth in HP's Business Critical Systems unit. But it's the Integrity line of RISC systems — which use HP's innovation of Itanium architecture — that must bolster the future HP-UX.

The revival of the Labs could be a sign that HP remains willing to keep up the innovation that an HP Unix platform is going to need. Without that kind of built-here-first engineering, HP's customers have to hunt harder for reasons to keep using solutions that lock users in HP's technology. CEO Mark Hurd is hosting tomorrow's event, a signal that Hewlett-Packard is willing to give its scientists room to run up bills, spending aimed at delivering knockout computing choices.

The Labs are now being run by Prith Banerjee, who left his post as dean of the college of engineering at the University of Illinois last summer. While the 46-year-old has earned scientific awards since those days when HP's RISC first became a product, Banerjee is said to have an eye on keeping the Labs pointed toward product-based research.

Soon-to-market products would be an innovation in the Labs. On the home page of the Labs' Web site, the group is promoting a science-fiction innovation: Painless injections using an HP skin patch. HP extended its printer designs several years ago to be able to create human skin with micro-needles that deliver smart doses of drugs. It's an alternative safer than being stuck with a needle, HP says, as well as a better skin patch which Irish firm Crospon hopes to sell by 2010.

The HP 3000 customers who stick with HP will want something less organic for their computing, of course. When HP introduced Banerjee as the Labs director, the vendor said his research interests "are in parallel and distributed computing, compilers, and VLSI computer-aided design."

All that sounds much closer to what the 3000 customer needs when choosing HP-UX and Integrity servers. Strong compilers are still crucial to the RISC computing process. Of course, as Labs director Banerjee won't be doing this work himself. But reports say that he's shelving some projects in the Labs in the reorganization that leads to tomorrow's revamp announcement. Pushing smart skin underneath the needs of a computer company would seem to flow from such shelving.

Banerjee has his own history of entrepreneur practices. In 2000 he founded AccelChip, a developer of software for building digital-signal processing (DSP) systems, which was sold in 2006 year to Xilinx. HP 3000 customers may remember that former HP business computing chief Wim Roelandts became Xilinx CEO after leaving HP.

The Labs still has some impact to deliver for the customer sticking with HP's computing. Recent projects cover developing computer chip circuitry to the atomic scale, software to automate data centers, and a utility computing center, where customers can get computing power based on changing needs.