Vendors, Researchers Set Pace To Benefit Broader MarketsVendors, Researchers Set Pace To Benefit Broader Markets

Life-science problems are supplying a spark for high-tech vendors as they shape emerging technologies in microprocessing, data visualization, and database queries. Biotechnology researchers for the past several years have pushed independent software vendors, IT departments, and universities to devise algorithms that run on large parallel computers, searching tables of DNA sequences, amino acids, and proteins for matching patterns that can indicate effective therapies. As the industry tackles even more complex problems, such as modeling proteins' behavior and determining genes' functions, vendors are building systems that balance parallel performance with floating-point capabilities and high-speed input/output.

The results could benefit broader markets. "Our strategy has been to play a role in scientific computing but do it in a way that's transferable to commercial computing, because the numbers are bigger," says David Turek, IBM VP of Linux and emerging technologies.

IBM's Blue Gene effort is accompanied by a project called Blue Lite, set up to commercialize learning from Blue Gene. Both systems use an architecture called cellular multiprocessing, in which thousands of "cells" of CPUs, memory, and embedded I/O circuits sit near data on a disk, so processors don't have to wait for central memory. If IBM can program a system like that effectively, it could learn about building computers that value response time vs. CPU utilization and keep running even as CPUs fail, says VP of systems Mark Dean. Applications could include pharmaceutical and agricultural research, broadband video delivery, and air-traffic control.

Hewlett-Packard this week will show the first data-visualization product based on an HP Labs and technical computing division research project, called Mindstorm, at the Siggraph 2001 conference in Los Angeles. The product, Visualization Center sv6, lets scientists and engineers crunch complex graphics on a server and display the results on a workstation connected via a LAN, says David Valenta, HP's life-science market development manager. Applications could include molecular modeling, industrial product design, and weapons simulations.

In some areas, IT vendors' work is just scratching the surface of users' needs. IBM, Oracle, and Sun Microsystems, along with Incyte Genomics, Millennium Pharmaceuticals, and others, formed in June the Interoperable Informatics Infrastructure Consortium (I3C) to create APIs and XML schema for life-science data queries. It's a pressing area, as researchers annotate structural gene-sequence data with video footage, articles, and lab notes.

Caprion Pharmaceuticals Inc. bio-informatics director Paul Kearney wants Sun, Oracle, and other suppliers to play an even bigger role. As Caprion negotiates a deal with a pharmaceutical company to explore causes of diabetes and cancer, its IT group must plan database queries that span customers' legacy systems and Caprion's Oracle system. The I3C's XML work is a start, Kearney says, "but it's work that's not completed."