SAN FRANCISCO — The electronics industry is in danger of missing the boat on using the Internet to boost design productivity.

The danger stems in part from a decline in funding for university-based EDA research by the Defense Advanced Research Projects Agency (Darpa), a prime U.S. government source, according to panelists at the 35th Design Automation Conference.

Sonny Maynard, program manager of the Information Technology Office at Darpa, confirmed that the agency's support for design automation has been diminishing rapidly since about 1992. "It's halving every two years," he said, due to changes in funding guidelines and because a number of long-term computing initiatives, which had funded EDA, were drawing to a close.

As panelists from industry and academia debated "Is the Internet a reasonable tool for computer-aided engineering?" it became clear that universities are having trouble getting funds to take their Java- and Web-based collaborative design experiments forward. Nor are the major EDA companies taking up the slack. It appeared that most are not researching Internet-based design to any great extent.

Further discussion of the issues raised by the panel will take place in a live online chat that will include some of the panelists, scheduled for June 24.

While the big EDA guns are taking an evolutionary approach — adding Internet support to established tools — some academics foresee a potential revolution catalyzed by startups and university spin-offs.

"I don't think there are any technical problems — ultimately," said panelist Richard Newton, a professor of electrical engineering and computer science at the University of California, Berkeley, who gave an impassioned overview of the potential of Internet-based design. "This is about change and whether we want to embrace it."

Newton pointed out that the EDA industry has already seen several upheavals in computing, from hardwired platforms such as Calma and Applicon machines, to Daisy, Mentor and Valid workstations, and on to Unix workstations and, now, Windows PCs.

"The Internet is a new platform and an opportunity to rethink the [EDA] problem," said Newton. "The universities drove the move to Unix and it was funded almost totally by Darpa. If you don't move to the [next] new platform, you won't simplify the design flow. We need to find someone to fund building of a prototype [design flow]."

One candidate for a prototype design flow was being demonstrated on the DAC show floor. The Vela project, funded by Darpa, is a collection of tools and methodologies being put together by six universities: MIT, North Carolina State, Mississippi State, University of Santa Cruz, Stanford and UC Berkeley.

Vela was launched as a 21-month project that was to start last September, with the goal of designing a complex, multimedia processor based on an ARM core. The aim was to be ready in time for a major demonstration at DAC in 1999.

Frank Brglez, research professor at North Carolina State, said the project is now due to finish next spring and will not necessarily be able to demo its progress at DAC next June. "We hope to get some more funds to allow us to attend DAC," he said.

Meanwhile, the Vela group last week showed an array of design-management and point tools running over the Internet. They included JavaCADD, a remote tool encapsulator used to drive Mentor place-and-route and Synopsys synthesis tools; OmniDesk/OmniFlow, a work-flow manager based on Tcl/Tk interfaces; and PPP, a low-power synthesis, optimization and simulation tool.

Raul Compasano, chief technology officer at Synopsys Inc., and Jim Rowson, a Cadence Design Systems Inc. fellow, saw the advent of EDA on the Internet as an incremental process, building on current design tools. Both said that evolution is already happening, with design support and design management provided over global corporate intranets.

Rowson went further, providing an analysis of a simulation tool in terms of its internal communications. He observed that the user interface maps well to an Internet browser, with the simulation engine suited to set up as a Web server. "But every tool needs to be analyzed to see what is best," he said. "It will take a while."

Lev Markov, chief scientific officer of Mentor Graphics Corp., also spoke in favor of a step-by-step approach. "We have to do so much hard work to persuade people to buy tools, with benchmarking and so on," he said. "How will you pay for that with an Internet tool? Will people access a $200,000 tool over the Net? I don't think so." Markov said that "the EDA industry spends 20 percent of revenue on R&D, but close to nothing is spent on tools over the Internet."

Mitch Mastellone, chief technology officer at Synchronicity Inc., was more upbeat about prospects for the Web. His company supplies design and data-management tools and has contributed technology to the Rapid industry organization's online catalog of virtual components. "The goal is a reuse-centric design environment," said Mastellone. "The Internet is the only practical vehicle."

"The pessimists tend to be the large players. The small companies tend to be the optimists," said Newton, the UC Berkeley professor. "But they can get caught up in meeting short-term needs so that the architectural issues get left behind."

Jan Rabaey, also a Berkeley EE and computer-science professor, touched on the potential for encapsulating remote EDA tools with browser-based applets that appear to users as a spreadsheet. "Each cell is a tool invocation," he said. "I send a query which gets me a number back."

In this category, Rabaey singled out WattWatcher, from Sente Inc. (Acton, Mass.). Designers using WattWatcher can generate a Web site representing an IC designed with Verilog or VHDL. The site provides information on designs analyzed with WattWatcher and can be accessed by anyone on a corporate intranet. It is intended to provide distributed project teams with the means to determine if power budgets have been met or where power issues exist.

"The Internet is not a solution," said Rabaey. "It is an enabler. And it is a necessity."

Rabaey expressed less concern than some of the panelists about university funding issues "There is quite a bit of funding," he said. "The systems and semiconductor companies are very worried, and that will prompt funding."