Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > EDA/IP

Interoperability: a sore point for tool suppliers, users

Posted: 21 Mar 2002 ?? ?Print Version ?Bookmark and Share

Keywords:interoperability? EDA? International Symposium on Quality of Electronic Design? ISQED? CAD tool?

Strongly divergent views of EDA interoperability surfaced at a panel discussion at the International Symposium on Quality of Electronic Design (ISQED), as users called for open standards and vendors expressed skepticism of such efforts. While design managers from Intel Corp. and Sun Microsystems Inc.made appeals for the OpenAccess Community, an Avanti Corp. executive said his company wouldn't join, and bluntly predicted its failure.

The user-backed OpenAccess Community effort has proposed a standard applications programming interface (API) and data model based on the Genesis database of Cadence Design Systems Inc. But Avanti and Synopsys Inc., which are due to merge this spring, have resisted participation. One issue has been the unavailability of Genesis source code, but the deeper issue appears to be a conflict between Avanti's Milkyway database as a de facto standard and Genesis as a new, openly available standard.

"The only thing that's user-driven is Milkyway," said Noel Strader, head of corporate product management at Avanti. "Genesis is a paper database with no tools." Strader said that OpenAccess "is committee-driven, not user-driven, and I think it's going to fail."

Avanti has opened Milkyway to customers and selected third-party EDA vendors, but will not provide it on an open-source basis, Strader said. "Nobody in OpenAccess believes that [Avanti] approach is truly open," retorted Greg Spirakis, vice president of the architecture group at Intel.

Spirakis said that OpenAccess strikes a "balance" between a committee-driven standard and a vendor-controlled, proprietary standard. He emphasized that multiple databases could potentially be implemented under the OpenAccess API, and called for Avanti to join the initiative.

Support for the OpenAccess approach also came from Sunil Joshi, vice president of design automation at Sun Microsystems, who stressed the need to go beyond file-format translations. And David Lan, manager of design automation for TSMC North America, said his company needs an open standard in order to work with customers using multiple tools.

Skepticism was expressed by Joe Hutt, vice president for technical sales at Magma Design Automation Inc., who said that too little is known about next-generation process technologies to develop a standard. And Pallab Chatterjee, president of design services firm SiliconMap LLC, said the real problem is a lack of design methodologies, not lack of a design data standard.

Point of agreement

The one point that panelists seemed to agree upon was the need for an integrated database. Spirakis said a "common run-time data platform" is essential to lower integration costs, and speed time-to-market. "It's not our job to put CAD tools together. It's our job to put products out that people want to buy," he said.

"I don't think we can survive without something like [OpenAccess]," said Joshi. "The days of file formats are over."

Lan said that multiple vendors, tools and databases cause a "nightmare" for TSMC engineers who must come up with technology files for customers. "Every time we put together a new RTL-to-GDSII flow, we spend up to 20 percent of our development time on interoperability," he said.

Strader said that an integrated database is needed for design closure, and he presented Milkyway as the obvious choice. "There are good standards, but they all tend to be de facto standards," he said. "Today, thousands of [Milkyway] users are using a database that's open to them."

Hutt noted that a good deal of information beyond the scope of OpenAccess remains proprietary, such as foundry design rules, company-specific design data, internal CAD tools, and EDA vendor data models and interfaces. For a standard to really work, he said, large users such as Intel and IBM will have to open up proprietary data, as will EDA vendors. "I don't believe that's going to happen," he said.

Hutt also said the industry's rate of change argues against the viability of a standard. "We don't know how to design 70-nanometer technology," he said. "Until that stabilizes, I don't see that interoperability standards will be stable."

"When technology slows down it will be too late," responded Joshi. "The time to do something is right now."

Chatterjee said that, in his experience, interoperability is not the "gating item" that stands in the way of tapeouts. The real problem, he said, is "the lack of a design methodology behind the tools." And OpenAccess is not promoting a methodology to go along with its call for a standard data model, he said.

Asked if their companies will refuse to buy tools that don't conform to the OpenAccess API, both Spirakis and Joshi gave qualified responses. "It's not a checklist item, but I don't think we'll be able to solve our technology problems without interoperability," Spirakis said. "And what drives our buying decisions is whether tools will solve our technology problems."

"With a common data format, time-to-market is so much faster," said Joshi. "I think performance gains are so significant that we'll automatically migrate in that direction."

? Richard Goering

EE Times

Article Comments - Interoperability: a sore point for t...
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top