Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > Manufacturing/Packaging

Coalition regroups in design-to-mask bid

Posted: 29 Nov 2005 ?? ?Print Version ?Bookmark and Share

Keywords:Silicon Integration Initiative? design-to-mask? coalition? universal data model? IC design?

The Silicon Integration Initiative launched a design-to-mask coalition last year with a grand visiona "universal data model" that would extend from IC design through manufacturing. But the challenge proved harder than it looked, and is now taking the coalition into a second, more pragmatic, phase.

The goal has not changedthe Silicon Integration Initiative (Si2) still believes in the concept of a universal data model. But now the focus is shifting from anecdotal beliefs to facts that can be confirmed in benchmarks, thus building a stronger case for what could be a highly disruptive technology for the semiconductor industry.

Roadblocks in today's design-to-mask data flow are widely acknowledged and may be costing the industry billions of dollars annually, some observers say. An OpenAccess-based universal data model and associated application programming interface could, in theory, bring design intent into the manufacturing world, and transport manufacturing and process data back to designers.

But the coalition's first phase made it clear that such a far-reaching idea needs a strong foundation. "We went too fast," said Si2 president Steve Schulz. "We went too far to a conclusion that the industry was not yet ready for. What they're crying out for is to see more data and to find out how much value it actually offers."

The coalition's early work was a "learning phase," said Don Cottrell, fellow of emerging technologies at Si2. "We came to realize there were a lot of anecdotal requests for better communication, but little proven information. There were disagreements between various players over what the important elements are."

"Most importantly, a majority of players that are affected by DFM [design-for-manufacturability] are still jockeying for position in this game," Cottrell said. "We could get very good resonance on what the long-term goals should be, but there seemed to be little agreement on the schedule or tactics to get there, and people didn't know what they wanted to open up to the world at large."

No tangible deliverables came out of that first phase, Cottrell said; a proposed engineering plan was developed, but never published. Companies involved included Applied Materials, Cadence Design Systems, DuPont Photomasks, Freescale, LSI Logic, Sagantec and Silicon Navigator.

One problem, said Cottrell, was insufficient EDA vendor involvement in phase one. He said that six EDA companies will be involved in phase two, including Cadence and Sagantec, but declined to name the others. One, apparently, is Ponte Solutions, which announced Nov. 17 that it has joined the coalition and has licensed its yield-modeling platform to Si2.

Phase two begins in January and will focus on "real metrics and real facts," Cottrell said. In this phase, the coalition plans to run benchmarks, capture information requirements from successful benchmarks, formalize the information model, develop a test case library through member contributions, and offer events and reports that help raise industry awareness of design-for-manufacturability issues.

Several key areas are on the table for phase two, Cottrell said. One involves looking at yield and lithography models, and determining how to communicate these models upstream into the design flow. Another will seek ways to pass more specifics about design intent into back-end processes such as optical proximity correction. For example, Cottrell said, information about slack times on critical nets would help determine how complete optical proximity correction needs to be.

A third area seeks to speed mask inspection and repair by giving the mask house more design information, such as which observed faults will actually cause failures in the final wafer.

While no timetable has been set, Cottrell said, Si2 hopes to deliver benchmark results "in a very short amount of time," meaning next year. But the creation and adoption of a universal data model, which will require extensions to OpenAccess, will be a more incremental process, he said. Adoption could be relatively fast in design environments that already use OpenAccess, he said, but slower for mask shops and fabs that do not.

"It could be a disruptive technology that takes some time for the industry to adapt to," said Schulz.

"If benchmarks show demonstrable value, the actual adoption of technology extensions to OpenAccess could probably happen in a year's time frame. We would probably have early adopters a year after, and mainstream adopters a year or two after that."

- Richard Goering
EE Times

Article Comments - Coalition regroups in design-to-mask...
*? 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