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

Users back OpenAccess standard, but Synopsys refrains

Posted: 14 Jun 2002 ?? ?Print Version ?Bookmark and Share

Keywords:OpenAccess? Design Automation Conference? chip design? EDA?

The OpenAccess design data standardization effort is crucial for next-generation chip designs, according to users at an Interoperability Workshop at the 39th Design Automation Conference Monday (June 10). But Synopsys Inc. is still concerned that the OpenAccess Coalition leading the effort is a "gated community," said Rich Goldman, Synopsys' vice president of strategic market development.

The OpenAccess Coalition, which works under the auspices of the Silicon Integration Initiative (Si2), has made considerable progress since last year's DAC. A design data API based on Cadence Design Systems Inc.'s Genesis database, now called the OpenAccess database, is available free, and Cadence recently donated the source code for OpenAccess version 2 to coalition members. In addition, a 12-member "change team" for overseeing modifications to the database and API is now in place.

But the OpenAccess Coalition only has a handful of EDA vendor members: Cadence, Simplex Solutions, Nassda, Numerical, Synplicity and Verisity, with the latter four joined the initiative only this week. Synopsys, in particular, has been vocal in its objections to what it views as excessive control by Cadence.

User support

At the Monday workshop, representatives of six powerful user companies?Ericsson, IBM, Intel, LSI Logic, Motorola and STMicroelectronics?emphasized the need for a standard design data API and called on the EDA community to line up behind OpenAccess. All said that interoperability in shared memory is far preferable to today's file format translations.

"File-based systems are dead?we just haven't stuck a fork in them yet," said Greg Spirakis, general manager for design technology at Intel Corp. Tightly integrated systems are the wave of the future, he said, and the only question is whether there will be an open standard or several closed, proprietary systems.

OpenAccess, said Spirakis, is "the only open alternative that's out there." Otherwise, he said, the EDA industry could create a scenario similar to Unix, where vendors fought among themselves, resulting in chaos.

Dale Hoffman, director of EDA at IBM Corp., said that concerns such as area, performance, power, noise and yield are driving the need for interoperability at the data model level. "What's really important is tight integration between tools, and an ability to do incremental processing," he said, noting that IBM plans to use the OpenAccess API.

In a pointed warning to EDA vendors, Hoffman said he has often declined to buy commercial tools because the cost of integrating them into IBM's design environment would have been too high, even in cases where the tools were clearly better than those IBM was already using.

"OpenAccess is just the right thing for us. It allows us to work more easily with our partners outside," said Sumit Dasgupta, director of system-on-chip and IP design systems at Motorola Inc. "We are setting a priority to move to the new API." Dasgupta questioned, however, whether one data model can serve all design needs.

Sudhakar Sabada, vice president of design technology development for LSI Logic Corp., spoke of the problems with today's file-based systems, including inefficient file transfers, huge application startup overhead, lack of scalability and inconsistencies in name mapping. OpenAccess, he said, will enable startups and foster competition. "Interoperability means the best-in-class tools win," he said.

STMicroelectronics has successfully been using its own APIs, but no longer has time to develop interfaces, said Jean-Pierre Geronomi, the company's director of CAD. He said the OpenAccess API needs to be supported by all EDA vendors, needs to provide sufficient run-times, and should be kept up-to-date as technology changes.

Qualified support

Representatives of Cadence, Nassda, Numerical and Verisity all stated their support for OpenAccess as well. The only critique of OpenAccess came from Synopsys' Goldman, who made it clear that Synopsys doesn't oppose the idea behind OpenAccess.

"Our industry needs an open API," he said. "This is too important to get it wrong. We've got to get it right." He then asked how many audience members have read the contract agreement between Si2 and Cadence. As it turned out, very few had.

Goldman noted that non-coalition members don't get the OpenAccess source code until sometime next year. He complained that Cadence has one of two "system architect" positions on the change team "forever," and that no other EDA vendor can ever hold this position. He also noted that Cadence has veto power over changes to OpenAccess until July 2004.

Goldman also said that while Cadence is not assigning any patent rights, anyone else who contributes to OpenAccess will have to do so. Thus, he said, other vendors may be opening themselves up to patent litigation. He also said Cadence can create its own "fork" of OpenAccess, while others cannot.

In the ensuing discussion, Don Cottrell, Si2's vice president of technology, said the coalition's system architects don't hold any special power, and noted that Cadence is granting royalty-free rights to use any code that embodies patents, and said that Cadence does not have the right to "fork" the code. Goldman expressed some relief, but remained concerned about the patent situation. He said that Synopsys might join OpenAccess if there is sufficient "momentum" towards solving the problems he cited.

Goldman also said that Synopsys has made no decisions about how to open the Milkyway database it has just acquired in its acquisition of Avanti Corp., and said that "every option is open." One option, presumably, is donating Milkyway to OpenAccess.

Charlie Huang, vice president of marketing at Cadence, said that OpenAccess is really standardizing an API, and multiple databases could operate underneath it.

"An alternative implementation would be very viable and very welcome," he said, without citing Milkyway by name.

Francine Ferguson, vice president of marketing at Verisity Design Inc., said her company would like to help extend OpenAccess?which currently addresses physical design?to verification. "Many tools make up the verification flow, and most link in programmatically," Ferguson said. "If you have an interoperability issue, it often becomes a show-stopper for verification."

Two are better than none

Even without Synopsys' participation, Intel's Spirakis said the coalition's efforts are better than not implementing the standard. "Having two solutions, OpenAccess vs. the one from Synopsys, is better than having 25 different solutions," he said. "Just because Synopsys is not joining now, doesn't mean they won't join later when their concerns are addressed."

Motorola's Dasgupta concurred: "We've been at it for seven years trying to arrive at an API standard, and we have done more progress in the last year than in the first six," he said. "It's an amazing progress and we need to keep going."

? Richard Goering

EE Times

With additional reporting by Nicolas Mokhoff.

Article Comments - Users back OpenAccess standard, but ...
*? 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