Tool vendors dispute report of downturn in ESL

The recently released EDA Market Trends report from Gartner Dataquest says yes. But ESL providers largely disagree, and their reactions reveal broad differences over how this nascent market should be defined and categorized.

The report was written by Gary Smith, chief EDA analyst at Gartner Dataquest, who has frequently described ESL as an essential technology for those designing 100 million-gate chips with 90-nanometer design rules. Last year's report anticipated strong 2003 growth in three ESL tool categories: design and simulation; test and verification; and behavioral synthesis.

Then came a shock. This year's report says that ESL design and simulation revenues declined by 20 percent to $44.7 million in 2003, while ESL test and verification was down 18.2 percent to $70.8 million. No separate forecast was given for behavioral synthesis, which is "still waiting for the right tools," Dataquest said.

The good news for suppliers is in the report's long-range forecasts, which anticipate ESL design and simulation will grow to $279.3 million in 2008, and ESL test and verification to $296.5 million. But for now, Smith explained, the market is hobbled because power users are still waiting for 65- and 45-nm tools to emerge.

ESL design and simulation is defined as "design at the conceptual level, including hardware/software co-design, design partitioning and specification." The 2003 numbers show Cadence Design Systems Inc. with a 28 percent market share, Synopsys Inc. at 28 percent, CoWare Inc. with 26 percent and others with a total of 18 percent.

ESL test and verification is defined as "tools that introduce the system test requirements into the design flow." Dataquest shows Verisity Design Inc. with 67 percent in 2003, Synopsys with 17 percent, Forte Design Systems with 9 percent and Cadence with 7 percent.

Dataquest recently discovered a new category of "system architects" who primarily use Matlab or Simulink products from The MathWorks. If the revenue for those products were included in its report, Dataquest said that The MathWorks would claim 41 percent of the ESL design and simulation market. But even with that revenue, the ESL market would have grown only 1.8 percent in 2003, the report noted.

Categorical disagreement

Mentor Graphics Corp. is making a major effort in ESL, but wasn't even mentioned in the 2003 market share numbers from Dataquest. Serge Leef, general manager of the system-on-chip verification division, said Mentor sees ESL as including six distinct categories of tools: platform-based design, transaction-level modeling, performance optimization, hardware/software co-verification, C-based simulation and C-based synthesis. He questioned whether there should be an "ESL test and verification" market as described by Dataquest. Leef said that 2003 revenues were "mixed" for the tools that fall into Mentor's six ESL categories, with a "slight increase" in 2004.

ESL revenues were "on track" for Cadence Design Systems in 2003, said Mitch Weaver, vice president and manager for system and functional verification. Cadence believes the market did not decline, "but rather fragmented a bit more than the way the market is defined by the Gartner report," he said.

Synopsys profoundly disagrees with Dataquest's ESL outlook. Farhad Hayat, vice president of marketing for Synopsys' verification group, said Dataquest's market share data in this segment is "inaccurate" and that the "market segmentation is wrong." Further, he said, Dataquest is missing an important trend toward a unified functional-verification platform.

"Synopsys owned approximately one-third of the testbench automation market in 2003," Hayat said. "We believe Dataquest makes apples-to-oranges comparisons by excluding verification IP [intellectual property] and related services for all players with the exception of Verisity."

Verisity, which, according to the Market Trends report, increased its share in 2003, believes there was actually growth in ESL test and verification last year. Steve Glaser, senior vice president of worldwide marketing, said that Verisity's licenses grew from 10,000 in 2002 to 13,000 in 2003, but that the company's revenue remained relatively flat because of a shift to time-based licenses. He also said that "transition issues associated with the shift to SystemC" may have dampened the ESL design market in 2003.

Mark Milligan, vice president of marketing at CoWare, also saw 2003 as "a year of transition" as the industry moved to SystemC. "CoWare did well during 2003, but license revenue specifically had modest growth during this product transition to open standards," he said. CoWare expects rapid growth in 2004 and beyond, he added.

Forte's ESL revenue grew during 2003 and is enjoying a major jump in 2004 after the release of the company's SystemC synthesizer, said Brett Cline, vice president of marketing at Forte. And Summit Design Inc. had a "good year" with positive cash flow in 2003, said president and CEO Emil Girczyc.

Celoxica Ltd. believes its 2003 revenue should have placed it as one of the top ESL companies in the Dataquest market share charts, said vice president of marketing Jeff Jussel. But the company wasn't adequately counted, he said, because many of its customers are "systems designers" who are outside of the traditional EDA customer base. Perhaps these customers are among the "system architects" cited in the Dataquest report, but not counted in the 2003 ESL revenue numbers. With the inclusion of that group in next year's report, along with revenues from The MathWorks, the market share numbers could look very different.