Designed from the ground up as a new high-level design language, Superlog provides an alternative to the recent C/C++ language proposals from the likes of Synopsys, CoWare and CynApps.

Co-Design will announce its initial backers in the Superlog 2000 Partners Program (S2K) today (Nov. 22). The S2K companies are chartered with promoting the use of Superlog and overseeing development of the language and its placement in the public domain.

The partners are Arexsys, Denali Software, Expressive Systems, Interra, Magma Design Automation, Get2Chips (formerly Meropa), Novas Software, Sente, Silicon Forest, Summit Design, Verilog simulation startup Tharas Systems and Viewlogic Systems.

Co-Design claims Superlog can support the specification, logical design, implementation and verification of digital designs by combining the best features of Verilog and C/C++ while allowing interfaces to, and translation to and from, C/C++, Verilog and VHDL. The 12 companies are taking key positions in a hypothetical design flow based on Superlog, although none of them, including Co-Design, has yet announced tools that work with designs expressed in the language.

"All [the S2K companies] are committed to produce, or to adapt, tools to make use of Superlog," said Simon Davidmann, president and chief executive officer of Co-Design, based here.

Davidmann said the S2K companies are expected to release products supporting Superlog during the first half of 2000 and that his own company's Superlog simulator is working well at beta sites and will ship commercially at the end of the first quarter. He declined to name any of the beta site customers.

"We wanted to have a Superlog-based design flow that would be of greater benefit than the sum of its parts," said Davidmann. The companies are said to be applying Superlog in areas from architectural planning and analysis, graphical design entry and manipulation, synthesis, verification, simulation and hardware acceleration of simulation, through to layout, prototyping and chip assembly.

"This industry endorsement from multiple EDA tool providers reinforces our commitment to place Superlog in the public domain during the year 2000," Davidmann said.

But there are many EDA industry participants who say a new language is not the way to go. "I'm hearing pretty well no one asking for a new language," said Pete Hardee, director of product marketing at CoWare. The company, along with many others including Synopsys Inc., is pushing the Open SystemC initiative, which is currently in the throes of rewriting its licensing agreement. "It has to be C or C++-based because that's what the hardware guys are using."

Among the S2K Superlog backers are three companies — Denali, Magma and Viewlogic — that have also signed on to the Open SystemC initiative. Indeed, despite apparent competition, Co-Design too is a member of the SystemC initiative, largely so the company can accept models written using that language.

Jim Griffeth, vice president of marketing at Viewlogic Systems, said his company will likely add support for both Superlog and SystemC to its eArchitect product.

"We need to solve some of the tough issues addressed by the system-level design language proposals," Griffeth said. "We at Viewlogic are supporting that broadly."

"As an open EDA tool provider, we believe we have to work with all the high-level design solutions and we will provide interfaces linking all of these to our Blast Fusion tool," said Rajeev Madhavan, president and chief executive officer of Magma Design Automation. Madhavan personally is an investor in both Co-Design Automation and CynApps, which is fielding its own C++ library proposal.

Sanjay Srivastava, president of memory-modeling tool vendor Denali Software, said his company is endorsing all "legitimate" efforts to develop system-level languages and will create tools and links supporting them as customers demand.

Bernd Braun, chairman, president and chief executive officer of Get2Chips Inc. (San Jose, Calif.), currently in the process of merging with Meropa Inc., said his company would bring Meropa's behavioral-synthesis tool, called Fast, to market together with other tools — combined with options on Internet access and pay-per-use licensing models — with an ability to support Superlog-based design.

"We will be language-independent but we will focus on Superlog and behavioral Verilog," Braun said. "We are working with Simon [Davidmann]'s team to help him define Superlog."

Braun, who was previously a senior vice president at Mentor Graphics Corp. (Wilsonville, Ore.) responsible for intellectual property, emulation, and debugger product lines, said: "We would be prepared to support C language design if it became significant, but right now it's behavioral Verilog and Superlog. As a startup we have to focus."

But the leading three EDA companies in terms of revenue — Synopsys, Cadence Design Systems and Mentor Graphics — are conspicuous by their absence from the S2K list. "We have shared information with Synopsys, Cadence and Mentor but we are not in a position to announce anything at this time," said Davidmann.

Stan Krolikoski, group director for system-level product marketing at Cadence, said his company "applauds" Co-Design's commitment to working through standards bodies. But he said Cadence hasn't had a chance to examine Superlog with enough "due diligence" to endorse it.

Cadence and Mentor, meanwhile, have welcomed the CynLib C++ class library proposal from CynApps Inc. (Santa Clara, Calif.).

Gary Smith, the chief EDA analyst with research concern Dataquest Inc. (San Jose), said the list of EDA companies endorsing Superlog gave the language a "medium amount" of credibility, but that getting tools out to users was more important.

"Meropa has been one of the hot secrets in EDA. They seem to have by far the best behavioral-synthesis tool around, by an order of magnitude or more. Of anybody on that list that company is the important one," said Smith. "I don't believe having Synopsys' support is essential [for Co-Design], although it would be nice."

However, Smith dismissed rival design methods based on C and C++. "SystemC was introduced as a high-level modeling solution. It did not say anything about design," he said. "Trying to use C for design is a Band-Aid on the language problem."

Smith said that if the purpose-written Superlog can support concurrency as well as many other system-level features claimed for it, it has a chance in the market. "The only problem is that it takes a long time, three to six years, to establish a language."

— Additional reporting by Richard Goering and Michael Santarini