For Increased Productivity, New Design Languages Are Essential to Success

Electronic News, August 27, 2001 by Dennis Brophy

LIKE A FAVORITE SPORTS team, to some people design languages garner an emotional following. And in the competition over which language will win industry support, one of the most visible of contests was between VHDL and Verilog. The languages are now united through the standards organization Accellera, but in other arenas, the one-upsmanship continues.

Debate rages over which language is superior for next-generation design: Verilog or VHDL, C/C or the Open SystemC Initiative, SpecC or Cynlib. Some suggest an evolutionary approach to Verilog is best and that through Co-Design Automation Inc.'s Superlog. But design languages are not enough for next-generation designs. Verification languages also are in the running, as witnessed by Open VERA and the Accellera working group, which has four technology donations in this area.

These debates are born out of the competitive nature of language standardization. Competition is healthy--it drives innovation and ensures interoperability. It also provides the best degree of openness for users.

The truth is that different languages are better for different applications. Designers use C/C for system modeling and software implementation, Verilog and VHDL for hardware design and other languages for verification and test.

Short- and long-term approaches should also be considered. The Accellera working groups are now engaged in extensions to Verilog to give designers added flexibility to do higher-level modeling using the Co-Design Automation donation of their Superlog Extended Synthe-sizable Subset. This, combined with assertion and verification testbench standards, will help designers deliver higher-quality products and serve as the basis for general methodology improvements.

And while any one language will not immediately address true system-level design, long-term work on one unified structure must be in place to link the many application-specific languages.

Accellera's System Level Design Standards group has several subcommittees addressing this. Accellera has the C/C Semantics Working Group, the goal of which is to define the semantic base from which to express design languages. This will lead to an example of a C class library for design, verification and implementation of electronic systems. The open definition will include the best features of all current offerings since the group collaborates with other industry groups.

Once languages become the domain of industry consortia, challenges still exist. It is imperative that participants commit and can donate significant time to see the venture to fruition. The industry must also understand that adoption of new languages will take time as some require shifts to new methodologies and all require vendors to create new tools.

At the EDA: Front-to-Back conference (www.edafronttoback.com) to be held Sept. 24-26 in San Jose, Accellera will host Language Day, focusing not on arguments about design languages, but on moving to the next level.

Accellera will integrate input received during these sessions to help drive the next leap in design productivity. A survey as well as conference information is online at www.edafronttoback.com. Results from the survey will be reported during Language Day. Your input is valuable if the industry is to move to the next level.

Dennis Brophy is director of strategic business development at Model Technology, a Mentor Graphics Corp. company, Portland, Ore. He also is the chairman of Accellera, based in Los Gatos, Calif.