Columnist: June 23, 1994

The shape of things to come?

John Cooley,
EDA Consumer Advocate & ESNUG Founder

I couldn't help but think about overly idealistic college activists while listening to Andy Graham, president of the CAD Framework Initiative (CFI), wrap up CFI's first conference on EDA integration and interoperability (EII) with a call for users to pressure EDA vendors to make their products more interoperable with their competitor's. I felt as if I were witnessing some sort of modern-day scene from Don Quixote, where a well-intentioned management-type was expecting normally overworked engineers to suddenly drop what they're doing, spontaneously unite, and assault the buildings where evil, scheming EDA vendors work—all in an attempt to force them to become CFI compliant.

Doesn't Andy know that it's not human nature for people to get worked up over something they already have a solution for? Sure, most EDA tools from different vendors choke a little when you try to get them to work together, but there's always a way to find a workaround. Trying to invoke major political changes with EDA vendors is a slow, messy, and uncertain process—and it doesn't solve that "not all EDIFs are alike" problem for me right now, when I need the solution.

Then my mind wandered into all the political realities involved with having fully interoperable EDA tools. EDA vendors would love it because it offers them one standard framework to design their competitive, hot, and innovative tools that would steal business from the big EDA vendors. Realizing this, big EDA vendors are infamous for providing all sorts of lip service to promoting interoperability and things like CFI standards—but allocate next to nothing in funding to make it happen. Customers would love it because there would be no pesky interoperability issues to occasionally hack through.

Of course, as any salesman will tell you, customers can be very fickle: They aren't willing to pay an extra cent for interoperability, but if it's free they'll take it. When I asked all five CAD managers on the EII panel the question, "Given a choice of either a screamingly new EDA tool that helped you do new functionality or complete interoperability of your current tools, which would you choose?", all five EDA customers unequivocally chose getting the new EDA tool over interoperability.

Zero tolerance

Also, unlike how they feel about tools offering sexy new EDA functionality, customers have zero tolerance for any bugs in a framework. The big EDA vendors who sunk a little money into making their own proprietary frameworks (ie, Mentor with the Falcon Integrator and Cadence with Cadence Frameworks) got burned big time by "displeased" customers who ran into major and minor framework bugs. Most of the customers stopped using the framework and bombarded suppliers with nasty phone calls, which is another valid reason the big EDA vendors give frameworks tons of lip service but not much more.

The final effect an industry-wide interoperability solution would have would be that soup-to-nuts "total-solution" EDA vendors (like Mentor) would be threatened, because EDA tools would become commodity items. (Customer loyalty lasts all of 30 seconds in the EDA business.) EDA vendors would be forced to offer quality tools at dirt-cheap prices because the competition would be cutthroat! On paper, this sounds like a great idea for EDA customers—and a living hell for EDA vendors. Maybe the engineers of the world should unite for CFI! But take this scenario a little further and you'll find that those software pirates known as EDA vendors are not necessarily as evil as they appear to be on this issue.

The future's a funny thing. Although in the short run customers would benefit from commodity-priced EDA tools, who's going to pay for the blockbuster next generation of EDA tools? Sure, a lot of great EDA tools start from two guys in a garage, but what bootstraps real development is the high profit margin these new companies get with their hot tools. For example, Chronologic Simulations couldn't have made a screaming Verilog simulator if they could only get $3000 per copy early on. The same is true for the ambitious projects the big EDA vendors take on: Synopsys couldn't have made a behavioral synthesizer without the financing from sales of its RTL level synthesizer, and Racal-Redac had to sell its behavioral-synthesizer research—along with most of its remaining EDA product line—because it wasn't making money in the EDA business.

No matter how good EDA tools get, there's a limited market for this type of software, and somehow the R&D money for next-generation tools has to come from customers.

For those who doubt my reasoning, look at the commodity PC-clone market compared with the old, big computer makers. Sure, the big computer makers were big, piggy, and financially greedy, but these were the places where a lot of hardware design concepts were conceived and developed. Scan, JTAG, the use of large ASICs, early FPGAs, and lots of EDA tools were financed by sales to companies like HP, DEC, IBM, Sun, Tandem, SGI, Intergraph, etc. Compared with the number of technological breakthroughs the Taiwanese PC-clone makers made in the hardware industry as a whole, you'll see where I'm coming from.

Please don't misunderstand: I'm not advocating ignoring the need for better interoperability among tools from different vendors, but as long as there's some viable workaround, you're not going to hear me complaining (much).