There are three kinds of lock-in in life: assisted living, high-end automobiles, and EDA software. The assisted living folks have got you the minute you move an elderly parent into their facility. It's so hard moving the parent in, you'll never force the parent to move out again, no matter how much the facility raises the monthly fees. That's lock-in.
High-end auto manufacturers have got you the minute you reach that rung on the corporate ladder requiring you to drive a Mercedes, BMW, or Lexus; your status metric says you'll never get caught behind the wheel of a Ford Festiva again, no matter how much the high-end manufacturers gouge you to play in their league. That's lock-in.
The EDA software vendors have got you the minute you decide to design a chip. You need their tools, even though they're proprietary, difficult to use, and costly to maintain. The EDA vendors will continue to feed your addiction and you'll continue to buy, no matter the level of cost or disappointment, because there is no other option. Until now.
Today, through a nexus of the open source software movement and the difficulty of developing tools to address the problems of leading-edge designs, there are EDA tools out in cyberspace that you can download for free, customize, and utilize to create your latest designs.
Or so I thought, until I spoke with Chitlesh Goorah, who quickly disabused me of the Utopian nonsense in the previous paragraph. Goorah manages the online Fedora Electronic Lab [FEL], and says his organization is not in competition with Synopsys, Cadence, Mentor Graphics, or anybody else. Instead the free, open source EDA tools available on the Red Hat-sponsored FEL website "complement" the tools from the major EDA software providers (even as they span the range of applications from synthesis and layout, to simulation and verification).
Goorah insists, "You have to distinguish between open source software and open source EDA software. There's a big difference. For open source software, you just need to create an application, but for open source EDA software you also need to have a scientific background, need to understand physical design, and have to have some mathematical algorithms to code. In the case of a synthesis tool, for instance, you have to be able to minimize the gate count while also improving the timing closure. Cadence or Synopsys spend a lot of money on research developing these types of features, a [level of resource] completely lacking in the open source EDA community."
There are two other important issues as well, according to Goorah, preventing widespread progress in open source EDA software today: marketing and mindset. He says, "The open source software community, driven by IBM and Red Hat for instance, is successful because they want to satisfy their customers and partners. As far as the open source EDA software community is concerned, however, we're completely lacking that kind of support. At FEL, we're trying to find new ways to market open source tools such that the EDA industry is convinced it can benefit and learn by example from us, but the process is difficult."
The mindset issue is even thornier, Goorah says, because it relates to the developers themselves: "The EDA vendor community understands today that interoperability is very important, but this is frequently a weakness in the open source community. Open source EDA developers often try to invent their own file formats. Very few of them use industry standard formats to avoid having to start from scratch. Through FEL, we are trying to change that, however, to convince developers to work so they can interchange their data easily. If we can share data between the different applications developed by the open source EDA community and the vendor tools, it will be a way to encourage tool users and designers to use and contribute to the open source tools."
Goorah notes that the increasing availability of various open source formats from the commercial EDA vendors contributes to the necessity of convincing open source software developers to conform to these useful industry standards. For now, however, he may have his work cut out for him, that proverbial 'herding cats' thing.
Given these various issues, and despite the sincerity of the effort going into the Fedora Electronic Lab, it may appear to some that open source EDA software poses no real threat to the market share or domination of the big industry vendors. Nonetheless, it's intriguing to contemplate what the future may hold should a different model of collaboration emerge.
If a shared-code mechanism evolves between the EDA vendors and the open source EDA developers, there may come a time someday when open source-like solutions are applied to problems in electronic design as effectively as open source solutions today tackle problems as diverse as accounting, graphics, and office productivity.
With the release of Fedora 11 on June 9th, and all that it represents, perhaps that Utopian vision of open source EDA is not so crazy after all. In reality, lock-in is never really a forever thing (except in assisted living and high-end automobiles, of course).
Peggy Aycinena is editor of EDA Confidential, and a contributing editor to EDA Weekly and the DACeZine.