Between The Lines

EDA veteran Eric Filseth has joined next-generation pcell language company Ciranova as its new CEO after spending many years at Cadence Design Systems, which ironically has been the bane of Ciranova’s existence (and vice versa). I spoke with Filseth briefly today and the first thing he asked is if I think he is idealistic or insane (or both)? I guess that remains to be seen.

Filseth had various stints at Cadence and started working for Cadence in the late ‘80s after spending his just-out-of-college-years as an analog designer at National Semiconductor (lots of early Cadence folks came from National including Joe Costello and Jim Solomon) and couple of startups. In his first stint at Cadence, Filseth was at one time in charge of marketing for the analog tools group but then in the mid '90s got the entrepreneurial itch and left to work for power tools startup Sente and later Silicon Perspective. Both companies were acquired, the latter by Cadence in 2001. The SPC First Encounter tool pulled Cadence out of the frying pan and out of the fire in the digital implementation tool space, after Synopsys took the lead in that tool area thanks in large part to its acquisition of Avanti Corp. Indeed, the Encounter tool was such a success that many Cadence tools today carry the “Encounter” name. “I’m very proud of what we did with Encounter,” says Filseth. “We helped Cadence rebuild and now I think it is one of the best if not the best tools on the market.” Filseth stayed on at Cadence as the vice president of marketing for its implementation tools and DFM division until last month.

Filseth said with mission accomplished on the Encounter front “it was time to move on to new challenges.”

Filseth’s choice of employer, Ciranova, certainly seems like a big challenge, as Ciranova is essentially trying to help the rest of the EDA community break into Cadence’s seemingly heavily guarded analog tool shed. Cadence has held dominant marketshare in analog layout with its Virtuoso tool for almost two decades and indeed a Cadence executive once reported that the company draws 1/3 of its revenue yearly from analog tool sales (new and renewals). Much of that dominance has been largely due to Cadence owning a key format called SKILL for describing analog library elements called PCells. No layout editor other than Cadence Virtuoso can effectively read and open SKILL based PCELL data. Because SKILL for PCell development is a proprietary and patent protected language, Virtuoso has remained the dominant analog layout tool. This dominance has been aided by foundries, which tend to develop process design kits in SKILL for their new IC processes long before they develop them in other languages.

When Cadence helped found the Open Access common database effort for tool interoperability (largely digital tool interoperability to create an open alternative to Avanti’s Milkyway database), Ciranova, backed by Cadence founder Jim Solomon, saw an opportunity to replace SKILL language with a more modern Python-based language for describing PCells. The company started giving away a tool for Pycell development. Then last year, Ciranova came up with an inventive way to use OpenAccess’ caching mechanism to allow other tools to access legacy Skill Pcell data. That Ciranova tool, PCell Xtreme, essentially reads skill data from Open Access cache and translates that data so any other OpenAccess compliant layout tool can read legacy pcell data.

Thus, Ciranova seemingly poses a big threat to Cadence’s analog layout cash cow. But it’s not like Cadence to standby and watch it happen, as Cadence in the past has shown itself to be very protective of this market and even sued Cooper and Chyan Technologies and failing that acquired it after CCT found a way to crack and read skill PCells. Indeed, borrowing a saying from Intel, “only the paranoid survive,” there’s been rumblings in the EDA space that Cadence has been working on a way to kill the Ciranova cache mechanism either in the newest version of Virtuoso, version 6.2, or perhaps even seemingly covertly via changes to the OpenAccess software (Cadence engineers do the majority of OA maintenance)—failing that Cadence could always go CCT and buy Ciranova out.

Filseth says he comes to Ciranova with no inside knowledge of Cadence’s plans in the analog arena (because in his most recent post he wasn’t in that division) and downplays any past turbulence between Ciranova and Cadence. He said he believes that Cadence as well as the rest of the EDA industry has much to gain from having open, interoperable pcells and believes it will open up new avenues of analog tool automation and thus EDA growth, for Cadence too. Filseth says analog designers can benefit from many levels of automation. “The analog EDA and analog EDA layout market should be a lot bigger than it is,” says Filseth. “If you look at the size of the logic design market (ASSP, ASIC standard cell based design market), there’s a fairly substantial market that supports that $100 billion semiconductor segment. The analog market as part of SoC design is between $40 billion and $60 billion a year, yet the analog tools market is only $250 million a year--it aught to be twice that size. I think the reason it isn’t bigger is that we in EDA haven’t provided the value and automation that would enable it to grow to that size. I think the established companies out there have provided a lot of value in providing an environment to do things manually but I think it’s time to take the next step and provide more kinds of value. We shouldn’t be focused on providing another manual layout editor, we should be focusing on ‘what can we bring to the world that doesn’t exist yet?’”

Indeed, it will be interesting to see how Ciranova shakes out under Filseth, who is also taking on the challenge of the CEO role for the first time in his career. He helped build and sell one blockbuster tool company to Cadence at a time when Cadence was desperate for a solution—a task seemingly as difficult as selling a cold bottle of Perrier (a big one) to a millionaire who has been stranded in a dessert for five days. Considering one of Ciranova’s tools is freeware and the other one appears to be at the mercy of Cadence, Ciranova certainly looks to be a bit more challenging for Eric. With the idea of making PCells interoperable, it also appears Eric’s taking up a noble cause. Good luck!