February 26, 2007
EDA Rising - Cool, Calm, & Collected
Please note that contributed articles, blog entries, and comments posted on EDACafe.com are the views and opinion of the author and do not necessarily represent the views and opinions of the management and staff of Internet Business Systems and its subsidiary web-sites.
| by Peggy Aycinena - Contributing Editor
Posted anew every four weeks or so, the EDA WEEKLY delivers to its readers information concerning the latest happenings in the EDA industry, covering vendors, products, finances and new developments. Frequently, feature articles on selected public or private EDA companies are presented. Brought to you by EDACafe.com. If we miss a story or subject that you feel deserves to be included, or you just want to suggest a future topic, please contact us! Questions? Feedback? Click here. Thank you!
It was a day like no other day in EDA. If you had 12 non-stop hours to spend at DVCon 2007 and the annual EDAC CEO Forecast event, both in Silicon Valley on February 22nd, I think you would have seen what many others saw over the course of the day and evening. The EDA industry is, at last, showing signs of maturing into adulthood.
Between Richard Goering's lunchtime panel on the emerging power standard(s), John Cooley's eclectic afternoon panel showcasing the noisy (not) show-stopping issues in EDA, and Jay Vleeschower's evening panel of EDAC BigWigs, the message was clear:
The EDA industry can indeed get along, resolve its differences, move forward to contribute even more to the success of the larger semiconductor industry, and finally, finally grow up.
Why now? Perhaps it's because people in EDA are making money again. Industry growth is (possibly) moving back into the double-digit range after 6 long, frustrating years. Perhaps it's because the world won't wait for Provincial EDA to become Profound EDA. Customers are threatening to solve the problems themselves if EDA can't step up to the challenge. Perhaps it's because the litigious Party of the First Part and the equally litigious Party of the Second Part seem to see the light at the end of their highly publicized legal tunnel. Perhaps it's because the technologist-turned-CEO of the top-earning company in the industry appears to have hit his stride here in his third year at the helm.
Perhaps it's as simple as the anchor tenant in the industry turning 21.
But for whatever the reason and with the caveat that not all will agree, I would suggest that if the EDA industry can take advantage of this unique moment in time, it's poised to move ahead, to expand into larger markets, and to assume a position of greater influence within the semiconductor industry. Most importantly, it's ready to shed the mantel of martyrdom it has worn so well for these last few years of adolescence. The EDA industry is ready to stop feeling sorry for itself, and ready to start saying to the world - we want to be your partner and share in your both your wins and your losses.
It doesn't get much more adult than that.
EETimes Richard Goering quipped that the panel he moderated at DVCon took the term “power lunch” to its ultimate definition. With 5 panelists from EDA and 1 panelist from the user community (Synopsys' Mike Keating, Mentor's Dennis Brophy, Cadence's Pankaj Mayor, ArchPro's Srikanth Jadcherla, ChipVision's Thomas Blaesi, and LSI Logic's Gary Delp), Goering walked the group through a thorny discussion of the current impasse in the world of power standards, an impasse that Goering himself has documented extensively over the last few months.
There are two standards currently under consideration - the Common Power Format (CPF), championed by Cadence and Si2, and the Unified Power Format (UPF), championed by Mentor, Synopsys, Magma, and Accellera. Accellera released UPF 1.0 on the very day of the Goering DVCon panel. Si2 released CPF 1.0 a number of weeks ago. The Low Power Coalition (LPC) was founded by Si2 and says it's examining both. If you're not a member of LPC, you can't look at a full release of CPF. If you're mad, or like UPF better, you don't want to anyway. If you don't like alphabet soup, you shouldn't even be reading this.
It's a mess, but a temporary mess, because there's too much money and customer goodwill at stake to leave this hanging out there. At least, that's the take-away I garnered from Goering's comment before opening up the mic to the 6 panelists: “Let's not look back, but look forward at what we should do now.”
EDA customer Delp was first up, stated the obvious, and ended with a plea: “Standards must identify and manage tasks, and a single power standard will allow interoperability. So let's avoid alike in every intent, and different in every detail.“ Delp is an active member of LPC, so it's not clear if he's already made up his mind, but I'm guessing he and the user community will have significant influence on the final decision.
The next 3 speakers were EDA guys and all members of LPC - Chip Vision, ArchPro, and Cadence. They all basically said, my company is very special, standards are next to godliness, and may the best man win. Well, actually, Cadence's Mayor said CPF should win: “CPF was developed at Cadence over the last 2 years. We went through extensive user verification. Convergence is of interest to Cadence and we're looking to Si2 to help with that convergence.”
The next 2 EDA guys are not members of LPC, haven't had full access to CPF, and are pretty annoyed. When pressed, they said you have to pay to join LPC and they don't think they should have to 'pay to view' CPF. Cadence offered to take up a collection if the companies were short on cash. Everybody laughed and then Synopsys' Keating said, “UPF and CPF drive the same conceptual model... [so] it will be interesting to see how this plays out.”
Mentor's Brophy pointed out that if the brick & mortar industry can have a single standard for bricks, it's a no-brainer that the EDA industry can have a single standard for power. He said Accellera's already merged 8 of the 9 known power formats into UPF 1.0, so let's get on with the obvious final step and get CPF merged in there as well.
The Magma guy sitting at my table during the panel said to me afterwards, “And don't forget to mention that Magma supports Accellera and UPF.” Done.
So, what do you all think? Impasse or tempest in a teapot? I'd say tempest in teapot, for 2 reasons. Firstly, everybody assured me after the panel that this is going to get resolved, that it's not going to be a repeat of the VHDL/Verilog brouhaha. And secondly, the body language at the EDAC CEO panel Thursday night indicated that the BigWigs in EDA are buds now and nobody's going to wreck that chemistry by allowing the guys swimming down in the alphabet soup to be duking it out for long.
At least for now.
It's not a surprise that great jazz was playing in the background during cocktails at the annual EDAC CEO Forecast Panel held at the Santa Clara TechMart on the evening of February 22nd. After all, Aart de Geus is Chairman of EDAC and an accomplished jazz musician. It was a surprise, however, that the event was on the same evening as DVCon. Any folks who needed to be manning the last hour of the DVCon Exhibit Hall floor over at the DoubleTree in San Jose couldn't be in two places as once. Oh well.
There are 110+ companies in EDAC. Five of them fielded CEOs for the EDAC panel this year including, not surprisingly, Synopsys' Aart de Geus, Cadence's Mike Fister, and Mentor Graphics' Wally Rhines. CoWare's Alan Naumann and Novas' Scott Sandler completed the roster.
Both during cocktails and throughout the panel, it was clear that a new chemistry and maturity have developed within the leadership of EDAC. They're courteous to each other, listening to each other, and clearly committed to the serious business of presenting an intelligent and unified roadmap both to the larger semiconductor industry and to The Street. As I mentioned earlier, it undoubtedly helps that they're all making money, but there's more to it than just that. A new era of steely-eyed calm has clearly arrived. “It's time to get down to the business of business” seems to be the new message du jour from EDAC.
Following a welcome from HP's Norm Reini, Aart de Geus offered opening comments: “EDAC's ambition is to add positive energy around what [the EDA industry] has been trying to accomplish. Bob Gartner is now in place as Executive Director, and our EDAC board meeting today was a very positive one because of the work of the entire team to move things forward. We all want to accomplish things together. Fundamentally this is an intellectual industry and therefore contributions from everyone are essential.”
Asking for a minute of silence to remember the late Dr. Richard Newton, de Geus noted that a number of EDA companies and private individuals, as well as EDAC itself, have teamed up to fund a chair in synthetic biology at U.C. Berkeley in honor of Newton's passion for this emerging technology. In addition, de Geus said, “Richard Newton will be missed as the voice of the Kaufman Award. The award is proceeding nonetheless, and the IEEE Council on EDA, CEDA, has now decided to team up with EDAC as co-sponsor. The award is based on your nominations, so please submit names.”
Following de Geus' comments, Merrill Lynch's Jay Vleeschower took over. He said that in the 2 years since his last appearance as panel moderator, a surprising number of things have happened in EDA, not the least being that the industry has become noticeably profitable with 17-percent profit margins. Although, per Vleeschower, other software industries do even better, the recent improvements in EDA revenues are welcome news. He also noted that Cadence, Synopsys, Mentor, and Magma now represent 80 percent of the revenue for the industry.
Each CEO then had a shot at the podium and in turn, celebrated the New Golden Age of Cooperation in EDA, admonished the industry to pull together to reach even greater heights, touted their own companies, or in the case of Dr. Rhines, went a long ways toward earning an honorary doctorate in economics with a complex presentation of data and graphs that correlated cycles in the semiconductor industry with cycles in EDA. Assuming his regression coefficients and leading/lagging indicators are correct, Rhines' conservative estimate puts EDA on a 7-percent AGR for the coming year. But, he warned, the reality may be prove to be even better. Maybe even into double digits by the time the real
numbers are in for the year, and therefore drinks are on the house.
Speaking to both the technology and the business of EDA, the panel seemed to endorse ESL, DFM, enterprise-wide viewpoints, the value of vendor-customer relationships, consumer electronics, engineers, innovation, small companies, big companies, new problems, new solutions, evolving business models, and the ability of everybody's favorite Wall Street analyst to advance PowerPoint presentations.
But lest I be perceived as making light of anything that transpired at the EDAC event, I will reiterate - the evening clearly illustrated that the tenor of the discussion within the organization has changed. These men seem to have realized that they have little to gain, but a great deal to lose, if they don't finally speak with one voice. They didn't articulate it in as many words, but the body language spoke loud and clear. Within EDA, co-opetition is back.
Within EDA, Real Men make money and lead the way.
That brings us to John Cooley's afternoon panel at DVCon. Cooley had 8 participants this year on the DVCon Troublemaker's Panel (formerly known as BigWigs), and in the brief span of 90 minutes managed (once again) to showcase the seamiest and sorriest aspects of EDA, albeit while also (once again) delivering a measurable dollop of well-received entertainment. Given my earlier thesis that EDA has come of age, I will predict that this is the last time this panel/circus actually unfolds before a live audience. Hence this may be last time I transcribe the script for this event (see below).
If the EDA industry wants to be perceived as an adult within the Semiconductor Brotherhood, and I believe it truly does, it can no longer afford the luxury of an event which portrays the industry as a gaggle of bickering jokesters.
Having said that, let me be clear about one thing. The people who participated on the Troublemaker's Panel are all technologists with admirable track records. It's not that they themselves are culpable for portraying the industry as less than the sum of its parts (some of them probably didn't want to be there at all), it's that the EDA industry as a whole has encouraged this kind of self-effacing rough-housing as a way of building camaraderie with the user community. Well, it's true - belittling yourself is one way to endear yourself to folks. But it's not the only way.
The EDA industry wants/needs to wrap itself in dignity, wisdom, and a sense of visionary technology that will help, not hinder, the globalized and highly profitable/complex semiconductor industry. Everything that the Troublemaker's Panel highlights is counter to that. Hence, the joke is over and the jig is up. Going forward, the industry can no longer afford to laugh at itself in this way. It can no longer hold up its internecine struggles as the norm that defines the industry.
Real Men make money and lead the way. Period.
The DVCon Troublemaker's Panel
John Cooley, DeepChip & EETimes
Rajeev Madhavan, CEO at Magma
John Chilton, Senior VP for Marketing & Strategic Development at Synopsys
Jue-Hsien Chern, VP of Analog/Mixed-Signal Design at Mentor Graphics
Ted Vucurevich, CTO at Cadence
Vic Kulkarni, CEO at Sequence Design
Atul Sharan, CEO of Clear Shape
Brett Cline, VP of Marketing at Forte Design Automation
Gary Smith, Chief Analysts at Gary Smith EDA
The crowd was thick as thieves and upwards of 300 strong at the annual John Cooley DVCon Panel at the DoubleTree Hotel in San Jose on a rainy February afternoon. Everything beforehand was hey-how-are-you and cheerful banter and milling about. Of course, the only one without a tie was the moderator himself, the self-styled Bad Boy of EDA. Meanwhile, the thing was being filmed up the yin-yang, huge lights, cameras, action - a real ripsnorter of an industry event. We all knew the panelists had seen the questions beforehand posted on DeepChip. We all knew they'd been coached on how to respond. We didn't know which questions Cooley had decided to use.
At 3:30, Cooley gestured to the 8 panelists chatting en masse below the stage and said, “I would say, 'Gentlemen, it's time to begin,' but that phrase doesn't apply. So I'll just say, 'EDA People, please come up onto the stage. It's time to begin.'”
The panelists took their seats at the table, the crowd hushed up, and Cooley began.
Cooley - First rule of Fight Club. Don't talk about Fight Club. Second rule of Fight Club. Don't talk about Fight Club. This used to be the BigWig Panel, but it was decided it would be better to tell the truth, so we changed the name.
Cooley introduced his panelists one by one and lobbed his first question at Jue-Hsien Chern from Mentor.
Cooley - Jue-Hsien, why are you here instead of Joe Sawicki?
Chern - Joe came down with a cold and had to cancel. I had no excuse. I had to be here.
Cooley - Thanks a lot, we're flattered.
Cooley - John Chilton of Synopsys, so why are you here?
Chilton - I'm here instead of Antun Domic.
Cooley - Actually they slipped him in at the last minute and I'm okay with that.
Chilton - I became head marketing guy after running the IP business at Synopsys. I know the chip problems. And Antun's done this panel for two year, so it was somebody else's turn.
Cooley - John, why should I trust you? You're marketing.
Cooley - Rajeev, why do you say you're the best?
Madhavan - For any product niche in EDA, if you're the leader, you've got 50 percent.
Cooley - What's your market share?
Madhavan - It's 33 percent in the RTL-to-GDSII market.
Cooley - I thought you had 100 percent in some weird sliver of the market.
Madhavan - I know it's whatever Gary said, and now Erach [Desai] has made up some other numbers.
Cooley quotes market stats from Gary Smith
Cooley - Gary, is Rajeev telling the truth?
Smith - He's off by a few hundred million, but yeah - he is growing.
Chilton - That's not what the public numbers say.
Madhavan - You adjusted your numbers when you changed your licensing model, but when we changed our licensing model and looked at the numbers, we were still growing.
Chilton - So you're talking bookings.
Cooley - Synopsys, is Rajeev outgrowing you?
Chilton - That's hard to believe. We're 4 times Magma, so I don't get their numbers.
Cooley - If you take place & route?
Chilton - Still the numbers don't make sense.
Cooley - So who's telling the truth?
Chilton - It's hard to lie to the SEC these days.
Cooley - Gary?
Smith - There are the numbers that get reported, but we scrub the hell out of those numbers. Magma is outgrowing the market.
Chilton - I thought it was RTL to GDSII.
Cooley - I was just saying place & route.
Chilton - Okay.
Cooley - So Blast Fusion is outgrowing the market.
Cooley - Vic, why do you advocate moving all jobs to India? Why do you hate America? Why do you want to outsource everything?
Kulkarni - Firstly, that's the dumbest question I've ever heard in my life.
Kulkarni - I came to Cincinnati 33 years ago for graduate school and then here to Silicon Valley. I have my family here. But nonetheless, this is not about India. Whoever's asking that question has to think about globalization issues. It's not just in India, the U.K., or Japan. It's wherever R&D is happening. It's about where are the customers. There are 130 design companies in Bangalore, Hyderabad - companies like TI, Infineon, Cisco. And there are 15 EDA companies in India employing 2000 engineers there. Companies like Cadence, Synopsys, Bluespec, Synfora, and so on. Really, we're talking here about globalization trends. The key issue is how do we leverage the globalized workforce
and create products.
Cooley - Atul, why are you giving talks about how to outsource?
Sharan - We don't have facilities in India.
Cooley - Vic will help you set them up.
Sharan - But, we do have direct operations in Taiwan.
Cooley - Why are you giving talks like Vic about outsourcing?
Sharan - We do serve our customers in India.
Cooley - Atul, what's your exit strategy. Do you have one?
Sharan - You have to ask the broader question which is how to focus on finding value. You do that by 1) providing IP, 2) by being hugely cash flow positive, which Denali has done a great job of, and 3) by getting acquired. That's the order of our preference, but the reality may be just the opposite. In the EDA segment, acquisition is the way to go. But the larger question is still one of valuation. We all suffer because the EDA market valuation is so compressed versus other industries.
Cooley - What about the Brion acquisition?
Sharan - That was great, especially because of what ASML paid for Brion. But it's not clear who's going to acquire other DFM companies in the sense that ASML is now competing directly with Synopsys and Cadence in OPC.
Cooley - Which DFM company will die next?
Sharan - Definitely not us!
Sharan - Who died?
Cooley - There was the Aprio fire-sale acquisition by Blaze.
Sharan - I missed that because I was preparing for this panel.
Sharan - But the issue is differentiation, not what needs to be done because there's consensus on that. People now are following what we say should be done. Even with some of these zero-dollar acquisitions, we still have the lead.
Cooley - Just name names of who will die next.
Sharan - That's a tough question.
Cooley - Should we start with Clear Shape?
Sharan - You should stop with Clear Shape. I have a question for you, John. Are there any other tools than Clear Shape's that have been written about on Deep Chip?
Cooley - I have to think about that.
Sharan - Think fast, and next time prepare better for this panel.
Sharan - The answer is no. I would encourage the audience to visit John's website.
Cooley - Ted, why not join the TCL era? It's only been 20 years.
Vucurevich - John, do you realize that our DFT implementation strategy has supported TCL forever?
Cooley - [reading a question] Ted, when do you anticipate the TCL era? My neighbor in the next cube is always learning SKIL. I have nothing against SKIL, but it's becoming the French of the EDA world.
Vucurevich - This individual is obviously one of our Virtuoso customers. We have no intention of modifying that capability. We've enabled other companies like Cira Nova, for instance, for doing PCell-like development.
Cooley - Are you happy about that?
Vucurevich - I would say that we are neither unhappy nor displeased with the fact that the interoperability that we've enabled has allowed people in the market to do something.
Cooley - But guys say it's a ruse, that Open Access is not really open.
Vucurevich - It's certainly the case that Open Access was an attempt to do something good. Over 200 engineering years of R&D have been made available to the industry without charge, fourth-generation technology.
Cooley - What do you mean?
Vucurevich - The reference implementation is what we ship.
Cooley - Rajeev, did you know there is free engineering at Cadence?
Madhavan - To me, it's Open Access but keeping the PCells closed is not open. There may be a start-up that may see that PCell is available in Open Access, but Cora Nova has achieved it through a back-door approach. Putting 200 years into development for the use of a database is a moot point. The whole model of OA as a database is becoming moot unless you enable competition to happen.
Cooley - Is Open Access really open, Synopsys?
Chilton - It's still controlled by Cadence, but it's useful. Of course, it's not useful for us because we do complex digital design requiring something more specialized than that.
Cooley - Cadence can't do digital design, but you guys can?
Chilton - It would be more interesting if SKIL and PCells were open, then the last citadel would be gone.
Vucurevich - To put it bluntly, that's not going to happen. We have to differentiate between what is a competitive line. There were many people in management who said it was nuts to do anything like this at all, but we said here's a chance to do something in an open forum. Opening a data model like Open Access is a big deal. But at the same time, from a competitive point of view, to say here's a free ride for coming in and taking accounts doesn't make sense.
Cooley - That makes sense. I don't see Synopsys or Magma enabling the competition.
Mike Santarini [from the floor] - But wasn't opening SKIL the FTC requirement?
Vucurevich - There's a lot of obfuscation about what progression happened. To set the time line - in 1991, Cadence opened the Connections Program to allow customers who had other vendor tools to have a method to provide some form of interoperability. Since '91, that has grown to over 130 companies driven by customers who want to interoperate between vendors. In 1997, we acquired the Cooper & Chen technology, and due diligence from the FTC said that to do routing technology Cadence must agree to a consent decree.
Santarini - But the FTC specifically called out SKIL.
Vucurevich - For the purposes of routing. There was a part of the Connections Program which had the ability for SKIL and through a challenge to the FTC of that review, it was concluded that Cadence has been very consistent in supporting routing technology. That's what the consent agree said, and it's about to expire.
Cooley - Yeah? Is the Connections Program going away?
Vucurevich - No, not at all. It will continue because it's our intention to have good integration of technology across companies. It's a bit complicated because there's been a lot of FUD.
Cooley - Synopsys? Why hire lawyers when you're losing a tech fight? Nassda? Magma? Is Sierra next?
Chilton - We don't bring out the lawyers, and hopefully we're forcing the end of questions like that at this panel. This is an industry built on figuring out customer problems and developing solutions. But you have a fiduciary responsibility to protect your IP. That's no more so in our industry than any other. It's an unfortunate part of business. People are forced from fiduciary responsibility to protect their IP. It's in the hands of the capable courts now and we'll see.
Cooley - Rajeev?
Madhavan - There is one answer for multiple questions. This industry has to show to its shareholders that we prefer not to confuse our customers as much as we can. To compete on our technical strengths. We have not done that as an industry
Cooley - Are you two kissing and making up?
Madhavan - We look forward, like John said, to winding down. It takes two to tango, and we're happy with the progress in the courts.
Chilton - The technical battle is completely orthogonal to what's going on in the lawsuit. We're very happy with our position in P&R, synthesis, static timing, test, and so on. We're solving the toughest problems and we're as strong as ever. We're recognized as high end and that doesn't change. Law suits for some are a smoke screen. It may be realized that every 3 years, there are lawsuits.
Madhavan - Let's not waste money in the industry. We are bickering [but] if technology is all that matters, then let's fight on the technology.
Cooley - You guys are practically singing Kumbaya to each other. Your lawyers must have briefed you to be good.
Cooley - Kissy, kissy? I'm happy for all of you!
Cooley - Brett, we're still hearing that every year will be the year of SystemC. So will next year be the year of SystemC?
Cline - Last year was the year of SystemC.
Cooley - So it's already passed? It's on the decline now?
Cline - Last year was the first year SystemC was being used in wide deployment. I get asked a lot if the first year of SystemC is the same year that VHDL and Verilog die, but things don't work that way. Things stick around. SystemC and SystemC synthesis are like that. The HDLs won't go away.
Cooley - It takes 6 months for C++ training. [reading a question] Why take a great Verilog engineer and make him a mediocre SystemC programmer?
Cline - If you're a hardware designer, you'll be a good hardware designer no matter the language. We benchmarked this by taking a young, hot stud C++ designer and putting him up against a Verilog guy working in SystemC. The Verilog guy kicked his butt because he knew how to make hardware work.
Cooley - Mentor? What are you doing in SystemC?
Chern - We started with some companies and customers in Europe. Then we started to broadly engage other companies. We're doing very well in Japan and the U.S with 63 customers.
Cooley - 63 sites or 63 customers?
Chern - 63 customers. We're very confident about the adoption. The biggest focus is not about competing between C and ANSI C. We're focused on design starts and adoption.
Cooley - Synopsys, why isn't SystemC taking over the world?
Chilton - We find that hardware engineers like Verilog and SystemVerilog. We're huge supporters of SystemC and find great utility for SystemC for virtual platforms, especially for mobile companies. We like that the software can be up and running before the hardware is available. EDA is a little overboard on interoperability, but we just donated some technology to OSCI. I just don't think, looking at our services and our IP, that coding up the RTL is really the hard part of the design. Changing that from Verilog to SystemVerilog is great because you can lose down to one third to one tenth of the lines of code. But we haven't really seen great SystemC implementation there.
Cooley - Then why are Mentor and Forte making money?
Chilton - You can make a little bit of money in small markets. We just don't think it's mainstream.
Cooley - Ted, what about SystemC at Cadence?
Vucurevich - We believe in the technology, plus we contributed to the SystemC standards and constructs. We've followed that on the verification side. The applications of the high-level languages for synthesis are being explored by Forte and [Mentor's] Catapult, but we said the big opportunities for implementation are the TLMs.
Cooley - So synthesis is not worth it?
Vucurevich - We don't say that at all, contrary to Synopsys. We believe in the applicability of SystemC. Watch this space!
Cooley - Gary, I think you're dying to say something.
Smith - I have only one disagreement with Brett. 2005 was the first year of ESL, not 2006. It was the “1987” of ESL. The methodology that Daya [Nadamuni] and I did back in 2004 helped with that, people agreed with us, and we've got the direction now.
Cooley - [reading a question] Now that you're running your own business, do you have to work on real markets?
Smith - Whoever wrote that question, they should contact me between 2010 and 2013. At that point, I'll help them get a job in the ESL market because there will be no more jobs in RTL markets.
Cooley - Atul, how do you compete against the Big Guys? Mentor, Synopsys, Cadence, Magma and [in a falsetto] Clear Shape?
Sharan - We don't see they all have solutions in that space.
Cooley - [reading a question] Are you and Cadence “Brokeback” buddies? Who pitches and who catches?
Sharan - Speaking of dumb questions!
Sharan - What does that question do except to insult a large constituency near to here?
Cooley - Do you mean the Indians?
Sharan - In terms of competing with other big companies, if you look at substantive evidence of who's competing, we're the only company certified at companies and fabs. It's a really big problem - people are fighting resolution problems. You're still getting silicon that's different from what you designed and modeled. We're the only company certified by the big fabs because we've got a solution that fits into any flow, and we don't see any competition. Clearly Synopsys has done an acquisition, but they still need to integrate. We've got a great relationship with Cadence. We're good partners.
Cooley - But as time goes on, Ted will say they've got great solutions. Then what?
Sharan - That's just a hypothetical.
Cooley - I'll bet it happens in a year.
Vucurevich - I'll take that bet.
Sharan - Our success is not dependent on Cadence's ability to integrate.
Vucurevich - I would love to follow Atul up on this. It's an extremely important example of how the EDA industry will evolve. In today's world, if you want to go IPO, The Street won't give you good multiples, so that's a hard question to answer. The cash flow model is a great model, but how to do that at scale? Before it was only through acquisition. I get acquired and have some value add - large sales and support organization can be immediately applied to my technology. What you're seeing now is a new dynamic of how that can happen. A high-value, specialized capability can be maintained over time by integrating into a larger environment. Clear Shape is a terrific role model of this
type of relationships. Don't be surprised if in the future, other people on this panel do the same.
Cooley - Well, either you get married or you don't.
Vucurevich - Atul's got some great technology, but at this point in time there's nothing that's driving either company in that direction.
Cooley - Jue-Hsien, is Mentor waiting until the Magma/Synopsys lawsuit is settled before acquiring Magma?
Chern - Another dumb question!
Jue-Hsien - We're not going to comment on rumors and we're not going to address them. I don't know where that rumor comes from.
Cooley - It's talking about the hole in your flow.
Jue-Hsien - We're always asked when Mentor is going to have solutions, so the rumor is when you are going to acquire this or that company. Always it's the assumption that the customers prefer one-stop shopping. But that's not the correct solution from the customers' point of view.
Cooley - But Synopsys and Cadence say it all time. You can buy from one vendor and improve efficiency.
Madhavan - Nobody in EDA goes with one-stop shopping. It's about technology improvements. Where that's been true, Mentor and Magma have worked together. On at least 3 accounts, we've made it happen. Those things won't make change happen. Ultimately, customers control it from the quality of the simulations they get.
Cooley - You should team up with Synopsys!
Madhavan - Why should we? We've got better synthesis tools!
Madhavan - EDA is not built on one-stop shopping. It's not as commoditized as people make it out to be.
Cooley - Mentor, are you saying there's no real need for a place & route tool in your flow?
Chern - Clearly we can't do everything. We can make a difference. We choose areas where we feel we can make a difference, where we're able to identify areas of discontinuities.
Cooley - So you're not working on a place & route tool?
Chern - No.
Cooley - Ted, is anybody using the 6.1 Framework?
Vucurevich - You're talking about the 6.1 Virtuoso Framework? Yes!
Cooley - Thanks!
Vucurevich - We've had a number of customers who were first-generation users moving from Edge to Opus database technology. In our early adopter program, we had a number of customers. And, talking about SKIL, people were saying, “Gee, I would love to have access to SKIL.” Trust me, you have no idea! We can say that our early adopters
[looking out to someone in the audience]
actually I don't know how many of them I can mention.
Cooley - You've got puppet masters off to the side?
Vucurevich - Yeah, we've got early adopters bringing it into their flows. There's no going back. There was a question about Open Access. The 6.1 release represents no going back.
Cooley - So you're dammed if you do, and damned if you don't?
Vucurevich - Yes, you have to make sure you've got resources committed to going back in the field. It's interesting. I would like to make at least a little bet here. I'll bet Rajeev that you won't get to 50 percent by 2008. I love all the analog and mixed-signal competition. It's great! But 6.1 is due to internal innovation even though people say it can't be done in big companies.
Madhavan - Yeah, we heard that back when Integration Ensemble was going to obliterate Magma, but it didn't happen.
Vucurevich - That not's true.
Cooley -Liar, liar pants on fire?
Vucurevich - The integration and technology have been maturing over the last 3 years. It's the world-class power flow in the industry today. Period. We'll benchmark against anybody and we'll continue to do so in all specs going forward.
Cooley - Why can't there be just one standard in power?
Vucurevich - That's a great question. I look at Brett and ask why can't there be just one standard. Look what we did.
Cooley - But it seems you and Si2 decided to just make a standard that's just not Synopsys.
Vucurevich - In SystemC, for example, we decided to donate to just make one standard work. In this power area, we put out a standard for something that actually worked.
Cooley - So you're saying the Cadence solution works?
Madhavan - Cadence came out and said let ours be the standard. But we worked together with Synopsys and Mentor - three vendors worked together and came up with a format. Ted said that PCell is not open for business reasons, but here are three companies here who are ready to take their formats and make them open. The three of us have demonstrated that. UPF is a universal standard today. We will all support this.
Vucurevich - Our basis was to work with customers and to prove that technology for the entire flow.
Madhavan - It was 2 customers, TI and Nokia, that drove it.
Vucurevich - That's not what you just said. Our flow is not just 80 percent, but 100 percent.
Madhavan - Open PCells, CPF!
Vucurevich - It's open!
Madhavan - Not true!
Chilton - We're legally prevented from seeing the CPF spec today. That's not very open.
Vucurevich - I don't understand where you can be legally bound.
Madhavan - That's what's happening
Cooley - Why can't you read it?
Chilton - The whole process is convoluted. At the beginning we were asked to join LPC, but with an NDA. That wasn't very exciting. Then we were enticed with a read-only version, but couldn't do anything with it. That hasn't been a real impetus to join LPC. If you're not in LPC, you don't have the right to look at CPF. But CPF and UPF are in the same family.
Vucurevich - So what you're saying is that if you were a member of LPC, you would not have trouble modifying the standard?
Cooley - But you're restricted?
Chilton - Then we would have been able to see it in November and December, but now UPF is done.
Cooley - Ted, don't you think it's weird that Magma and Synopsys are agreeing?
Vucurevich - Whatever training they've had in other areas is leaking over.
Cooley - Why not just open CPF?
Vucurevich - It's simple to have access.
Chilton - But only 6 people have access!
Madhavan - We wanted to bring CPF into the fold.
Vucurevich - What does “into the fold” mean?
Madhavan - It means everybody looks at the formats and discusses them.
Chilton - Accellera just voted the UPF standard in today.
Vucurevich - Good for Accellera.
Chilton - It's done!
Cooley - Vic, what do you support?
Kulkarni - Whatever the customers need. We've been advocating that through Accellera, Si2, CPF, and UPF. We want CUPF, a Common Universal Power Format.
Kulkarni - There's a danger here of the old Verilog/VHDL issue, but we support Cadence, Synopsys, and Magma - 40 percent of our customers use Cadence, 20 percent use Synopsys, 20 percent use Magma. But we want to champion just one standard.
Cooley - Or support none?
Chilton - I'm confident we'll get to one. My prediction is that once CPF is open, the technology team will get together. They won't find huge technical difference and it won't be possible to sustain multiple efforts. I'm hoping everything will come together.
Cooley - Isn't this the kiss of death, Vic, for your company?
Kulkarni - If that were the case, it would also be so for innovation and best-of-class solutions. It turns out that for us, power is not a single silver bullet. You need 20 different techniques. The Big Guys have tremendous implementation flows, and people like ourselves have to add value on top of that. This year we're growing 22 percent, after growing 18 percent last year. For things like power and timing, luckily there are holes in the flow for us to fill.
Cooley - I'd ask you what your exit strategy is, but clearly you've been around too long for an exit strategy.
Kulkarni - It's the journey. Great innovation is always required to meet the customer's needs. What is an exit strategy anyway? M&A is just another point in the journey. There's really no such thing as an exit, it's really about solving problems.
Cooley - Brett, why did the Forte CEO leave after less than a year?
Cline - This question from a guy who changes his underwear more often than he changes panelists!
Cooley - I don't change my underwear that often.
Cline - We hired David Sears to organize the company so the growth path would be faster. Then in Q3 2006 we got $5.5 million funding, which was a nice round. David and the Board got together at that point and decided the person who needed to lead the company was an EDA technologist. So we looked around and found Sean Dart who was our vice president of engineering. Sean knows the customers and products and now has a war chest. And he understands ESL in general.
Cooley - Why did Forte bail out on the Bluespec challenge?
Cline -I categorize that under the “EDA Duel” category. Some small EDA company needs notoriety, so they challenge somebody to a duel. I liken it to a hyperactive 4-year-old kicking somebody and looking for attention. The answer is, we've got a lot of resources and we're growing our business. Bluespec is not our number one competition, and neither is Mentor. Sitting and bashing each other is not where we should be.
Cooley - Synopsys? Talking about companies kicking each other - is PrimeTime a monopoly?
Chilton - We don't like the word monopoly.
Cooley - Gary, what's the PrimeTime market share?
Smith - 95 percent.
Chilton - So, what are we going to do in PrimeTime? We're going to meet with customers as always, ratcheting up the performance of the tool, and working on variability problems that have been around for a while.
Cooley - Isn't Cadence trying to do this?
Chilton - But this is their third time around.
Cooley - But they've got a huge sale force.
Chilton - So we take this seriously. The only thing we can do is work like hell to make things a better and better product for our customers. It's lucky that people come to us and talk in detail about their issues. If we stay engaged and work real hard in R&D - there aren't any guarantees and there are smart people at Cadence, but our strategy is to work as hard as we can.
Cooley - Ted?
Vucurevich - We look forward to the competition. John is right. PrimeTime's been a workhorse for them and the golden standard. But we have every resource available to us to put out capability. Quite frankly that core piece is part of implementation. If it were just signoff, we wouldn't bother. But timing is an integral part of the whole process. We have to provide assurance to the people who use our implementation flows. We've got very serious investments in this area, in modeling and analysis, plus a serious focus on integration of this technology deep into the implementation flow. As a result, we do have focus on the signoff step. Quite frankly, the industry will benefit from our
serious people working on this problem
Chilton - I can't resist saying this, but we've done everything we can do to promote interoperability, unlike PCells. On a positive note, people have to have great products. It's all about correlation into silicon.
Cooley - I know I welcome it. The best thing that ever happened to Design Compiler was Ambit. The best thing for PrimeTime will be Cadence tools.
Chilton - But PrimeTime is a great tool!
Cooley - Rajeev, I've got an email from a guy in Munich who said his timing closure is going to take 4 weeks, but the press release said it should only take 2 days.
Madhavan - You sent me that email, and it was about a bug and a misunderstanding. I look forward to seeing him in Munich next time I'm there.
Cooley - So you found a bug that was a show-stopper? You didn't put a contract out on his head instead?
Madhavan - We solved the problem.
Brett Cline interrupts
Brett - The panelists decided that John Cooley also needs to answer a couple of edgy questions.
Cooley - This is not what Brett said he would do [if I gave him some time].
Cline - First of all, it seems like ESNUG has become the National Enquirer of EDA? But what's with this fascination with Mike Fister's shoes, John? And how come we don't hear about the sheep on your farm anymore? Did you guys break up?
Cooley - Note to self - Brett not to be on panel next year.
Brett Cline presents John Cooley with a white t-shirt emblazoned “Troublemaker”
Smith - Remember your girlfriends and the white T-shirts? You said your DAC T-shirts made great nightgowns for them?
Santarini - You put T-shirts on your sheep?
Laughter until Cooley stops blushing/stammering
Cooley - [reading a question] Ted, the new licensing for analog tools? Just a price raise?
Vucurevich - The only thing that has changed is the packaging option. We've gone to segmentation. Maybe the individual who sent this question is having trouble mapping into the new segments?
Cooley - Rajeev, Quartz fell off the map?
Madhavan - Not at all. We have several papers coming up at MUSIC [Magma Users Summit on ICs] next week. We have customers standardizing on 65-nanometer designs. We have 25 customers. It's met all expectations at 10-percent growth. It can't grow any faster than that.
Cooley - Mentor, are you losing market share to Magma?
Chern - I refer to your favorite article where your wrote the early benchmark and the article. Last year we released NanometerDRC. The customers adopted it and find this deal is even better than before. If that's how somebody goes about reducing market share, we'll gladly take it.
Cooley - Ted,?
Vucurevich - NanoCaliber is a good update of the Caliber franchise. They've raised the bar. We're in the process of making sure our technology meets the new bar. We're taking a bit longer than we would have liked, but we're not backing off.
Cooley - Gary, now that Gartner is gone, you can tell us - what's the worst stats you've seen from companies running amuck?
Smith - They've been doing that for years, so I haven't been paying that much attention lately. I've been too busy setting up my new company. The worst case from the past, however - Cadence reported 120 percent of their revenue. Then we reminded them about their SEC obligations.
Smith - It was also difficult when Synopsys sent in 5 quarters for one year.
Cooley - They had to keep up with Cadence?
Smith - They had moved their fiscal year out a quarter. We really had to dig in to really sort it out.
Chilton - I think Mentor's doing that this year.
Smith - Those were the two worst cases. That's why we scrub the numbers.
Cooley - Isn't that done by the companies?
Smith - Just buy our report.
Cooley - [reading a question] What's the difference between your numbers and EDAC's?
Smith - EDAC's numbers not scrubbed. They're supposedly checked against the SEC numbers, but since there are only a few public companies and not all revenue is known
We just follow product numbers, not maintenance. And you've also got to separate between currencies.
Cooley - Do you think the EDA industry is stagnating?
Smith - Not at all. I don't see any real reason why any company in the EDA market would want to standardize on one flow. You've got 3 tool sets that can do almost the same thing and they've been competing on price for 5 years. So for the vendors, it's just about how do you get the money.
Cooley - So they're keeping price discipline?
Smith - No! There was hesitation for 6 months when Fister came on board [at Cadence]. How were they going to compete? They didn't want to work on the technology, so they become propriety. Just try to take a Cadence PCB design and take it over to Mentor, or vice versa.
Cooley - So the customers are locked in?
Smith - They're locked in!
Cooley - Ted, has EDA gone flat?
Vucurevich - It absolutely has not gone flat, but it has gotten more sophisticated in how you invest. In the past, it was a Wild West. Every 18-to-36 months you'd have some event with significant re-tooling. But we haven't seen those kinds of events coming lately. We're probably good in the industry for 3 or 4 more scaling, but then it will be competing not on scaling, but on innovation.
Cooley - So you'll see growth?
Vucurevich - No, it will cause improvements in the technology, not commoditization. We've all looked at what the next algorithm is, but the next question is what is the next integration that will guarantee convergence. And the verification space is still a huge market.
Cooley - Is Rajeev in trouble without verification tools?
Vucurevich - Rajeev needs to get those tools.
Madhavan - We've continued to innovate. And EDA is not flat because all 4 public companies have had really good years.
Cooley - But there's no crazy “woo-hoo” going on right now?
Madhavan - Yeah, but growth is steady. In fact, all 4 companies have done really well!
Chilton - We're announcing good growth!
Cooley - And the overall industry?
Chilton - There are designs out there that lots of tools can do, but we work on problems that the customers are really working hard on. There is still lots of innovation going on. It's a revolution out there, giving tremendous value to the customers. You do see great tools with the vendors competing with each, but if you look at efforts to develops tools and provide support, we're a long ways from done. It's not the rocket market it was when everyone was moving from full custom to standard cells, but a lot of people would like [to be part of] a market like EDA today, one that's grows 10-to-15 percent a year!
Cooley - Thank you! Free beer downstairs in the Exhibit Hall!
Peggy Aycinena is Editor of EDA Confidential and a Contributing Editor in EDA Weekly.
You can find the full EDACafe event calendar here
To read more news, click here
-- Peggy Aycinena, EDACafe.com Contributing Editor.