The word of the year: CHANGE

by Gabe Moretti

As all of you know, the United States is in the midst of a political campaign that will ultimately decide who will be its next President.  The word most often used by all candidates is: Change.  Everyone is for change, although everyone defines it differently.  It seems a bit strange to me that candidates should have to spend so much energy telling everyone that they will bring change.  After all, there is an election, the incumbent is not running, and thus there will be change, even if not desired.  Yet, someone has figured out that there cannot be progress without change, and so the word of the year has been identified: Change.  EDA is one of the most dynamic industries in existence, and its rapid changes are driven by just as dynamic an industry: electronics design.  So it will not surprise you that this month's articles are all about change.

DAC Leadership
In her career, Limor Fix has proven to be an outstanding technical manager.  You have to be one in order to achieve her present position at Intel.  This year she is using her talents to lead her team as they work to transform the Design Automation Conference to take advantage of the latest communication tools while maintaining the very high level of technical content and the vibrant showcase of EDA products that have made it the uncontested primary conference for our industry.  As Peggy Aycinena relates in her profile of this year's DAC Chair, those attending DAC-2008 in Anaheim will see a conference that, while retaining the characteristics that have served the industry well in the past, will offer new approaches and new opportunities to learn and network.  In addition, Limor introduces the readers to ways in which DAC is transforming itself from an event to a year-round tool for both academics and practitioners in the electronics design industry.

EDA Startups
Ed Sperling sat down with several EDA leaders who have each started more than one EDA company in their careers so far, and asked them to share their views on the prospects and problems new EDA startups face in the present business climate.  The answers, at times quite personal, will be very helpful to those considering a startup, as well as those of us looking for insight into the current state of the EDA industry.  Let me know what you think; you can be part of this discussion by writing to me.  The Letters to the Editor section is always available.

Speaking of the Letters to the Editor, we are publishing two of them this month. They comment on the articles on analog design in last month's issue.  As you can read, analog design is receiving a lot of attention, and we have not yet reached consensus on how to improve tools and methods for that segment of electronics design.

Protecting electronics contents
In his Viewpoint, Craig Rawlings of Kilopass Technology explores one way to improve security in digital storage that has proven to be commercially successful.  The issue of Property Rights Protection is a critical problem as distribution of all forms of artistic expressions is relying more and more on electronic means as opposed to mechanical ones.

Read the rest of the article

********************************************

The DACeZine also has a Letters to the Editor section to allow for shorter contributions to the contents and directions of the publication. When necessary, answers to the letters will come from the appropriate member of the team (including our readers), since I do not (yet) hold the total knowledge of the industry within me. I encourage all of you to write, either a viewpoint or a letter, and state your opinions on matters that impact our industry, the contents of this publication, or, for that matter, the publication itself. Send your letters to: dacezine@dac.com.

I hope you enjoy this DACeZine issue and pass it along to your friends and colleagues: I am sure they will want to subscribe as well. They can do so by visiting the www.dac.com web page.

The material in the two articles on analog design resonated with our readers, and I am glad it did.  I picked two letters that represent the general sentiment of our correspondents.

The first letter is from Eric Filseth, as follows:

There seems to be an ongoing dialogue in the EDA industry, touched on in your January 4 issue, to the effect of, “analog designers don’t want tools."

I don’t believe this is true.  With consumer and communications now accounting for something like half the world’s non-memory wafer shipments, analog, RF and mixed-signal design are more relevant today than any time in the last 20 years.  Most analog engineers I know are busy all the time, and usually under pressure.

These folks aren’t asleep.  If they don’t use tools then it’s largely because we, the EDA industry, have not given them tools they want to use.  In an age where analog layout is still done by hand; where designs take multiple respins to work right just at shipment, never mind over their operating life; where nobody can predict parasitics until late in the design cycle; where each process port means a near-complete design; and where engineers draw polygons by hand because their PCells can’t handle DFM rules and don’t work in their verification tools; it’s not for want of analog, RF and mixed-signal problems to solve.

Eric Filseth
CEO, Ciranova

Dear Eric,

Analog design has always had an aura of mystery surrounding it.  It seems to defy reusability; it seems to require invention and creativity well beyond that of digital design.  And yet, every implementation is an approximation of the ideal algorithm that constitutes the solution to the design requirements.  The fact that there exists a healthy market for analog standard parts is to me a sign that reusability is possible in analog, just as it is in digital designs.  At the same time, I continue to hear that each mixed-signal chip is unique and requires handcrafted analog circuitry.  The realities are that both design and verification are expensive, that the probability of on-time delivery is less than optimal, and that EDA companies do not perceive the existence of a market that would justify the projected development costs to serve analog designers beyond what tools are available today.

The second letter comes from Mike Demler, who was quoted in Geoffrey James' article.

It’s good to see the coverage of analog and AMS on DACeZine.  I appreciated the opportunity to be interviewed by Geoffrey James for his article, but (unfortunately) only one sentence from that conversation was included in the publication online, and the topic apparently changed considerably from the briefing topic provided before the interview.  Now that I see what was published regarding ESL, analog synthesis, …, Luddites (?) I just had to write.  There is so much more for me to comment on if these are the topics of interest.

Rather than re-hash recent articles I have written on the topic, I invite you to visit my blog, especially my post on “Analog design is NOT black magic… but it is VERY hard.” In that post I especially take issue with the contention in Geoffrey’s article regarding “The fact that EDA vendors have so far lagged in helping analog designers may be the result of not enough "out of the box thinking."    I dispute that alleged “fact” quite strongly while highlighting some of the advances that have been made in EDA tools for analog design.  

You saw that James Lin regards simulation as the biggest problem for his designers at National.  Well, 20 years ago (the same time when Synopsys began), it was nearly impossible to simulate a PLL, a complete ADC, a SerDes, or any RF circuit for that matter.  All of these circuits are routinely handled by the industry-leading SPICE and Fast-SPICE simulators today, without a second thought in most cases.  While you may not think of it as a paradigm shift akin to Design Compiler, this represents orders of magnitude improvement in design productivity!  And that’s just simulation.  What about schematic-driven layout, parasitic extraction and DRC, and many other tools that are routinely used today for transistor-level design?  Twenty years ago I had to push every polygon manually, and as far as post-layout analysis goes… forget about it! You wouldn’t have any of today’s mixed-signal devices if EDA vendors had lagged in the analog tools.

The topic of analog synthesis is especially significant to me.  I was the architect for Analog Synthesis at Antrim Design Systems, for which I have been granted five patents.  The conversation on this topic should start with those who do or have done analog design, not your typical EDA marketeers or digital synthesis experts.  What problems need to be solved?  What are the biggest areas needing productivity improvement?  James Lin said it is simulation, a particularly hot area in the EDA industry at present.  It may be that custom analog design is not a “plague” (ouch!) on the industry.  Many studies have shown that verification is the biggest bottleneck—another area that the industry and Synopsys especially is focusing on.

Thanks for stimulating the dialogue.  Let’s keep it going!

Regards,
Mike

To refresh everyone's memory, this is the sentence from Mike quoted in the article:
"Automation is possible but it's very difficult to change the methodology for analog designers because they are stubborn beasts for the most part."

I am grateful to Mike for his heart-felt defense of analog designers and EDA tools, but I also must take issue with some of what he states.  Yes, it is true that simulation of most analog circuitry is possible now versus 20 years ago, but much of the improvement is due to huge advances in computational power, and better compiler tools coupled with better programming languages.  SPICE, in its various incarnations, is as "leading edge" as it was in the '70s.  And I also agree that there have been quite a few inventions in the area of analog synthesis; on the other hand, most of the startups that targeted this market have gone out of business, not for lack of engineering talent, but for lack of profits.  It may be that analog designers are not averse to new tools, but that it takes longer than anyone has so far planned, and therefore more investments, to establish a viable market.

Change Agent Limor Fix: YouTube & the Next Generation

by Peggy Aycinena

Limor Fix, 2008 General Chair for the Design Automation Conference, is only asking one thing of her various committees’ heads: Do something different this year. Just one thing that will shake things up, evolve the conference, help move the industry in a new and appropriate direction. Not surprisingly, her committee chairs have responded.

Although Fix partially credits her long years in management-previously as head of an internal CAD R&D group at Intel, and now as director of an Intel research lablet in Pittsburgh—for giving her the leadership skills to promote out-of the box thinking from the 2008 DAC team, she argues the larger credit for their innovative ideas comes from the caliber of the team itself. “It’s always about the team in technology, and the 2008 DAC team is top notch. Best ever, in fact,” Fix told me in a recent phone call.

Few would take issue with Fix’s boast. Just look at the team: U.C. San Diego’s Andrew Kahng; IBM’s Leon Stok; Magma Design Automation’s Patrick Groeneveld; Virage Logic’s Yervant Zorian; University of Pittsburgh’s Steve Levitan; Synopsys’ Narendra Shenoy; University of Minnesota’s Sachin Sapatnekar; Purdue University’s Kaushik Roy; Mentor Graphics’ Dennis Brophy; NVC Marketing & Public Relations’ Nanette Collins; Katholieke University’s Georges Gielen; Kyushu University’s Yusuke Matsunaga; Carnegie Mellon University’s Diana Marculescu; Qualcomm’s Ramesh Chandra; Mentor Graphics’ Anne Cirkel; MP Associates’ Kevin Lepine; and MP Associates’ Lee Wood.

Fix says the 2008 DAC Dream Team is historic in its talent and energies: “This team is doing more than I ever could have imagined possible, and the conference is going to be supremely successful because of them. In previous years, the Executive Committee used to meet every three months, but so many new ideas are being brought online in 2008 that the committee is happily meeting every month in the run-up to the conference in June in Anaheim.

“We are crafting a Design Automation Conference that will reflect ongoing changes in the industry. Change is happening in design automation at an even faster pace than what we’ve seen in the past—and this in an industry that’s legendary for its pace of change. There is a change in tools on the horizon and a change in the way design is done that’s creating a crisis for design managers all over the world. They’re having to say to their teams, and to their management, ‘Okay. This time as we start our newest project, we’re going to have to do things differently. We’re going to have to move things to a higher level of abstraction and, therefore, be willing to integrate even more external IP, or IP from different groups within the company, to make our new project succeed.’

“And,” Fix adds, “Not only are the designers changing their methodologies and attitudes, the computer aided design companies—the EDA vendors—are changing their attitudes, as well. They’re developing new technologies and methodologies that are working at the new, higher abstraction levels needed by their customers, while also acknowledging that the fine tuning of new technologies and methodologies cannot happen until the design teams take it upon themselves to absorb the risk associated with the new tools. It’s a two-way street—out of necessity, the customers have to use the new tools and methodologies, while at the very same time, the new offerings from the vendors are being stress-tested on real projects with real deadlines.”

Fix says there’s enough risk in all of this to sink a [corporate] ship, but she also senses great optimism: “I am seeing more and more willingness in the design companies to be open to new technologies and methodologies. Amidst everything the industry, and the DAC Executive Committee, are talking about—system-level design, new levels of abstraction, additional use of IP, DFM, and new verification technologies—you can see the adoption of all of these things is actually starting to happen across the globe. It’s taken years for this to become a reality, but adoption is definitely kicking in.”

Managing Change

As admirable as all of this forward thinking may be, Fix lists several specific innovations related to DAC 2008. “First of all,” she says, “The DAC website is sporting a fresh new look and the DACeZine, newly launched this year, is also adding a refreshing new voice to the industry.

“Second of all, the exhibitors will welcome a host of new ideas and updated features in the conference that are promising to enrich their experiences and ROI. To that end, every day at DAC in Anaheim in June 2008 will be a "Free Day" on the exhibit hall floor. Attendees will have to register a few days in advance of the conference, but if they do, the floor will be open to them – which will be a tremendous boon for the exhibitors.

“Third, DAC attendee feedback from past years has generated additional program content for 2008 that will look more closely at the methodology of design through a variety of ‘how-to’ sessions that are being integrated into the schedule. [Hence], we are looking at opportunities for learning to take place at DAC for all types of technologists.

“I believe when a person comes to DAC, they are already expert in some sub-domain within the technology. While DAC can give that individual an updated perspective on current trends in their own area, it can also give them more in-depth exposure to other domains, as well. If you’re in formal verification, for instance, DAC will give you multiple sessions to round out your knowledge, but will also offer extended collocated tutorials that offer additional exposure to advancements in other, complementary technologies.”

Finally, and most profoundly, Fix says, “The entire Executive Committee is revamping their thinking about the conference itself. No longer are they thinking of DAC as a single event that happens just once each year, sometime in the late spring or early summer. Instead, they are envisioning DAC as a year-round phenomenon – an ongoing, always vital source of information where academia and industry, students, professors, and industry leaders can exchange ideas and develop a dialog via DAC’s monthly newsletter DACeZine and DAC’s website that provides a single focal point for the industry’s bloggers.

“In conjunction with that, we are developing a database of expert speakers and well-known people who have been invited to speak and/or participate on panels at the conference. This database will be a very valuable resource for future DAC organizers, and it serves as a repository of names and expertise that companies can populate on their own initiative in order to be able to expose their technical leaders and high-level management to DAC’s organizers and attendees.

“But, this year’s Executive Committee is thinking on an even bigger scale,” Fix says. “In terms of marketing the conference and reaching out to a larger ‘user base,’ we are looking at all of the new media in the world today—everything from Facebook to YouTube, and all of those socialization and information tools that our children are busy with all the time. I hear about them from my own children and you probably hear about them from yours, as well. For the new generation of engineers—if we really want to reach them—we need to do it through the communication channels and tools that they use in their everyday lives. For example, we are considering inviting the authors of accepted papers to promote their work and encourage DAC’s participants to attend their paper presentations by uploading short videos on YouTube that would highlight the main contributions of their papers. The DAC website will contain links to these videos.”

Read the rest of the article

Will there be more startups in EDA?

Long considered the lifeblood of innovation, their focus, their exit strategies, and their funding sources are changing.

DACeZine sat down to discuss startups and entrepreneurship in the EDA market with some industry exponents each of whom have launched more than one in their careers. Participants to the discussion were: Rajeev Madhavan, chairman and CEO of Magma Design Automation; George Janac, CEO and founder of Silicon Navigator; Andrew Yang, president and CEO of Apache Design; and John Sanguinetti, chief technology officer at Forte Design Systems. What follows are excerpts of that conversation.

by Ed Sperling

Q: In the past, the majority of startups in electronics occurred in the EDA sector. What's so unique about this sector, and will the trend continue?
Madhavan: In EDA, the most important thing is the experience. You do not need a bunch of guys with different talents on a large team. As a result of that, things can be done quicker than in a lot of other areas. It's much easier to get the startup off the ground because of that. In addition, there's a lot more money in EDA from venture funding and other sources than there used to be 10 years ago.
Sanguinetti: Ten or 15 years ago, it didn't take as much money as it does today, though.
Madhavan: But now there also are a lot more entrepreneurs putting money back into EDA. That makes it a lot easier to get access to capital. The success of a few entrepreneurs makes it easier to get money—and to get some advice, as well.
Yang: That is an 'Easy come, easy go' model. It's easy to come into the market, but it has been easy to go out of business as well.

Q: Does it make good business sense to fund these companies?
Madhavan: That's a question that isn't always asked.
Sanguinetti: People like us are easy marks for other guys who are trying to start companies. It took me a long time before I got the point where I said 'No' to someone asking for funding.
Janac: The biggest change in EDA is the exit strategy. That eventually will change the capital that goes into it.

Q: How has it changed?
Janac: The number now is 2x. I'm paying some of the same people twice as much as I was paying them 10 years ago. One of the nice things about EDA is that you don't need an army of people to develop stuff. You do need an army to run the business.
Sanguinetti: But haven't the problems gotten harder?
Madhavan: The problems have gotten harder and the channel has gotten more expensive. Our customers expect a lot more in support than they did 15 years ago.

Q: What is the exit strategy for today? Who has gone public besides Magma and Verisity?
Janac: There may be two or three companies that could do it, but the number is now about $40 million in revenue.

Q: Is that still the number, or has it gone higher than that?
Madhavan: It's higher now. You now need to spend more money on lawyers and accountants with Sarbanes-Oxley.
Yang: About five percent of the profit margin goes to the IPO process, so if you can't get above 15 to 20 percent in your operating margin it's very difficult [to go public]. What's changed is that several years back you needed a big idea to take on a bigger TAM (total available market). But it's a maturing industry, so it's very difficult to get a big idea today. Entrepreneurs should focus on differentiable smaller ideas. They need to make money at them, be successful with them, and have several of them. Combined together you can do just as well.

Q: Will that attract the same caliber of entrepreneurs if the return on each investment isn't as high?
Madhavan: There's nothing wrong with five or six people getting together and selling the company for $20 or $30 million. There will be fewer very large acquisitions, and fewer at the lower end as well, because there nobody wins from an investment or entrepreneur perspective. But that also means you can't take too much money in outside investments. You can't take the highest amount available in series A, B and C and still be financially successful.
Janac: Nowadays, there are a lot of things you have to question if you're going to fund $5 million or $10 million in EDA because you need to know what you're going to get out. Acquisition is the most likely exit strategy. IPO is the most difficult. But there is going to be a new class, which is cash flow companies. So there will be a whole other set of startups that will not have an exit strategy.
Sanguinetti: In EDA, there are very few companies that can survive on cash flow. The only one I can think of is Denali.
Janac: The market isn't there yet. But it will be soon.
Madhavan: If you're a successful company generating cash, is there a big value in going IPO? Why not pay dividends to people?
Janac: Investors aren't ready for that.
Madhavan: Maybe we need a new class of investors.

Q: When all of you started down the entrepreneur path, was there a bigger reward at the end—or at least the promise of one—than there is today in EDA?
Janac: Even at $30 million, the founders and the employees will make over $1 million.
Yang: Financial rewards are not the only motivation in starting companies. It's one of them, but as the company progresses what's more important is to take care of the people to guarantee their financial return. That, along with the passion, is what drives people like us. Being driven entirely by financial rewards is very risky, because that means we tend to make decisions at the expense of other, often more important, factors.

Q: But if you're an entrepreneur, would you go into EDA today or would you go into Web 2.0 or some other nascent market that's ready to explode?
Sanguinetti: If you're an engineer and you've spent your life doing chip design, you're not going to do a Web 2.0.

Q: Agreed, but how about if you're just coming out of school and looking for a specialty?
Sanguinetti: It's pretty unlikely that a 24-year-old entrepreneur is going to come into EDA and be successful.
Janac: Unless they've done their Master's or Ph.D. in the field.
Sanguinetti: That's possible.
Yang: This is an interesting and different industry segment. It requires a lot of hard skills. You don't need an MBA to become an entrepreneur in EDA, but you do need a very deep understanding of the technology and the product trends. Then you need some soft skills to run the business. I don't think a 24-year-old would have those hard and soft skills.
Madhavan: That's why there hasn't been a 24-year-old coming in, taking over, and busting out like Google did. The chips are extremely complex and the customers are extremely demanding. It's not just one algorithm. No algorithm works out of the box in this business.
Janac: There are two dimensions to this problem. One is how to do a startup in an era of mega-bundling, where the big semiconductor guys are going to buy from one vendor. The other thing is what do you fund? One group of startups provides something that the mega-bundlers don't have. Another is focused on something that is deadly competitive. Another alters the game of how EDA will be done. Everything else is a me-too, and it's difficult to figure out how to make money out of this segment.

Q: So what happens to the entrepreneurial talent pool?
Yang: The flip side of solving a difficult problem is that we have to encourage people. Being difficult sometimes is good because it means there is a problem to be solved. It's worse if there were no problem left to solve. That would encourage an exit of people out of this industry.
Madhavan: That brings up another issue. If you're coming out of school, do you choose EDA or something else? In the late '80s or early '90s, we got them all. That's declining now because they might choose research in Web 2.0. That's our biggest challenge. The reality is that Wall Street has not rewarded this industry enough for the amount of work we've done. We need to sell that value proposition. If EDA were to shut down today there would be no chips and no consumer electronics. There's a big difference between the value we bring and the value we derive.
Yang: But is the value measured by how much money we make or the long-term impact on the semiconductor industry? Are you just going in for a quick hit where you take out money and the product never makes an impact? That is not a successful entrepreneur. 

Q: Aren't the problems that startups deal with relatively short-term, though?
Janac: No, the problems stick around in this field.
Sanguinetti: The scales change, the levels of abstraction change, but the fundamental problems of analysis and synthesis are always going to be around.

Read the rest of the article