|
|
EDA Confidential
Ravi Subramanian: by Peggy Aycinena March 25, 2009 The principle subject of the DVCon 2009 Executive Panel in San Jose in February was "EDA: Dead or Alive?" The seven speakers on the panel were unequivocal in the answer: EDA is definitely alive, because it’s the single most important pre-requsitive for progress in electronics. The electronics industry will never die and, therefore, neither will EDA. In the weeks after the event, I spoke at length with all seven panelists individually to follow up on various topics discussed at DVCon. This article recaps my conversation with Berkeley Design Automation executive, Ravi Subramanian. Links to the rest of the conversations in the series can be found at the end of this article. Setting the EDA roadmap ... Ravi Subramaninan – In my view, as long as electronics is alive, it means electronic products must be designed and maintained, and therefore EDA is alive. With what the electronics companies are going through today, however, the face of EDA will certainly change. There will be some death [in the industry] and some new life. Certainly, I am cautiously optimistic about the future, although I think this is going to be a pretty long downturn and some technology sectors in EDA will not survive. Those companies targeting markets of less than a half billion dollars will be the most challenged in the next 12 to 18 months. They’re going to have a very tough go. Markets need to be big enough to continue to draw investments from both the EDA company and the customer, working together on problems that really matter. Q – How do we know which problems those are? Ravi Subramaninan – There are multiple trends in electronics to track. There are many vertical markets, communications, computing, networking, wireless, and so on. For each of these, the electronics and semiconductor industry is dramatically changing. Who are the top three companies, and who constitutes the rest of the pack? That list of companies will dramatically change. It’s accepted that Qualcomm, NXP, and MediaTek are the leaders now, but just 3 years ago it was TI, Philips, and Infineon. For EDA companies who do not yet have these companies as customers, who do not have strong collaborative relationships with these companies, those vendors need to dramatically change these relationships. If you take wireless for example, and the types of platforms involved, there’s the ultra-low-cost segment and then there are smart phones. There is a dramatic difference between the chips in these devices. Understanding the different market segments is especially important for EDA companies to be successful. Similarly in the memory market, NAND Flash had certain design flows at 90, 65, and 45 nanometers. Below 45, from our point of view, we’re seeing dramatic changes in design and verification because of problems that never existed before. In NAND flash memory, the portion of the die area doing analog work is becoming significantly larger. Those circuits need to operate at very large voltages, even though the power supply to the chip is very low. You need 10 to 20 volts to program the analog, but you only have a 1-volt supply. You have to use charge-pump circuits to get those high voltages, and you need on-chip precision. These are big challenges and are keeping a lot of analog engineers busy today. Also in NAND Flash, the I/O is getting faster and faster because of the high data rates coming in. What used to be a few megabits per second is now tens of megabits per second. You need higher-speed circuits built onto the chips, with bits moving on and off the chip reliably because of error protection rates. There is also stacked-die technology in memory. To get those chips stacked and working, you need a type of analysis to get good estimates of reliability and to measure catastrophic failure limits. Who is providing that functionality today? No commercial vendors, yet. Typically it takes 18 months to flesh out new design systems and tools. These things will need to be done in concert with the leading flash memory providers. The types of products that are high volume/high margin are significantly changing with all of this, and companies have to dramatically change the mix of what they’re designing as a result. Companies are completely revamping their products to address the emerging markets. The best example is TI. They used to have a high-performance business unit, but have pulled out a separate business unit to just deal with analog. Q – So everything is changing. How does a vendor respond? Ravi Subramaninan – The number one thing we need to do is to be very aware of underlying trends that drive change in these platforms. For example, RF CMOS has been a big buzzword for a long time, building RF circuits on digital CMOS. It’s the right solution for certain products, but not for all products. Putting everything on CMOS doesn’t mean it’s necessarily cheaper, although CMOS is important for ultra-low-cost markets in rural India, China, and Africa where the silicon must cost less than twenty dollars – the cost of a handset in those places. The most successful companies will be very savvy about trends and end markets, and have the ability to develop [relevant] tools. EDA companies need to look at implementation and design complexity. They must focus on this type of analysis to survive. Q – Meanwhile, are there more and more EDA companies today going after fewer and fewer customers? Ravi Subramanian – The number of companies designing products for the electronic community will dramatically decrease, with consolidation being the reason. The number of engineers doing design is decreasing, and things will have to go through a cycle of death and rebirth as a result. New companies will be started, but it will be a long and painful process. It’s going to be a bloodbath. Q – Who’s going to survive the bloodbath? Ravi Subramanian – The answer to that question has three parts: 1) Who ever is financially strong enough to survive the downturn. That tends to favor the larger companies. 2) Whoever has differentiated technology that’s critical for the success of the semiconductor companies. 3) Whoever is able to adapt and move the quickest at the industry dynamics play themselves out. Numbers 2 and 3 tend to favor the smaller companies in EDA. If you look at several key areas in EDA today, there are startups providing excellent solutions. The customers are making suggesting, and the vendors are providing cost savings to the customers as well as productivity improvements. The big companies are able to this in certain areas as well, but they need more time to respond. The larger companies have the capability to meet the needs of the larger customers, but often it’s the small company that initially identifies that need. The smaller EDA companies can identify the pain because we are closer to the customers. We’re able to identify the problems because of that closeness, then respond quickly. It’s not just market research, but it is discovering the problems and creating the solutions in partnership with the customers. Q – What markets do you compete in? Ravi Subramaninan – Signal-to-noise ratio became significant at 90 nanometers and below. Signal levels were dropping, but noise levels were not. We’re recognized by a number of leading semiconductor companies for providing silicon accurate noise analysis technology, and have the leadership in the industry in this area. We’re using stochastic methods and applying new mathematics to electrical circuits. Q – What kind of new math? Ravi Subramanian – Simply put, the traditional analysis of ICs using SPICE technology relied on making circuit behavior liner, and ignoring flicker and white-noise types of impairments that come with semiconductor devices. Our technology solves these types of circuit equations by not making linear assumptions. We retain the nonlinear nature of the problem, and keep track of the stochastic noise of each transistor. We do create more complex equations to solve, but we have the technology to solve them and to yield the type of analysis you could not do before. Q – What’s the difference, in your opinion, between today and ten years ago in the way you describe things? Ravi Subramanian – Things have changed a great deal. Ten years ago, the critical mass required for a smaller company to survive was much less. Today, the big EDA customers ask, why can’t we just buy this stuff from Cadence or Synopsys. It’s a heroic task for the smaller companies to compete in that market. Ten years ago, the companies were willing entertain smaller [point] solutions from smaller companies, but buying one copy of a tool is not enough today to keep an EDA company in business. Today, however, it is almost impossible to dislodge the leaders. If you are a student of Clayton Christensen, look at the structure of the EDA industry and the history of the evolution of semiconductor complexity. Much of the change in EDA happened in earlier decades at a point in time when evolution was happening quickly. New companies could be born effectively then to take advantage of that evolution. Over the last 10 years, however, the amount of complexity has reached a point that the big EDA companies control the design platforms. The EDA technology the customers purchase has to fit into those platforms, whether they’re from Cadence, Mentor, Magma, or Synopsys. For the small EDA vendors, fitting into that platform is the key to getting adoption. This whole situation was very different 10 or 20 years ago. Q – So, the Big 3 companies control the platforms and evolution in EDA – is that good for the world? Ravi Subramanian – In the most global sense, it may be good for the advancement of humankind, but I’d say it’s not an ideal situation for innovation in EDA. It’s the deck of cards we’ve been dealt, however, and we need to deal with it. We can be resigned and say it’s not possible to innovate in this environment, or we can focus on growing a sustainable business. Before, you had to have an exit strategy as an EDA startup, either IPO or being acquired. Today, you have to be able to build a real business, instead, and prove that it’s healthy. You do that by going after the market segments that are growing. We believe we can do that by focusing on building a successful business. Our number one job is to survive and adapt. Bio ... Ravi Subramanian is the CEO of Berkeley Design Automation, a private EDA company recognized by leading semiconductor companies for it’s high-precision unified verification platform for analog, mixed-signal, and RF IC design. Ravi received his BSEE from the California Institute of Technology, and his Ph.D. in EECS from the University of California at Berkeley. He brings experience in the wireless communications, semiconductor, and EDA industries with leadership roles at TCSI, AT&T Bell Laboratories, Synopsys, Morphics (acquired by Infineon). ____________________________________________________ The DVCon 2009 Panel Series * Atrenta’s Ajoy Bose & Mike Gianfagna:Recrafting the concept of EDA * Javelin Design’s Diana Feng Raggett:Government funding for R&D in EDA * Synopsys’ Gary Meyers:FPGAs versus ASICs * EVE’s Lauro Rizzatti:EDA in Europe * Berkeley Design Automation’s Ravi Subramanian:Setting the EDA roadmap * SpringSoft’s Scott Sandler:EDA dead or alive? * Calypto’s Tom Sandoval:Reaching the EDA customer ________________________________________________________ Peggy Aycinena owns and operates EDA Confidential: Copyright (c) 2009, Peggy Aycinena. All rights reserved. |