News and New Products

Voices: Magma Design Automation’s Rajeev Madhavan

Rajeev Madhavan discusses the evolving EDA business, the state of the semiconductor industry, and the future of education.

By Rick Nelson, Editor-in-Chief -- EDN, 2/3/2009

The evolving EDA business, the depressed economy, the state of the semiconductor industry, and the future of education in the US were all on the mind of Rajeev Madhavan, chairman and chief executive officer of Magma Design Automation, as he spoke with EDN.

What’s your impression of the state of the industry?

The semiconductor industry in the last few quarters has been in a really bad situation. October and November were the worst months I have ever seen in my life. Customers were just were frozen up, with deer-in-the-headlight syndrome. They just froze all purchase orders. People deploying new technology just stopped halfway through their projects.

What’s going to happen next?

I think people were shell-shocked given what the implications were of the two bad quarters. But now some companies are doing some analysis to determine what they are to become. I think a lot of transition is happening in our customer community.

Could you elaborate?

Companies are asking what things are optional—what things that aren’t really important? TI, for example, even wanted to sell some of its wireless business and has moved more and more into analog and mixed signal. [Texas Instruments announced on Jan 27, 2009, that it would not sell its merchant broadband business because it could not get the price it wanted (see "TI cuts 3400 jobs, no longer plans to sell merchant baseband biz").] Right now, semiconductor companies are shedding what they think is bad for them, and many are moving away from large digital SOCs (systems on chips) and toward mixed signal.

Why is that scenario occurring?

It’s just a reflection of the fact that just putting a digital SOC together by slapping in a lot of IP (intellectual property) into design and coming up with a chip that is sufficiently differentiated from the competition is very difficult. If it’s a pure standard chip that you’re deriving, you might have a little better cost than everybody else, but fighting ultimately on the cost margins in a recession is a bad idea for anybody. So, many companies have changed directions, and the focus seems to be on key architectural differences in the form of mixed-signal designs.

What application areas do you see the semiconductor companies moving into?

As the recession prolongs, our semiconductor-customer community will look at what Uncle Sam is investing in. Uncle Sam is going to invest in green energy—whether it’s electrical-grid control, electric vehicles, or more optimized gasoline engines—all of which require mixed-signal designs. A whole new category of electric vehicles are on the cusp of getting done and whether the automobile companies get some money or not will probably be tied at some point to whether they come out with electric vehicles. So, we are going to see big injection of capital from Uncle Sam in these areas, which means some of these companies will more and more migrate toward mixed-signal SOCs.

What are the consequences for the EDA industry?

EDA will have to gravitate toward chasing these new markets and not just focus on big digital SOCs. We [in the EDA industry] need to focus on the ability to put together mixed-signal designs in a very fast, transferable fashion. Most of our customers have gone fabless, so you are not going to see these specialized fabs that analog companies used to have. Now, our customers need to figure out how to do analog- and mixed-signal CMOS in such a fashion that designs can all survive and be transferred from one technology node to the other in an automated fashion.

Could you comment on Magma’s recent evolution?

Magma came out with a physical-synthesis tool in 2005, and then we went through a period where we had a lot of litigation expenses. But when we came out of that, we actually had a new tool, Talus, which represented a significant improvement, but it wasn’t stable when it came out in 2007. It took us a year, but I’m glad to say that, in June 2008, we finished the transition in terms of technology, and, as of November, pretty much every customer has migrated over to the Talus platform. But what has helped us through the transition period is our analog technology, which has kicked into very high gear. We have some unique specialized technologies that let us mix analog and mixed signal into one single chip with a single integrated tool.

How have you managed to fund R&D during the transition to Talus?

We have controlled our expenses, which is always a challenge when you have a lot of new technologies coming onboard. When the economy is in a recession, you have to fund initiatives almost by stealing from Peter to fund Paul. You cannot just make arbitrary investments; we have to be very agile in terms of investing in each of our technologies. We as a company have changed more and more to what I consider start-up mentality. Each group is very much fiscally responsible for only having resources necessary to complete its engagements—to take its product to the next level. Our investments have to be tempered with actual economic facts. In fact, we’ve gotten out of things that might help complete our product portfolio but will not result in a huge return on investment for our shareholders and our customers. Fortunately, completion of the transition to Talus frees up a lot of resources. We are going through optimization of our resources in such a fashion that only the necessary cost burdens are in place to handle the business we have.

You mention backing away from tools that would complete your product portfolio but not offer significant return on investment. Isn’t it important to be a one-stop company that offers everything?

I know there are EDA companies who say, “We can package everything and sell everything together,” and I think that’s never going to fly. Just having a “me-too” technology and saying “I can provide everything” is going to be a disaster for those companies in the longer term. A me-too technology has no value in EDA. The technology has to be differentiated. Packaging me-too technologies actually limits the price of each tool, limits the growth of the total market, and ends up hurting the big players who play that game. I think an economic recession will set them right. They need to realize that each product has to be profitable on its own merit.

But isn’t it important to have a single set of integrated tools?

Integration in my mind means one tool that actually sees all the way from the beginning of the design to the end, from RTL to GDSII, with full knowledge of parasitics and timing and other deep submicron effects. Companies may be claiming integration, but all they mean is that their field people have catalogs listing every tool they offer. That is not integration. No software company has ever succeeded by just offering everything.

What’s your strategy for handling the recession?

In this recession, we have to be more focused on staying back and saying, “What are the products where we can provide huge differentiation and provide huge productivity improvements for our customers.”

What are the challenges in the mixed-signal-EDA market?

If you look at the history mixed-signal and analog tools, there has really been only one player, which is Cadence. I participated in the development of the Cadence tools in 1991, and very little has changed. The flow is very manual. Analog designers are unique individuals with a lot of talent in terms of the circuit knowledge. But there was never an automated way of transferring the knowledge of a great circuit-design engineer to the layout.

What’s involved in the manual process?

Once you design a circuit, it must be simulated for its multiple corner cases for the deep submicron world, and that is a horrendous task. And once you simulate it and get a clean schematic, you call a bunch of polygon designers and you say, “Can you please do this in a given process?” Then, you find that you can never transition that product from an old process to a new one, so sometimes what happens is you have to sustain an older fab—which otherwise you would love to kill—just to produce that one part.

What’s the alternative to sustaining an old fab?

If you want to move same IP to a new process, you face a restart of the design, and if the architect of the chip is gone, tough luck, you’ll be starting from scratch. And the interesting thing is that, the second time around, you never get a better design.

What has Magma done to address the analog-design challenges?

So, what Magma has done, rather than competing head on with Cadence, has been to provide automated capabilities in our Titan tool. It is extremely fast and allows high-caliber circuit designers to capture a design, verify the design, and then in an automated fashion transfer it to a given process.

How are the prices for analog tools—is it a profitable business?

The good news is, in mixed-signal, the pricing of the EDA tools is pretty good because of the monopoly that has existed. And, as I said, we don’t really compete with Cadence; we provide a very different methodology. Yes, we have tools that provide some of the same capabilities, but that’s not what we are selling. We are selling a unique and value-added technology. So, I think the market will grow in mixed-signal as the number of chips in this area increases. By the by, I’ve always disagreed with the argument of a lot of people, who have counted the number of design starts and said, “Oh the digital tools are going down because the number of design starts are going down.” It’s not the design starts that makes any difference whatsoever in the number of licenses that are sold, be it mixed signal or digital. It’s chip complexity that matters.

You established a presence in Beijing and Shanghai last year—what’s the status of those operations?

We have tamped down those types of initiatives somewhat. During a growth period there is a tendency for every company including Magma to start offices in different places and look at R&D cost optimization. But in today’s economy, what you need to look at is closer-knit teams being more productive. We’ve gone for more, higher quality people—not higher quantities of people.

In fall 2007 you sat down with an Electronic Business editor and some other executives to discuss education (see "Where do engineers get the best education?"), and you said that, in the United States, too few engineers are graduating and that math and science have to be elevated in the middle- and high-school levels. Have you seen any improvement over the past year?

Some of the problems of finding enough engineering graduates to hire have been eased by the recession, but the reality is that as far as education in the United States goes, we’ve got a lot of work to do. I see it with my own kids—even though they attend a good private school, there are a lot of improvements that need to be made.

What are the prospects for making those improvements?

I am very hopeful of what the new administration is going to do. At least they are cognizant of the problem that exists—that there is very little emphasis on getting kids to go into technology areas. It’s very difficult to get kids excited about engineering, and that is a very big challenge that we as a country face. We need to be the creators of newer and newer technology—whether it’s in the area of electric vehicles or some other field. We need people skilled in biotech; we need people skilled in green energy aspects. In many of these areas we need what I call mixed engineering/educational systems, which are still very lacking. Instead, we have traditional engineering focuses in most places. It’s time we changed that and got back in leadership in every area of education—from kindergarten to college engineering programs.

What’s the solution—spend more money?

Money is one aspect. In addition the metrics of how we measure student performance are somewhat arbitrary. And our teachers don’t foster competitiveness. We want to accommodate the sensitivities of our students, but the competitive spirit is sometimes good. When our students graduate, they’ll be in a competitive world. I’m not an educator, but I think we can introduce that competition much earlier in the system without having a negative impact on the kids.