Spotty Connections in Wireless
EDA vendors view market as big growth opportunity, but gaining a firm foothold is proving harder than expected.
by Ed Sperling
Wireless may look like a vast, untapped market opportunity for EDA vendors, but so far it hasn't proved to be a particularly lucrative one.
This is especially frustrating to the EDA industry, which prior to the 2001 downturn was considered a bright spot in a maturing electronics industry. But growth has been lackluster since then, leaving EDA vendors with two options --sell more tools to existing or new customers, and find new markets. They are taking both approaches very seriously.
With one billion handsets sold in 2006 and the number of wireless devices on the rise, wireless seems to be the clear-cut choice for the new market side of that equation. Dig a little further into the market dynamics, however, and the opportunity may be less robust than it appears from the outside. The problem is that most wireless chip makers have been developing their own tools for years, and as that market consolidates EDA vendors must either displace existing tool flows or find new customers in a consolidating industry. Chip makers say the dynamics in this market are starting to change, and there are indeed opportunities around the fringe of the wireless market. But exactly how big those opportunities ultimately become remains a big question mark.
Need for Proprietary Advantages
Paul Carson, Texas Instruments' director of VLSI design for communications infrastructure and voice, said his company expects to use more third-party EDA tools in wireless chip design as TI relies increasingly on foundries for its digital chip production and processes. But those merchant tools will be relegated to the fringes of TI's wireless design flow, largely because that flow was created before any other tools were available and the company wants to ensure backward compatibility. That means TI will continue to use its own tools for creating IP libraries and methods to increase hardware performance.
"We were the first to introduce Viterbi acceleration," said Carson. "We are certainly open to using outside tools, but in-house we have developed highly differentiated things. If something is considered a commodity, we're better off procuring that externally. We look a lot at make versus buy, and we do try to buy tooling because we would rather have our personnel working on things like 4G cell libraries."
That leaves the door open for standardized intellectual property (IP), but Carson said it would be harder for EDA vendors to make inroads in the core acceleration algorithms.
EDA vendors discovered this conundrum at the end of the last decade, when both Mentor Graphics and Cadence released their first tools aimed at the wireless market. Both were hailed as major breakthroughs by the companies that made them but sales were weak. Since then, Mentor and Cadence have upped the ante in places where they perceive the opportunities to be in the wireless market.
Segment Specific EDA Tools
Anil Khanna, field marketing manager for Catapult in Mentor's ESL division, said Mentor's focus is to take wireless into a higher level of abstraction. "Companies like Ericsson and Nokia are looking for the most efficient methodology they can find," he said. "It's how they go from their algorithm -- which is their secret sauce and where the innovation happens -- to high-level synthesis." Mentor's previous entry in the wireless field was Monet, which addressed behavioral synthesis. Khanna said the market was not ready for Monet, and Monet did not offer the level of benefit to users that the current Catapult architecture offers. "What we're addressing is how you enable hardware and software, embedded functionality and verification. We've seen more customer tools in verification than synthesis." Mentor is hardly alone in smelling opportunity in new generations of networks, new network protocols and more functionality and complexity. Cadence, which is making a play for the broadest segment of the market, has been offering Wireless Design Kits for the past several years.
"This is our second year with a methodology program," said Paul Colestock, product marketing director at Cadence. "We're looking at cellular as the major component of wireless. By 2008, the TAM (total available market) for wireless chips will exceed the market for PCs. By 2010, it will be an order of magnitude bigger. The challenge is that even if different vendors have their own tools, it comes down to interoperability. They need to be able to take up more and more of the design flow so that all products work together. We've got RF, analog/mixed signal and deep submicron digital."
Colestock calls Cadence's push "a long-term strategy," noting that problems were evident in the first-generation kits. "We will be continually refreshing this," he said, adding that Cadence already has seen an increase in business in this sector.
Synopsys, meanwhile, has taken an altogether different route into the wireless world. Its standard IP plays directly into the needs of companies such as TI, which would prefer to automate development of standards blocks wherever possible. Joachim Kunkel, Synopsys' senior vice president for IP and system-level solutions, said the real challenge is lifting wireless design to the system level, complete with design tools and methods that also are optimized for low power.
"What's different in wireless is vendors tell you how much information you can squeeze through a channel with the least amount of power, but they do not tell you how to implement that. It's up to the designer to figure out the algorithms. The algorithms are the crown jewels, and all of them are suboptimal because the optimal solution is too expensive."
Kunkel said that traditional EDA has focused at the RTL level and below, but most of the wireless design is being done in C or C++. Synopsys acquired Virtio in May 2006, in part to address this discrepancy. But Kunkel noted that to be successful in wireless requires picking the right slice of the market.
"Development cycles are long, investments are huge and companies want to recoup their investment before they look at new technology," he said. "On top of that, there is little room for disruptive technology because it is all based on standards. Where we see the opportunity is in creating virtual platforms for customers -- mostly silicon vendors -- and increasingly device manufacturers. It's all the things that go around the chips that we're looking at."