You need all the pieces...........................
192.156.97.113
Oct 31 1997
EB 11/97: EDA face-off Q: What role should electronic design automation companies play in the intellectual property business? Walden C. Rhines (left), president and CEO of Mentor Graphics Corp., and Joseph Costello, former president and CEO of Cadence Design Systems Inc., express their views below. (Rhines photographs by Steve Bloch, Costello photographs by Robert Holmgren) EDA face off There's no arguing that the intellectual property (IP) business is growing rapidly. The market for system-level macros -- the building blocks of chip design -- is increasing at a compound annual growth rate of nearly 46%, according to San Jose-based market researcher Dataquest. As the chip industry moves toward building systems on silicon, these macros become more important. Not surprisingly, electronic design automation (EDA) vendors are looking to capitalize on this trend. ÿÿÿÿÿBut the waters are murky regarding the role EDA companies should play in this emerging business. Indeed, some chip customers are worried that their EDA vendors might become competitors as the EDA vendors expand into design, consulting and integration services along with their traditional offerings of software tools. After all, EDA companies have all the pieces of the puzzle at their fingertips--tools, design expertise and access to the IP to build products. ÿÿÿÿÿAt the center of the debate is who should own and manage the IP now and in the future. ÿÿÿÿÿWhat's the thinking among EDA companies? Electronic Business asked two heavyweights--Joseph Costello, formerly president and CEO of Cadence Design Systems Inc., San Jose, and Walden "Wally" C. Rhines, president and CEO of Mentor Graphics Corp., Wilsonville, OR--to give us their views on the future of the IP industry and the role that EDA companies should play. Ann Steffora By Joseph Costello Former President and CEO Cadence Design Systems Inc. San Jose, CA The advent of system-on-a-chip has created a lot of churn in the electronics industry as the various players scramble to stake out their turf in the new world order of how electronic product designs are architected and implemented. Semiconductor companies begin to look like systems companies, and vice versa. And design automation companies look for ways to leverage the opportunity afforded by unprecedented complexity and time-to-market pressures faced by both those sets of customers. Before we can figure out the right next step for implementing system-on-a-chip, we need a vision of where we are going. This vision of the future must go beyond the technical foundation of how things will get done. It must also address the roles and associated core competencies that determine who will get things done. Sound complicated? Well, in this case, it's actually quite simple. The electronics systems market has already dealt with many of the challenges we are now talking about. This includes how designers get access to an adequate supply of competitive components, how they mix and match (reuse) those components, how they integrate both third-party and proprietary software and even how they value and price those components. As we talk about true systems on silicon, why would we think that the world must look so different? Sure, the components must be virtual instead of physical, but so what? Just like the systems of today, the future of systems on silicon will require open access to the best virtual components in the market, designed to mix-and-match with hardware and software interface standards, priced at market value and at each point in time (not a locked-in lifetime royalty), for each component shipped in a system, and of course, with the usual volume discounts. Role playing With this view of the future, it's important to understand the roles of the various players. First, are the traditional "component" suppliers--semiconductor companies who want to leverage their unique IP across broader markets. For them, the notion of mix-and-match extends the opportunity for them to sell their products to more customers (acknowledging that their products may not be physical components). Then there are systems companies which may also have their own IP that they may want to commercialize on the open market. Both these groups can also be virtual component consumers, most notably acquiring building blocks from the emerging industry of independent core suppliers which specialize in a unique application or technology. In all three cases--the semiconductor company, the systems company and the IP company--there is a product to be sold. This fact represents the primary difference between the role those three factions play in the system-on-a-chip world, and the role design automation companies should play. "Just because the integration substrate for a system has changed from a PCB to silicon, why would one conclude that EDA suppliers are suddenly an optimal source for the next generation 'virtual' 3-D graphics chip sets, low-power RISC processors, MPEG-4 decoders and CDMA chip sets for PCS?" Just because the integration substrate for a system has changed from a printed circuit board (PCB) to silicon, why would one conclude that EDA suppliers are suddenly an optimal source for the next generation virtual 3-D graphics chip sets, low power reduced instruction set computer processors, Moving Picture Experts Group 4 decoders and code division multiple access chip sets for personal communication systems? The systems world saw the best component suppliers succeed by applying tremendous expertise with tremendous focus. None of the application-specific standard part (ASSP) companies like Cirrus, S3 and C-Cube, providing the valuable physical sub-systems of today, were founded from EDA experts or are part of an EDA company today. Most were founded by systems experts from systems companies, who relied on other specialists (e.g., EDA vendors, foundries, silicon design services) to get their IP into a form that could be leveraged. As those physical components now become the virtual components of system chips, why would this change? The commercial design automation world needs to focus on being the facilitator for system-on-a-chip: providing a new set of tools and methodologies that help authors of virtual components design and productize for reuse, while allowing integrators to rapidly evaluate and interface the co mponents. But, will the tools and components be enough? Unfortunately, even if the external component and EDA suppliers make the necessary adjustments, there is a tremendous shift required for most companies. Systems (and ASSP) companies are not only dealing with the blurring of system and IC organizational boundaries and design flows, but at the same time must deal with new consumer-oriented design cycles (including the need for rapid product derivatives), and the convergence of multiple types of applications coming into the same systems. These companies will need to make tough choices about where to focus, and find new partners for design. They will need help mixing and matching the best internal or third party systems and component IP, spanning multiple application domains, into convergence system-chips, within the time constraints of new consumerized markets. They will need continued access to the best tools, the best flows, the best methodologies and the best designers, and then have access to the best IP available in an open market to realize their dreams in silicon, before somebody else does. Focus vs. defocus Selling an electronic product--virtual or otherwise--is not a business Cadence intends to be in for a number of reasons, not the least of which is a potential to defocus from our core competency. Instead, Cadence has chosen to direct all of its focus on helping companies realize their visions of system-on-a-chip through architecting, building and, sometimes, running design factories. This includes state-of-the-art EDA technology and tools, efficient design methodologies and a full range of services to assist companies in getting the right product to market at the right price in the right timeframe. This does not include selling a "product" that will eventually make its way into an end product system-chip. Customers don't need EDA suppliers to contort themselves into the best IP suppliers. They need EDA suppliers to provide great authoring and integration tools and advanced methods, supporting open standard data formats so that one set of data for a virtual component can be used in any design flow. And they need neutral design partners to help them turn the big three obstacles ahead into the big three opportunities: the consumerization of electronics, the convergence of multiple applications and the siliconization of complete systems. By Walden C. Rhines President and CEO Mentor Graphics Corp. Wilsonville, OR A revolution is underway. It's a movement from custom-designed chips tied to specific manufacturing processes to reusable IP--the pre-set hard cores, reconfigurable soft cores and physical libraries that are critical to the next generation of wired and wireless communications, and computer and consumer applications. A key to winning in the revolution is knowing who is your friend and who is your foe. If EDA companies focus on what makes designers successful, they will provide the maximum benefit to the electronics industry and they will make reuse a reality. They will be your friend in the IP revolution. "EDA companies will work with the independents to make their IP supportable on a worldwide basis through reseller agreements or service contracts." If, instead, they choose to focus on outsourcing, performing systems integration, using IP to create systems-on-a-chip, or (heaven forbid) providing turn-key designs and wafers or chips, then the EDA companies will become your foe. They will become simply another semiconductor competitor and a poor competitor at that, because they will have strayed from what they do best--freeing customers to focus on differentiable designs and mitigating risk for them. Semiconductor companies will have the broadest role in this revolution. They will continue to be the primary creators of IP and will provide their primary value through system integration into standard and customizable systems-on-a-chip. In addition, they will make their IP in all forms--as chips, or as reusable design building blocks. System companies will continue to be the leading users of IP, as system integrators. They will create their own proprietary IP, as well as acquire IP from others. Occasionally, they will market their IP as software, to drive industry de facto standards, or to support specific customers, but usually their value will be realized in the system products they sell. New industry A whole new industry of independent IP creators will spring up. Their expertise will be in system or chip design for particular vertical markets. They will provide many of the most innovative design approaches and will market their products as reusable design IP. Their weakness will be supporting the hundreds of "views" required for IP to be reusable in a wide variety of customer design environments. They will lack the infrastructure to provide worldwide maintenance contracts, on-site support and engineers to provide special interfaces, or variants, of their IP. EDA companies will fill this void. They will work with the independents to make their IP supportable on a worldwide basis, through reseller agreements or service contracts. They will provide design tools, reuse methodologies and libraries of standard building block functions so the designer can focus on that which can be differentiated. What's in the arsenal? What do designers need from their EDA suppliers to win the IP revolution? Help securing customizable standard building blocks of IP that are readily accessible, testable and verifiable, and manufacturable. In our 10-plus years of experience providing customers with physical, soft and firm libraries, we have found that support, access and lower cost are the critical values. EDA companies can take up where the independents leave off. They will work with the IP independents to make their IP supportable on a worldwide basis, through reseller agreements or service contracts. They will provide design tools, reuse methodologies and libraries of standard building block functions so the designers can focus on developing the latest products instead of translating VHDL to Verilog, adding built-in self-test, and so on. The strategic arsenal required to win the IP revolution includes: Design tools and a reuse methodology so that designers can create their own reusable IP Sources of standard building blocks of IP (e.g. standard processors) so that designers can concentrate on assembling IP building blocks into differentiable designs A way to distribute, and provide support for, the IP that designers want to make available to designers at other companies A robust set of verification and test tools that meet the challenges associated with deep submicron (for instance, databases of hundreds of megabytes) and results in chips that work at first pass Support for the IP they acquire from other companies. This means 24-hour hotlines, worldwide application engineering, on-site support if needed and the hundreds of "views" required to make IP reusable in the wide variety of design flows that must be supported. EDA companies are uniquely capable of stocking this strategic arsenal. The EDA industry is a prime force in driving standards for reusable IP through the Virtual Socket Interface Alliance (VSIA). Last June, Mentor and Synopsys Inc., Mountain View, began moving the VSIA standardization effort to real applications by joining forces through the Design Reuse Partnership. Our two companies are publishing a practical "how to" design reuse methodology and applying it to the world's largest independent library of reusable soft IP processor cores. Now, a question to ponder: How are you going to verify, let alone use, this abundant IP? Again, we turn to the EDA suppliers. Most of the industry is developing tools that will facilitate system verification, now the largest single barrier to system-on-a-chip design. Allied forces You need a friend, not a foe, in the IP revolution. EDA companies will make excellent allies by focusing directly on the designer, contributing maximum benefit to the industry as a whole, and ensuring that reuse becomes reality. |
