Wednesday, December 15, 2010

Mentor’s Wally Rhines on global EDA industry challenges – II

This is the concluding part of my discussion with Dr. Walden (Wally) C. Rhines, chairman and CEO, Mentor Graphics.

EDA’s role in modeling and photomask correction
I asked Dr. Rhines about the future of EDA’s role in modeling and photomask correction. He said that in just a decade, resolution enhancement has grown from zero to over $200 million in annual revenue for the EDA industry.

“Almost all of this revenue is concentrated in two EDA companies. The value of this EDA software is clearly recognized by manufacturers. Mentor has many partnerships with manufacturers and a joint development program targeting 20nm resolution enhancement with IBM.”

Handling 22nm and sub-22nm levels
Next, with new process technology nodes becoming quite the talk of the desgin community, what does EDA now need to do at 22nm and sub 22nm levels.

Dr. Rhines said: “We have been working with our customers on this for quite some time now and are in fact well down this path. We think that most of the problems have been solved, or are solvable. Obviously, most of the issues here revolve around the lithography and manufacturability, but the EDA industry has been leading this since optical proximity correction became a key technology for the fabs quite some time ago.”

Density area savings
In an earlier discussion, the issue of how compelling would integration density area savings remain by going to new nodes had come up. I have to repeat this question, as it still seems to remain an issue.

So, how long will the integration density area savings you get by going to new nodes remain compelling?

“Hard to say!” noted Dr. Rhines. “We can see a path to 15nm with the traditional 193nm immersion lithography, and we usually surprise ourselves in our ability to go farther than we think we can. However, even if density slows down, this is but one way to achieve the continuous performance improvements that we’ve seen over the years in silicon.

“3D silicon, for instance, holds the promise of allowing us to continue to grow performance without necessarily doing it by just continuing the process shrink. Logic and memory have been on a predictable “learning curve” since the vacuum tube and I don’t expect that learning curve to deviate anytime in the foreseeable future.”

What about verification?
One of the challenges is that verification is falling further behind. What is the EDA industry doing to handle this situation?

He said: “Well, innovation isn’t continuous, it often comes in big jumps, and in the last couple of years, we have really made some great advances in verification. First, we have developed an intelligent testbench automation that can really reduce the time to achieve coverage objectives by orders of magnitude.

“Second, emulation is playing a more critical role at the leading-edge because of the orders of magnitude in performance improvement it can offer over simulation.

“Lastly, the development of OVM (Open Verification Methodology) and the industry standard verification methods around SystemVerilog have also improved the situation. Of course, designers need to learn these new tools and techniques, but the answers are out there now.”

Who will survive in the EDA industry?
During his presentation at the recenrly held U2U India, Dr, Rhines had pulled up a slide where Gary Smith, a noted EDA expert had answered a question: Will big EDA companies (Cadence, Mentor, Synopsys, Magma) keep going or change course?

Gary Smith is said to have replied: “With the exception of Mentor, they will change course or fade away. Mentor is well-positioned for the future, the rest are not.”

What makes Gary Smith say that Mentor should survive, rather than the other EDA vendors? Why won’t the other EDA vendors survive?

Dr. Rhines remarked: “They probably will! Gary’s point was that Mentor’s strategy gives him confidence, while other companies are still evolving their approaches to the future.

“Our perspective is that, to grow as a company and an industry, we need to find and solve new problems. That’s why we have spent a great deal of effort in translating our EDA expertise into adjacent markets like automotive networking where existing development and verification techniques just aren’t keeping up. We are confident in Mentor’s future.”

To be concluded.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.