The thing about an event like the Design Automation Conference is that it doesn’t really matter in what order you attend the events. Each panel, keynote, technical session, workshop, or tutorial is, in fact, a stand-alone presentation offering information and an opportunity to learn. You get to as many as you can, and hope that you’ve guessed right as to which set of events will optimize your experience. Unless, of course, you’re a vendor. Then, as one CEO of a small start-up recently told me, “You’re chained to the booth and attending any sessions whatsoever is out of the question.” That is a definitely a loss. There are great sessions going on all over the place at DAC, so pity the many vendors who were unable to attend any of them.
Following is Part 2 of my DAC Report. Part 1 was published on June 18th.
|[ Click to Enlarge ]|
The third day in San Diego was the first official day of DAC. The opening session early Tuesday morning was attended by upwards of 600 people gathered in the biggest hall in the Convention Center. Seated at the double row of tables on stage were the 20 members of the DAC Executive Committee in their distinctive yellow shirts. General Chair Steve Levitan welcomed the attendees to the conference and the meeting was off and running.
Before the Best Paper awards, scholarships, IEEE fellows, and lifetime achievement awards were celebrated, however, the session began with presentations in memory of U.C. Berkeley’s Dean of Engineering, Dr. Richard Newton.
Dr. Alberto Sangiovanni-Vincentelli stood mid-stage and admitted that talking about his lifelong friend and colleague was a painful and difficult task. Nonetheless, he spoke in detail about Newton’s work, his creativity and his drive, and their dynamic technology partnership. Alberto ended with a poignant: “Goodbye, Richard, my friend.”
A quote from Newton’s 1995 DAC keynote flashed across the screen: “If there is a single point I wish to make here today, it is that as a discipline, both in industry and in academia, we are just not taking enough risks today, and most certainly not from a technical perspective.”
Dr. Newton’s wife, Petra Michel, spoke next from the stage about their lives together, of their children and, in particular, of Dr. Newton’s work to improve the role of women at U.C. Berkeley and in the world.
Finally, Dr. Ellen Sentovich spoke about the Dean Richard Newton Memorial Professorship in Synthetic Biology that has been established at Cal based on a generous outpouring of funds from the industry. The first 30 minutes of DAC were emotional ones – and that tone continued on through the various awards and accolades for members of the industry who have contributed in every way to progress in the technology.
When it came time for the keynote address, Dr. Oh-Hyun Kwon, President of the System LSI Division at Samsung, took the podium and delivered a lengthy address complete with detailed statistics and slides – mandatory props given his huge topic: “A Perspective on the Future Semiconductor Industry: Challenges & Solutions.”
Unfortunately, those of us sitting in the first several rows could not hear him at all because of the acoustics in the hall, so I took my laptop and moved to the very back of the room. I sat on the floor and watched Dr. Kwon’s talk on the big screens and then could hear quite nicely. [The talk was taped and it would be time well spent to go listen to it online if you couldn’t be there.]
Above and beyond the staggering figures – 7.8 billion wafers currently rolling out of the fabs, with estimates of 15 billion wafers by 2015 – Kwon noted that only 100 products this year will actually enjoy a production run of over 100 million units: “The costs are therefore difficult to amortize over the lifetime of the product.”
Kwon acknowledged the global tech alliances required to push the next node into reality, whether it be 65, 45, or 30 nanometers: “The semiconductor market will continue to grow to new applications and higher performance demands. Huge capital expenditures will necessitate consolidation within the industry, and will increase the important of collaboration between key players in order to maximize ROI. Strategic alliances and standardization of both equipment and IP are going to be critical.”
Kwon ended his address with a stopgap solution to address today’s out-of-control design complexities. Forget SoCs where everything’s packaged onto one highly integrated (impossible to verify) chip. Instead go with: “Many-core designs and chip/package/board co-design!”
One of the highlights of my days at DAC took place immediately after the Tuesday morning plenary session, while dignitaries, honorees, keynote speakers, DAC Executive Committee members, etc., were milling around below the stage before rushing to their next appointment. I spoke at length in the emptying ballroom with U.C. Berkeley’s Jan Rabaey and IMEC’s Georges Gielen about the future of electronic design automation – initiatives that are nudging the industry towards system-level design and nanotechnologies, including carbon-based nano-scale systems.
Ours was a wide-ranging conversation that ended with a debate as to the optimal order of academic training, if it is to include electrical engineering, computer science, and biology. The debate was based on the premise that biological systems are offering some of the most compelling adjacencies’ with which to expand on applications of the knowledge base already established in design automation. My thesis was that biological systems are built on first principles in physics, chemistry, and self-replicating systems and, therefore, must be studied before man-made’ disciplines such as engineering and computer science.
Gielen and Rabaey disagreed with me completely. Their assertion was that engineering and computer science are tantamount to first principles’ when it comes to design automation. When it comes to creating and/or controlling carbon-based systems, optimizing design is the desired destination, not biological systems in themselves. Therefore, the disciplines of engineering and computer science must lead the curriculum, followed only then by formal introductions to biology and carbon-based structures.
The arguments on the Gielen/Rabaey side of the debate were compelling, but I was not totally convinced. I was convinced, however, that the greatest moment at DAC 2007 – especially for those truly interested in the future of EDA – was going to be taking place during Rabaey’s lunchtime keynote address on Thursday, June 7th. Unfortunately, I was slated to leave San Diego by end-of-day on Wednesday and was not going to be able to hear Rabaey’s address. The wise folks on the DAC Executive committee have posted Jan’s talk on the DAC website, however, and we can all go hear it now: “Design without Borders: A Tribute to the Legacy of A. Richard Newton.”
Late Tuesday morning, one of the most highly anticipated panels of the conference took place in Room 6C, “Mega Trends and EDA 2017.” Presenters included Argon Capital’s Jean Antonio Carballo, Synopsys’ Aart de Geus, TSMC’s Fu-Chieh Hsu, U.C. Berkeley’s Kurt Keutzser, and NEC’s Kazu Yamada. Given my usual scheduling conflicts, I was only able to sit in on Keutzer’s presentation.
Keutzer was happy to report that he published an article 6 years ago in EE Times that accurately predicted “Programmable Platforms will Rule.” Now Keutzer’s got a new prediction, and you probably won’t have to wait another 6 years to wait to see if he’s correct. Prof. Keutzer is predicting that the industry is in the process of abandoning its obsession with Moore’s Law, and is moving instead to a full and complete obsession with many-core processors. Briefly glancing through the amount of content in the DAC program that touched in one way or another on multi-core, multi-processor, multi-thread, multi-multi, and mega-multi, I’m predicting Keutzer is correct. In fact, his 2007 predictions for 2017 are probably already here.