DAC 2015: Day 5

I feel like the last man standing. The show floor is closed, most of the industry folks have gone and the other press is nowhere to be seen. The good news is that we are still here to cover the events of DAC and to bring you the whole show. The morning starts with a keynote entitled “Electronics for the Human Body.” John Rogers from Univ. of Illinois at Urbana-Champaign talked about the mismatch in properties between rigid silicon and soft and flexible biology.

He concentrated on integration with the skin although he pointed out that the same techniques would apply to many organs within the body. “The temporary tattoo would be a great vehicle,” he said. “It would be super thin – about 5um, light, stretchable and water permeable and waterproof and of low modulus.” Rogers talked about the option that could make this a possibility including nano ribbons of silicon. This keynote will appear in the research corner in the coming weeks.

Following this, in a co-located conference – ESLSyn, another keynote given by Forest Brewer from UC Santa Barbara titled “Trends and Open Problems in High Level Synthesis.” He categorized HLS as primarily being a way to explore RTL through a semi-automated mapping from some higher level model. But there are things missing in the flow and the models were badly defined meaning that the problem was made way more difficult than it should have been and because of that, the algorithms don’t scale well.

He pointed out that most of the time in CAD, the problem has been better defined and efforts have been made to reduce the complexity of the problem. “Can a middle ground be defined?” asked Brewer. Without an equivalence model, “benchmarking and pragmatic comparison becomes almost impossible.”

The problem is that variables in procedural languages are badly defined and because of this, external sequence matching becomes problematic. This observation was enough to cause Brewer’s research to change and for him to concentrate on interface definitions and protocols. The answer to all of this, according to Brewer, is memory and knowing the duration of variables. Would this put constraints of the software? Yes, but it may be worth it and companies like nVidia are doing a version of this in their graphics processors.

For the final day of DAC, the SKY talks moved from the show floor to the academic area. This didn’t seem to affect their popularity. The final one was titled “Back to Hogwarts – On the Dark Arts in Hardware Security”, presented by Saverio Fazzari from Booz-Allen.

How does one deal with supply chain issues and counterfeit devices? This is very acute when dealing with obsolete and replacement parts used for system maintenance. What may be surprising is that the US is a part of the problem.

Fazzari categorized attacks as being either design, hardware or logistics attacks and the programs (TRUST and IRIS) created to address them. Others programs are dealing with counterfeit detection and finding their origins. SHIELD is attempting to make counterfeiting too expensive or too difficult.

When asked about privacy, he admitted that the DOD and privacy did not have a good track record, but that you should think differently about the creation of technology and the way it is used.

The final session of the day and of the show was titled “Developing Safe and Reliable Automotive Electronics System.” Presenters from Infineon, Freescale and Tesla talked about the impact of power, noise and induced failures as weighed against the desire to integrate increasing amounts of electronics into the car.

And with that DAC is over for another year. While Anne Cirkel did not meet her stretch goal for attendance, she should be very proud of what she accomplished this year. She has set a high bar for next year in Austin. To bring you the details of the events, all of the editors at SemiEngineering now have many hours of transcriptions ahead of us. For those of you who could not attend, we will be bringing the most important aspects of the show to you.