Changing EDA-IP Relationships

Gary Smith at DAC 2014: How System Design is Changing Electronics

Everything you ever wanted to know about EDA but were afraid to ask! The basics on the business and competitive environment in the Electronic Design Automation market.

Price: $1,000

Download PDF

As Mark Twain once said “The reports of my death have been greatly exaggerated”, and so goes it with DATE. The move from Nice, and before that Paris, to Grenoble was as successful as the move from Munich to Dresden. Not that I won’t miss Paris and Munich; however both Dresden and Grenoble are semiconductor towns and therefore more fitting for an IC Design conference. Both Dresden and Grenoble are beautiful cities in their own right so we have lost little of the romance of Europe. It is a bit of an adventure getting to Grenoble, reminding me of my flights to other ski resorts, but well worth the trip.

Biologically-Inspired Massively-Parallel Architectures
Steve Furber, from the University of Manchester, gave a great talk on the future of computing. His topic was the most advanced computer known to man, his own brain. He took us from Baby, a 1948 vintage computer to today. Did you know that the ARM 968 is 50,000,000,000 times more power efficient than Baby, the name of that 1948 computer? Not bad, but at the same time, not nearly good enough. He unfortunately showed the out-of-date Cost of Design gap chart that was popular in the mid-1990s; I’ll address that on my third Design Research Viewpoints in May.

If you look at the brain you will find it has:
1. Massive parallelism (1011 neurons)
2. Massive connectivity (1015 synapses)
3. Excellent Power efficiency

It does this by using:
1. Low-performance components (~ 100 Hz)
2. Low-speed communications (~ meters/sec)
3. Adaptability – tolerant of component failure
4. Autonomous learning

A extremely thought-provoking keynote

How technology R&D leadership brings competitive advantage
Philippe Magarshack, Group VP Technology, STMicroelectroncs gave an important keynote on how STMicro is addressing the challenges of being an IDM (Integrated Device Manufacture) in this new world of Fab-Lite manufacturing. I was able to spend some time with Phillip going into further detail. He brought up two of what I consider myths that are circling around the semiconductor arena. The first was that you can’t automate the analog design flow. Phillip has been trying to find ways to automate at least part of the Analog flow. I told him about a presentation that I gave at EDP 2001 showing that an automated analog flow is possible. The next subject was the future of IDMs. I’m really getting tired of hearing analyst proclamations that there will be only three Fabs companies in the future. The reason of course is that fabs cost $6 billion dollars and only a few can afford to spend $6 billion on a fab. My prediction is that we will have around 50 IDMs in the future and that the answer to the $6 billion fab question is to build billion dollar (or less fabs). Only foundries, memory manufacturers and one microprocessor vendor that I know of need $6 billion fabs. Now I do believe that if we keep going in the direction we’re going we may end up with only three fabs, but then the semiconductor world has a lot of experience in changing direction. I’ll be writing about this in the near future.

DATE Continues to Grow
Conference attendance was up by 8%, but what impressed me was that DATE had to open up another Exhibition Floor to accommodate the number of new companies with booths. It was well they did for exhibition attendance grew by 40%! One of the interesting comments, that seemed to resonate with most of the attendees, was that the non-presence of the large EDA booths made it much easier to find the small, new start-ups where most of EDA’s innovation comes from.

All in all another great DATE.