Analog Design and ESL: We Are Not There Yet.

Two of the three articles in this month's issue deal with analog design challenges.  The third one hails a new season in ESL.

Analog Design
It is well known that analog issues have become important even in digital design.  So important, in fact, that tools developed to aid engineers to solve this type of problems have seeded a new and growing market segment, DFM.  Companies in this segment are building a bridge that allows designers trained solely in digital design to deal with the physics inherent in all electronic design.

Do We Need It?
Geoffrey James's article revisits the issue of analog synthesis and the resistance of the vast majority of analog designers to new tools.   The statement by Ashutosh Mauskar, vice president of product and business development at Magma Design Automation, brought me back to the late 1990s, when digital synthesis was being introduced as a commercial product.

In those days, the most prevalent justification by designers for not using the tool was that they could do better than the tool.  And in many cases, but not the majority, that was true.  But the tradeoff was time.  E ventually, acceptance grew, the products matured with greater use, and now synthesis is a natural step in digital design.
Looking back at commercial offerings from companies that pioneered analog synthesis (and went out of business for it), it is easy to see how analog design creation based on parameters manipulation does not solve all of the issues inherent in analog design.  But, as Gary Smith recounts, even this form of analog synthesis may address a significant portion, probably more than 50%, of analog designs, especially when a specific instance of a frequently used analog component is required.

Of course, as the article points out, developers of standard analog parts enjoy high rates of return , so the economic incentive for well paid designers to adopt new tools or methods is missing.  But how long are we going to continue to integrate discrete analog components in our designs?

I believe that what is needed is a fresh look at the analog design methodology.  Let's not try to copy the digital flow.  Just because it worked there does not mean it will work here.  DAC offers the perfect environment to discuss new ideas in analog design methods; I am looking forward to seeing what the dynamics at the 45th DAC in Anaheim will foster.

A Lack of Consensus
The roundtable hosted by Ed Sperling gives an indication of the economic difficulty of introducing a new tool in a market, the analog design market in this case.  One of the issues that has intrigued me for a long time is the question of standards.  Should standards precede the development of tools, or should a leading tool be the framework of a standard?  I believe that acceptance is not a question of the standard engineering excellence: it is a matter of marketing.  It is obvious from the article that people know, at least subconsciously, that an approach to improve efficiency in analog design following the model of digital design automation methods will not work.  But it seems apparent that we do not have consensus as to how to begin, let alone proceed with, the work of developing a solution.

The discussion leaves me with the impression that the economic urgency is not yet high enough to fuel the development of a new analog solution.  People complain, but do not revolt: in fact, they hardly even demonstrate.  Is this a non-problem?  Write to me (gmoretti@gabeoneda.com) and share your opinions.

ESL 2.0
Alan Naumann in his Viewpoint expresses the belief that we have turned the corner in ESL.  The tools, he states, are no longer just an interesting experiment; driven especially by the rapid emergence of heterogeneous multicore systems-on-chip, ESL is now part of accepted development methods.  I think that Alan is correct in identifying a new season in ESL, especially given the complexities of multicore systems design.  But the work is far from over: acceptance of the tools does not mean we have the best methods for system-level design yet.  Much more work is required, mostly in understanding what system-level design really means, and how we can integrate the design of the electronic portion into the total product design.  EDA is a wonderful industry: just when you think you have reached a plateau, you find new mountains loom on the horizon.  DAC, as usual, will provide many venues for designers to explore new heights.