Although vacations are supposed to be an opportunity to unplug from work-related activities, they also give me a way to re-organize my time. And so, I get the chance to do things I did not have time to do while immersed in the daily work routine. One of the things I got to do last week was read a book by Brian Bailey, Grant Martin, and Andrew Piziali: ESL Design and Verification published by Morgan Kaufman. If you have to learn about ESL, you cannot ask for better instructors. Individually each of the three authors can provide you with a robust understanding of ESL. Together they provide a formidable look at this quite misunderstood segment of EDA.
The book aims to be a prescription for ESL methodology, and talks mostly to designers. I am not sure that this segment of our industry is the one that needs to be convinced, although the book provides them with a wealth of background information and suggestions about methods that can lead to successful ESL design.
What I like best about the book is the organization of each chapter. After discussing the subject, the authors always add a section entitles: The prescription. There they share their editorial opinions about the material covered. This is a good attempts to make the reader think about the consequences of what he or she has just read and could well be the way to start a healthful discourse within the industry.
If there is a fault in the book it is the superficiality, and at times, absence, of economic considerations in the creation, growth, and future prospects for the ESL segment. Chapter four, for example, deals with ESL enablers and you cannot find any discussion in the entire chapter about the economic forces that would justify the growth of the sector. The reader is told about technology in great details: after all this is the aim of the book, but I think that designers, waiting for tools from EDA vendors that would help them implement the methodology suggested in the book, may like to know the reasons for the slow pace of tools development and introduction from the leading vendors, especially when startups seem to continue to miss the crucial point of the new methodology. The key work in Electronic System Level is not Electronic, it is System. The vendors that have shown success in this space, like The Mathworks and National Instruments have addressed the problem of system level design that encompasses electronic modules. The EDA industry has so far failed to realize the promised growth in this area specifically because they have focused on the electronic part, forgetting about the problem of system design.
System designers are of course concerned about implementing the electronic part of the system, but they are even more concerned with implementing the entire system, a notion that has escaped the provincial thinking of EDA companies. Unfortunately Brian, Grant, and Andrew are technologists, not businessmen. Designers getting to the end of the book may well think that EDA vendors "just do not get it", but cannot derive any support for their opinions from the authors. The last chapter: "Research, Emerging, and Future Prospects" clearly shows the background of the authors. Their prescriptions to the research community and designers are clear and helpful, but their advice to managers is general and weak. This is too bad because this is what the industry really needs if we can finally be persuaded to abandon the assumptions of a few years ago and finally address the real problem: the complexity of heterogeneous systems.
Non-the-less this is a book designers should read. It will guide them through the jungle of tools and methods that obfuscate the world of ESL at present.