GUEST EDITORIAL

For Designers: Opportunity

 

Michael J Flynn

Emeritus Professor, IEEE Fellow, ACM Fellow

Department of Electrical Engineering, Stanford University

 

Much has been made of the design productivity gap. Basically, this represents the difference between the rate of growth of design complexity (transistor density) and the rate of growth of designer productivity (measured in transistors designed per month). In a widely cited study by S. Melik of Sematech computes this gap as increasing at the rate of 37% / year. He notes that design complexity has and increased by 58% /year; while productivity has increased by 21% / year.  The productivity implication is that CAD tools and related technology enable designers to manage the complexity. But there are several dimensions to productivity: tools, organizational discipline and the designers themselves.

The gap

But first, is the gap real or imagined? The increase in transistor density is real enough, but does that in itself translate into complexity? Sometimes increasing density leads to design simplicity. This occurs when a significant piece of a design or design element can be captured and reduced to a single specific element. For capture to work the resultant element must be well understood, with input and output properties easily specified and well determined behavior.   Usually when capture occurs the number of input and outputs as a fraction of the total size (transistors, etc) of the design decreases noticeable.  Using capture effectively requires the designer to deal with larger sized design components (or design grains).  All this has its own specification, documentation and support overhead.

Still the complexity enables the designer to deal with larger units as components. So increased ¡°complexity¡± ¨C in terms of transistor count ¨C may actually result in simplicity. But when it does not, we have to count on the increased capability of our tool set.

The tools

There¡¯s no denying that design tools have significantly improved in both variety and robustness. Moreover our notion of a CAD tool has broadened. Ideas such as FPGAs and other types of programmable logic have developed to a stage where they become a real design asset either as a prototyping tool or as a per se design instrument. This movement towards softer designs is broader still and includes the addition of reconfiguration features and reuse considerations.

The notion of soft design comes to a logical conclusion in companies whose sole (or principle) product is design. Whether the design is a microprocessor (ARM microprocessors) or memory bus technology (RAMBUS) or DSPs (many specialized offerings), designs are now available. Some are better supported than others, but the variety is impressive.

The rise of the soft design companies doing proprietary design of ¡°captured¡± recognizable design elements is quite different from older design companies doing solely commissioned design.  Soft design companies start with a product idea, then implement, complete and test the design. A traditional hardware company would then manufacture and (in one way or another) market it. All this requires significant capital which the soft design company avoids. The skill in soft design is in design for reconfiguration and reuse. A quality soft design has a full repertoire of support tools (compilers, libraries, etc.) as well as well thought out plans for manufacturing support aids and reliability. What enables soft designs? The very complexity (transistor density) afforded by the technology and the need for designers to remain increasingly productive. The large die size enables targeted system applications (SOC) while soft designs enable the system designer to control design costs.        

The great entrepreneurial opportunity for the best designer is no longer product development and manufacturing, but simply product design and design support. Such enterprises focus attention on design, something the designer does best and avoid myriad other product concerns.

With a productivity gap, whatever the size, the simple law of supply and demand puts demand on the side of the designer. So how does the individual designer make the most of it? It seems to me that several strategies to maximize opportunity:

1) Continuing education¡­. It always pays to be prepared with the best knowledge of the field. This means advanced degrees where possible, attention to relevant technical literature and conferences and familiarity with the best CAD tools. Even as qualified designers are increasingly in demand the very best designers will command an even larger premium for their talent.

2) Global engineering¡­. The field of computer engineering is increasingly a multinational enterprise. So it pays to be fluent in language, custom and attitude with your major international partners.

3) Entrepreneurial and management awareness¡­. Opportunities will arise for skilled designers, but how to know good opportunities from bad? There¡¯s a basic set of business skills that an accomplished designer should have at their command. And since the designer deals primarily with ideas, the protection and valuing of intellectual property is required knowledge. Even more important is the ability to judge people, their talent, character and capability.  This is what we usually call management skill.

Putting it together equals opportunity

So the skilled and knowledgeable designer is faced with increasing demand for his talent. Opportunity abounds with soft design oriented business models, improving design tools and an advancing technology.  But the truly bright spot in the picture is that there¡¯s no saturation point in the marketplace, or at least there¡¯s no saturation point in the value of amplifying human intelligence which is after all what computer engineering is all about.