In the mid-1980s, in response to requirements for higher performance, lower cost, shorter product life time, and faster time-to-market, a market developed for application specific ICs (ASICs). ASICs are ICs taylored to meet the critical system requirements in an otherwise traditional system composed of ASICs and standard components. Typically, though not always, the volume is small. In 1988, in Denmark, an estimated number of 70-80 ASICs were designed.
In this model, the semiconductor company acts as a process provider, while the system designer takes on the design of the device. Such a division of work may not result in reaching the limit of technology.
According to the predictions of Moore's law2), an increasing number of transistors will be available in IC's in the forthcoming years. Thus, even more complex microprocessor devices can easily be foreseen. In the ASICs case, entire systems may be realized in a single chip. The obvious question is, how we may organize design to benefit from this potential complexity.
The answer to this question is abstraction. In the near future, the system designer may need the ability to select layout entities, which may be provided by providers independent from the process providers, in order to manage system complexity. This requires the emergence of entity providers, who hold the intellectual rights to specific layout entities but are independent of the semiconductor process houses. System design will consist of selecting layout entities from one or more entity providers.
Author:
Flemming Stassen
(http://www.it.dtu.dk/~stassen/Edu/49260/Historie)