The “S” in ASIC

What does it take to build an ASIC? What role does ESL play in this process? And what will things look like 20 years from now? In a recent post, Synopsys’ Frank Schirrmeister predicts that IP and re-use will play a “crucial role” and suggests that ASIC design may very well be simplified to a mere assembly of standardized components. In an analogy with urban planning, the Oakland harbor is compared to a USB interface. Why redesign standard and readily available features?

While there is no doubt that IP and re-use have become very important in modern design, I believe it is still a stretch before ASICs become “sea-of-IPs”. Elaborating on the harbor analogy, can we really build a “one-harbor-fits-all” type of construction? Are the same infrastructures needed for grain carriers and oil tankers? Can the same construction fit sea and fluvial requirements? What about sizing the harbor to traffic expectations? And would a blueprint designed a few years back pass new regulations?

Preparing Recommendations

IP is a necessity to manage complexity, but it isn’t just yet the silver bullet that will solve all design problems. Today, with the on-going convergence, it is not uncommon to find the same “feature” in a very wide and diverse range of devices such as laptops, set-top-boxes or smart phones. Yet, if the function is identical, the underlying implementation will be drastically different. All of the above mentioned devices can for instance play video - but the image size and quality requirements will vary, the power consumption constraints will be different, the available processing resources will differ, the available memory interfaces will change…

The ASIC design economy is about finding the right balance between cost and features. IP and re-use strategies obviously help reducing the engineering costs. But these approaches may have hidden manufacturing costs: generic re-usable components are often over-designed to accommodate more designs. And by relying too much on off-the-shelf IP, valuable product differentiation is lost.

The “S” in ASIC stands for Specific. This specificity is the purpose of ASIC design, and with it comes vital innovation and differentiation. IP has an important role to play, but not an exclusive one. Custom hardware design is not going away. What do you think?

Preparing Recommendations