Long Life Cycle Electronic Products
The Idea
1
Click on the image to view the keynote
The journey for this project started in the mid 1990s with the original PhD work at Harvard for reconfigurable computing. The focus of this work was to use delayed binding of function to do application specific optimization for performance or power. Over the years, this paradigm was useful for a number of small number of applications (cybersecurity, networking), however broad-based reconfigurable computing did not happen. Why? If someone wanted flexibility, they used a processor (software) and if they wanted performance/power, they built an ASIC. Rather, the work was actually more successfully used in configurable processors such as Tensilica or now RISC-V.
In 2011 with the role of Sr VP Strategy at Flextronics (EMS company), I was exposed to many customers who suffered from issues of semiconductor obsolescence or reliability. The reason? 90+ percent of semiconductor volume is focused on consumer marketplace. If you have a product which lasts more than five years (A&D, Industrial, Medical, Energy, etc), you are likely to face a lot of issues with the semiconductor supply chain. Is there a way to solve these problems through robust design? How about if the massive machinery of reconfigurable computing is used to build designs which are robust relative to supply chain churn and at the same time built in a way to handle future requirements volatility. In 2018, ACM group inducted the original work into the group Hall of Fame, and this concept was introduced to the group as a part of the HOF keynote address.
The Next Step
2
The next step was to define concrete flows within the context of existing design practices which used the power of reconfigurable computing and Artificial Intelligence to help Long Life Cycle customers. This resulted in three primary flows:
PCB Bottom-Up: In this flow, an existing PCB board design is in the field and has either obsolescence or reliability issues with semiconductor parts. The Anew design system takes the underlying PCB design and remaps it with a high degree of automation into an equivalent design with much better LLC properties.
System Top-Down: In this flow, a system design description can be optimized not only for power/performance, but also for robustness to supply chain issues (obsolete parts, reliability issues) and architectures for future design volatility.
Runtime Flow: In this flow, you have a piece of hardware which is hard/expensive/impossible to access. You can remotely change system function using the power of reconfiguration driven by the design system.
Launch
3
In 2020, a number of EDA veterans picked up the idea and are building a solution for LLC companies around this concept. As a part this work, EPSNews is publishing a series of articles which outline the space and solutions.