Robust Rapid & Wireless Chip Design

Two decades of relentless improvement in semiconductors, circuits, and software tools has created a set of dominant design styles for today's integrated circuits and systems. These circuit, system, and software styles comprise the design infrastructure for the discipline of microelectronics - the techniques we rely on to convert transistors into performance. This design infrastructure is now at risk. The radical and uncertain semiconductor technologies of tomorrow threaten to make obsolete many of today's most basic circuit design assumptions. The extreme-performance, ultracomplex systems of tomorrow threaten to overwhelm today's fragile, inefficient, often nonconvergent circuit design flows. The highly integrated systems of tomorrow threaten today's ad hoc and incomplete strategies for analog and other nondigital interface circuits.
Tomorrow's circuits must routinely move billions of bits per second through the air; perform billions of operations per milliwatt; access billions of bits of on-chip storage; interact with a rich environment of communicating electrical, mechanical, optical and biological systems; and offer a spectrum of soft-to-hard reconfiguration options. To convert tomorrow's transistors into this range of required performance requires a radical rethinking of today's design strategies. Today's design styles are either vastly wasteful of performance (e.g., ASIC, FPGA), enormously expensive in time and effort (e.g., full custom), or so entirely ad hoc (e.g., analog, sensors) that they maximize both design time and design risk. This bodes ill for the increasingly large, heterogeneous, time-constrained electronic products essential to the nation's future economic vitality.
We propose to reinvent today's at-risk circuits infrastructure with new circuit-, system- and software design strategies aimed at robust, rapid design with tomorrow's radically new semiconductor devices. We will demonstrate a set of coherent design methods applicable across a complete range of custom digital, analog, and interface circuits, and extremely-integrated heterogeneous systems. We will attack these problems in a distributed, multi-university Center for Circuits, Systems & Software (CCSS). CCSS will comprise a consortium of leading universities with demonstrated records of success in delivering real circuits, real systems, and real software.