The case for a configure-and-execute paradigm

Abstract

Tomorrow's silicon chips will hold more transistors than most embedded system designers could possibly use under the prevalent "describe-and-synthesize" design paradigm. Many have thus re-proposed the once popular "capture-and-simulate" paradigm, wherein pre-designed Intellectual Property software and hardware components are connected and co-simulated, to reduce this gap. However, major hurdles limit this paradigm to only very high-cost embedded systems. In this paper, we describe those hurdles and present a case for a new "configure-and-execute" paradigm for mainstream embedded systems, based on the idea of deconstructing rather than constructing systems, which takes advantage of the surplus transistors in a way that can overcome the hurdles and significantly reduce time-to-market.