Providing Compilers and Application Program Support for Reconfigurable SoCs: Radical but Overdue

After nearly half a century, the reconfigurable chip concept has transcended the time when it was first conceived and still remains very attractive today. However, for the most part of this history, the transition from a promising research idea to commercial success was a frozen evolution. Technology roadmaps of early-generation reconfigurable SoCs ended up in failures and extinctions. In our view, the lack of compilers and support at the applications programming level was responsible for this situation. This paper is devoted to that perspective. We reveal how a lack of compiler support consigned the first generation of reconfigurable SoCs to failure. We also present current progress and limitation. We conclude this paper with an outlook on the technology roadmap for reconfigurable systems-on-chip and an overview of the authors own research efforts in this area.