Nanocomputing Architectures with Bloom Filters as Defect Maps

Significant progress has been made in the development of the emerging nanoscale computing devices. While the ultimate manufacturing process for such devices is uncertain, it is abundantly clear that future technology nodes will see an increase in defect rates. Therefore, it is of paramount importance to construct new architectures and design methodologies that can tolerate large numbers of defects. Defect maps are a necessity in the future design flows, and research on their practical construction is essential. In this work, we study the use of Bloom filters as a data structure for defect maps. We show that Bloom filters provide the right tradeoff between accuracy and space-efficiency. In particular, they can help simplify the nanosystem design flow by embedding defect information within the nanosystem delivered by the manufacturers. Additionally, our method does not have falsepositive in defect identification, i.e. it will not report a faulty device as functional. We develop a novel nanoscale memory design that uses this concept. It does not rely on a voting strategy, and utilizes the device redundancy more effectively than existing approaches.