크로스바 어레이 구조를 갖는 저항변화 메모리에서 최악의 구동 시나리오를 찾기 위한 검색 알고리즘

Abstract

This thesis presents a search algorithm to find the worst operation scenario of an ReRAM crossbar array. According to the analysis, the previously believed worst operation scenario does not always result in the worst operating signal and thus the previous margin evaluation can have serious errors. Thus, to solve this problem, a search algorithm is proposed to systematically find the worst operation scenario by wisely sampling scenarios and adaptively searching for the worst one. In experiments, the proposed algorithm successfully found the worst operation scenario for various designs. Results show that the correct worst read scenario found by the proposed algorithm can be different from the previously believed one, and the correct sensing window can be as small as 1/14 of the previously evaluated one. This result clearly demonstrates the need of the proposed search algorithm for the worst operation scenario in ReRAM crossbar design.