Testing current high density memory chips using older algorithms is highly time consuming. March test of O(n) is the most widely used approach for its high fault coverage and systematic way of extending the test sequences. Size of the March test is guided by the number of fault models. Most of the March test generation algorithms reported so far, takes long time especially in case of number of operations exceeding seven. In this work, we propose a Particle Swarm Optimization (PSO) based March test generator which gives tests of high fault coverage, at a much lesser time compared to the existing approaches. Since the time to generate long March tests increases rapidly, in this work, we show that combining March tests of shorter lengths can replace the long March sequences without compromising fault coverage. Another bottleneck in memory testing is the power dissipation. The fitness function of the proposed PSO can be tuned to obtain March tests with high fault coverage, low power consumption, reduced peak power or a combination of any of these three. The experiments carried out confirm the effectiveness of our approach as it produces better results than those reported in the literature.
