The ability to simulate large-scale quantum circuits is important to developing quantum computing systems. We developed software and related performance optimization techniques to simulate large-scale quantum computing systems with non-volatile memory arrays. Compared with the practice of using solid-state disks as virtual memory to increase the simulation scale, we greatly improve the simulation speed by optimizing parallel computing scheduling, data access patterns, and re-grouping quantum circuits. The resulted speed is close to memory-based simulation with QuEST, but is capable of simulating much larger quantum computing systems, which is the most scalable and cost-effective quantum computing system simulation software available today.