View article

Today's memory technologies, such as DRAM, SRAM, and NAND Flash, are facing major challenges with regard to their continued scaling. For instance, ITRS projects that DRAM cannot scale easily below 40nm as the cost and energy/power are hard -if not impossible- to scale. Fortunately, the international memory technology community has been researching other alternative for more than fifteen years. Apparently, non-volatile resistive memories are promising to replace the today's memories for many reasons such as better scalability, low cost, higher capacity, lower energy, CMOS compatibility, better configurability, etc. This paper discusses and highlights three major aspects of resistive memories, especially memristor based memories: (a) technology and design constraints, (b) architectures, and (c) testing and design-for-test. It shows the opportunities and the challenges.