A Study on the Challenges and Prospects of PCM Based Main Memory Architectures

For several decades, computer’s memory has been volatile. High speed memories such as DRAM and SRAM have been used as main memory and cache memory respectively. Magnetic disks are used as the persistent secondary storage devices. The present DRAM based main memory has reached its energy limits. Due to the advent of nonvolatile memories like PCM, MRAM, RRAM, FeRAM and Flash, there is growing interest in incorporating them into the computers as the main memory alternates. Introducing the non-volatile storage devices in memory hierarchy is considered to reduce the energy consumption since nonvolatile memories have got no power leakage in memory cells. A major hurdle in this direction is the poor the endurance limit of most non-volatile memory technologies in comparison to the RAM technologies. Among the non-volatile memory technologies, PCM is the most attractive and the most investigated option due to it advantages over other memory types. But there are few issues related to PCM that need to be addressed in order to integrate it as a potential memory for computers. There have been intense researches being carried out to devise new memory architectures based on PCM. There are also methods identified to overcome the issues of PCM. This paper analyses the various challenges in integrating PCM into the system and the various PCM based main memory architectures that have been proposed so far.