Slides
Various new memory technologies, such as Phase Change Memory and Memristors, are being explored as alternatives to the current DRAM technology, whose scalability has stagnated in recent years. In the next generation of computers, these new memory technologies are expected to be available to the user-level programs as a large, byte-addressable, non-volatile random-access memory (NVRAM). For the foreseeable future, current programming languages such as C/C++ are expected to continue being used to program these future systems with NVRAM. In this talk, I will discuss numerous challenges related to programming these future systems with the current programming languages, along with the approaches to address these challenges.