Study into the architectures of next-generation computer memories by making use of emerging non-volatile memories and modelling techniques

Abstract

A contemporary approach to the design of computer systems puts a priority on energy economy. It is a widely held belief that leakage increases exponentially with smaller CMOS technology. This is due to the fact that standard CMOS scaling theory dictates that threshold and supply voltages are lowered in proportion to device sizes. Because of this, contemporary methods consider leaky power to be a rival to dynamic power. Without a wave of revolutionary technology that can completely change the game, the issue of power budget leakage never be able to be overcome. Within the realm of non-volatile memory technology, there have been a number of significant new advancements that have taken place. "ReRAM," "PCRAM," and "Spin-Torque-Transfer Random Access Memory" (MRAM, STTRAM) are some examples of popular examples of modern non-volatile memories that have desired properties such as low access energy, high cell compactness, and great access performance. These memories consist of a combination of these qualities. Therefore, it is excellent that these new technologies for non-volatile memory are used in the construction of future computers that are not only powerful but also efficient in terms of energy consumption. Due to the fact that these new non-volatile memory technologies are still in the research and development phase, further academic research is required in order to demonstrate their utility. Due to this, this study investigates three different approaches that may be used to facilitate the development of these new types of non-volatile memory. The first step is to create models of a few different types of nonvolatile memory, which include their space requirements, power consumption, and performance at the circuit level.