A Study on Factors Influencing Power Consumption in Multithreaded and Multicore CPU’s

The ever-growing demand for computational power and high performance has led to a rapid growth in the semiconductor industry. This evolution has seen a continuous increase in CPU performance and the number of transistors on a chip has roughly doubled every two years – proving Moore’s law. An inevitable consequence when achieving this is that more functional units, deeper pipelining and larger cache sizes have had to be implemented on the CPU chip. The result is a significant increase in the power consumption. Achieving high performance with low power consumption has been the traditional goal in high-end processors. In order to accomplish high performance, multithreaded and multicore CPUs have become the recent trend in semi-conductor technology. The purpose of this paper is to statistically analyze the various factors that affect power, to study their relationship to quantify their influence on power consumption in multithreaded and multicore CPUs. This paper explores the on-chip power modeling simulation techniques with the existing processors and compares the performance and power trade-off between multicore and multithreaded CPUs. In this paper, we also present review/tutorial of recent advancements in power savings through the implementation of power-limiting micro-architectural features (e.g. out-of-order execution, branch prediction, caching and prefetching) in contemporary multi-core processors, such as Intel Nehalem and AMD’s Istanbul processors. The results show that the statistical findings on power consumption are encouraging and useful for low power application and power-aware processor designers.