This paper analyzes various multi-core architectures that have been proposed for decades to increase the performance or power efficiency of processors. First, we investigate representatives of each type of multi-core architecture in detail. Then, we estimate the performance and power efficiency by modeling them into a cycle-accurate simulator. As a result, we conclude that each type of architecture shows different characteristics on the perspective of optimization point. Therefore, architects need to consider their purpose of designing processors, such as target applications, high-performance computing, and low-power computing.