In this paper, we have proposed a novel architectural level simulation platform set-up to study the temperature characteristics of a processor system, when applying different speed scheduling schemes to execute the desired workload. Our simulation platform set-up is very practical as it combines the most practical simulation models, and uses the industry-standardized benchmark SPEC CPU2000 as the input to the platform. On this novel platform set-up we have compared two speed scheduling techniques, namely m-oscillation and a traditional optimal dynamic energy reducing speed scheduling technique, and have shown that m-oscillation can be as effective as the traditional approach in reducing the peak temperature of the system. © 2011 IEEE.