Analysis of Power, Temperature, and Performance on Mobile Application Processor

Recent mobile devices have multiple cores and high operating frequencies. As a result, their performance has increased, along with their power consumption and temperature, which have become problems to solve. To understand how to operate a CPU efficiently while solving these problems, we study the relationship between power consumption, temperature, and performance as functions of the number of operating cores and operating frequency. We use CPU power consumption that we measured, use steady-state temperature of the CPU calculated by Therminator and use DMIPS that is an index of CPU performance measured by Dhrystone. The experimental results show that the performance increases and temperature decreases as the number of operating cores increases for the same power consumption. In addition, for the same performance, power consumption and temperature decrease as the number of operating cores increases. Consequentially, the quad-core shows a 75.54% performance improvement and 36.04% reductions in temperature compared to the single-core for the same power consumption. In addition, when at the same performance, the quad-core has decreased power consumption and temperature compared to the single-core, 49.42% and 53.94%, respectively. Therefore, operating by increasing the number of cores in the multi-core application processor will effectively increase the performance and lower the power consumption and temperature.