Abstract. This paper presents the design of a high performance energy-aware whole-system emulator --SkyEye. Several optimization and novel energy estimation methods used in SkyEye are proposed. By using novel searching strategy for Translated Block (TB), SkyEye save the time to find proper translated block. SkyEye uses Basic Equal Length Unit (B-ELU) method, and dynamic binary translation to reduce the simulation and energy evaluation time. The performance model of B-ELU is built to get the best length of translated block. In addition, the simulator automatically detects the voltage/frequency variation, and adjusts the energy estimation model accordingly. Using these methods, SkyEye which simulates ARM CPU based hardware system achieves marvelous performance and energy-aware statistic capability in experiments.