Abstract:With the wide application of intelligent devices in all the fields, their application system has the increasingly prominent issues in the aspects such as power consumption, efficiency and instantaneity, and such issues have become the main obstacle for the continuous development of the intelligent devices. How to improve the intelligent device application system performance has become an important topic in this field. In this paper, the technology is able to effectively improve the intelligent device application system performance in the way of core algorithm and design framework of the application software. First, it conducts the cross-platform transplant on the core algorithm and framework of the application software based on the technologies such as JNI and NDK; then, it improves the original algorithm with the native code (C or C++ language); finally, it utilizes the multi-core calculation and processing ability of the intelligent devices to appropriately increase the task quantity and improve the load balancing. The experimental result indicates that: if such technologies are appropriately integrated into the original application system, the overall performance of the intelligent devices can be improved substantially.