Research on Performance Evaluation Method for Embedded Intelligent Computing Platform

With the rapid development of artificial intelligence technology, different intelligent tasks have put forward different requirements for the performance of embedded intelligent computing platforms. Therefore, rapid and effective evaluation of the performance of embedded intelligent computing platforms plays an important role. This paper studies the performance evaluation method for diversified embedded intelligent platforms. Firstly, a performance evaluation indicator system is constructed by combining the characteristics of general computing and embedded intelligent computing platforms; Secondly, different test cases are designed for different intelligent perception and decision tasks; Finally, the comprehensive performance evaluation system for intelligent computing platform is designed to manage the evaluation indicators and test cases. It plays an important guiding role in optimizing and improving the performance and architecture design of embedded intelligent computing platforms.