Abstract: The computing product performance benchmark tool is an important guarantee to support the iterative optimization of computing product performance and improve the capability of the computing industry. In order to fill the gap in the current domestic CPU performance evaluation benchmark tool, CPUBench is developed, which is positioned to evaluate the comprehensive computing capacity of the CPU, memory subsystem and the compiler under the general computing scenario. The framework of CPUBench is developed by Python, modular design, and compatible with x86_64, aarch64, ppc64le, sw_64, loongarch64 and other pocessor architectures. The workloads comes from the computing intensive applications of typical business scenarios in various fields, and has a good representation of the actual business. From the perspective of architecture related or irrelevant characteristics such as instruction proportion and top-down, the characteristics of the workloads are obviously different. The characteristics of the whole test suite cover a wide range, and can fully represent the actual business characteristics under the current general computing scenario. At the same time, the microarchitecture characteristics of CPUBench and SPEC CPU2017 on the same test environment are compared by PCA analysis method.CPUBench basically covers the characteristic classes of SPEC CPU2017, and adds the characteristics of new applications such as big data and database.In addition, CPUBench and SPEC CPU2017 were tested and compared on 14 different tested platforms, and the score curve showed good trend consistency, indirectly proving the availability and rationality of CPUBench as a general CPU evaluation tool.In general, CPUBench has certain advantages in business representativeness, ease of use and maintainability, which can be used to guide the design optimization, specification selection and market procurement of general computing CPU, server and other computing product, and has important guiding significance for the development of the computing industry.