Advancing the Ecosystem of Domestic Processors: Development and Optimization of Function Libraries for Improved Performance

Abstract

The rapid advancement of the computer industry and the increasing need for information security have underscored the strategic importance of China's domestic processor ecosystem, especially in light of international trade restrictions. This paper explores the challenges faced by China's processor industry, including limited software ecosystem support, inadequate optimization for critical applications, and dependence on foreign technologies. By focusing on Kunpeng, Feiteng, and Loongson processors, this study highlights their potential and the urgent need for ecological development to enhance their competitiveness. Open-source software applications, such as CFL3D for fluid mechanics and GIMP for image processing, are analyzed for their role in advancing domestic processor ecosystems. Using tools like Turbo C and Authorware, a high-performance library of basic functions was developed, with implementations in C and assembly languages for DSP chips such as the FT-M7002. The research also demonstrates the successful deployment of optimized image and signal processing libraries on Huawei Kunpeng servers. These efforts mark a significant step towards addressing the "core shortage" problem and promoting the localization of China's processor industry.