YatCC-AI: Supercomputing + DeepSeek, Sun Yat-sen University starts practical teaching of "intelligent compilation"
Recently, DeepSeek has set off a wave in the global technology field, promoted the vigorous development of the domestic computing power industry, and rapidly triggered profound changes in various fields. How to effectively integrate these cutting-edge technologies into curriculum teaching and cultivate professional talents for the future intelligent era has become an important topic in the field of education. In response to this, Associate Professor Zhang Xianwei from the School of Computer Science and the Supercomputing Center team took the lead in coming up with ideas and innovative solutions, and successfully developed and launched the YatCC-AI intelligent compilation teaching practice platform (https://yatcc-ai.com) based on domestic supercomputers and DeepSeek . The platform deeply integrates high-performance computing (HPC) and AI technologies into computer system courses, providing a new intelligent and modern learning experience for the core courses "Compiler Principles" and "Compiler Construction Experiment" in the 2025 academic year.
YatCC-AI "intelligent compilation" teaching practice platform comprehensively optimizes the practical process and provides a one-stop, integrated and convenient experience. Users do not need to make complicated configurations, they can just open the cloud compiler development environment with one click through the web page. Whether in the classroom, dormitory or laboratory, they can seamlessly connect and start intelligent compilation experiments anytime and anywhere. The platform is equipped with a full-process AI empowerment system, supporting 7x24 hours of compilation knowledge Q&A to solve technical problems in seconds; providing AI-assisted code development and generating development suggestions in real time; enabling the exploration of cutting-edge compilation optimization solutions and helping incubate scientific research innovation. Based on the powerful empowerment of DeepSeek, students can quickly grow from novices to experts in the field, master the comprehensive capabilities of "using intelligence, building intelligence, and improving intelligence" in the AI era, and lay a solid foundation for future technological development.
Technology integration: supercomputing and AI empower teaching
The YatCC-AI platform deeply integrates supercomputing system technology and AI large model technology. Relying on the "Tianhe-Xingyi" advanced computing power of the National Supercomputing Guangzhou Center, the "Xingguang" supercomputing application platform and DeepSeek intelligent services, it integrates cutting-edge tools and technologies such as VSCode code space, high-performance computing containers and parallel file systems.
Through a one-stop online integrated development environment and supercomputer native DeepSeek-R1 access, the YatCC-AI platform has developed and built a one-stop multi-level intelligent practice environment in terms of API interface, LLVM knowledge enhancement, C++ built-in LLM calls, etc., providing full-process intelligent teaching practice support such as professional knowledge question and answer reasoning (LLM4Chatting), LLVM code development and tuning (LLM4Coding), and exploration of cutting-edge compilation solutions (LLM4Compiling).
Practice-oriented: seamless connection from classroom to production and research
Based on the previous version of YatCC, YatCC-AI continues the core concept of "practice-oriented, quick to get started, and exploring the frontier", and is committed to cultivating students' compilation practice ability and global vision of complex systems. The experiment is based on the Clang/LLVM compilation facility widely used in industry, emphasizing the engineering practice characteristics of "based on LLVM, learning LLVM, and comparing LLVM", guiding students to build a complete and practical compiler.
Through modular and decoupled design, YatCC-AI achieves good interaction and close collaboration with the LLVM framework. Students can refer to and compare the standard implementation of LLVM at any time during the development process to deepen their understanding and optimize the design. In addition, the platform introduces an automated real-time feedback evaluation mechanism and graded test cases to provide timely and fine-grained improvement suggestions to help students gradually complete the implementation of each stage of the compiler.
Through YatCC-AI's course experiments, students can not only master the basic skills of developing practical compilers, but also have in-depth contact with and get started with the LLVM open source community framework, have the ability to participate in cutting-edge engineering practice and research exploration, achieve seamless connection from classroom knowledge to industrial applications, and form an innovative ecosystem that integrates industry, academia, and research.
YatCC-AI was independently developed by a student team from the School of Computer Science. The team members include undergraduate, master and doctoral students, including graduate students such as Gu Yuhao and Zheng Tengyang, as well as junior undergraduates such as Chen Haoquan and Ouyang Yipeng, forming a cross-grade, multi-level collaboration model. The YatCC team not only focuses on basic cutting-edge research in high-performance computing and intelligent computing, but also actively engages in teaching practice, promoting the in-depth integration of scientific research and teaching, and achieving a two-way breakthrough in the publication of cutting-edge academic papers and the implementation of technological applications.
With the strong support of the School of Computer Science and the Supercomputing Center and the guidance of Professor Zhang Xianwei, the YatCC team adheres to the concept of "learning ability, thinking ability, and action ability" and carries out efficient division of labor and collaboration around architecture design, code development, technical testing, document writing, video production and other aspects to ensure that the practice platform has leading functions, stable operation and easy use.
In recent years, the YatCC team has made steady progress in the teaching reform of compilation systems. The results have been highly recognized both inside and outside the school. It has won multiple awards and honors in teaching and competitions, and won wide acclaim in the industry.
Awards and Honors
Conclusion
The launch of the YatCC-AI platform demonstrates the prospect of integrating supercomputing and AI technology in the teaching of basic computer courses, and provides reference and inspiration for cultivating future professionals in the field of computer science. Through this platform, you can not only easily master the core knowledge of compilation technology, but also unlock deep innovation and creativity based on large models in the AI era. A new semester, a new starting point. Come join YatCC-AI and start your intelligent compilation journey!
Visit YatCC-AI platform: https://yatcc-ai.comRegister
and contact email: yatcc-team@outlook.com
Source: Chen Haoquan, Ouyang Yipeng, Zhang Xianwei
Editor: Ma Ye , Xu Ying
First review: Wang Dongmei
Review: Yan Xiaohui
Review and release: Ma Xiao
