Recently, Professor Dai Hongjun, who holds dual appointments at the Institute of Intelligent Innovation and the School of Software, successfully led the basic software team in merging the first UEFI boot solution for RISC-V CPU servers into the mainline repository of the open-source Tianocore EDK2 community. They also completed the development of the first UEFI-compliant firmware for RISC-V servers. Following the X86 and ARM architectures, this achievement realizes that the RISC-V architecture can rapidly integrate into the server industrialization field with the support of the mature UEFI ecosystem. This marks that Shandong University has the international leading capability of core code research and development of basic software such as firmware and operating system kernel, and has become an important contributor to RISC-V key open source code.
With Professor Dai Hongjun as the chairman, the UEFI on RISC-V working group proposed a RISC-V firmware design approach that separates OpenSBI from UEFI, with each being maintained independently. This approach effectively enhances the source code of EDK2 and achieves a comprehensive boot solution "EDK2 – GRUB2 – Linux kernel" on the SG2042 RISC-V CPU. It has successfully run the domestically developed operating system distribution, OpenKylin. Through extensive collaboration and continuous improvements with engineers from Intel, Ventana, Qualcomm, RedHat, Sugon, and others, after multiple rounds of reviews, it has been successfully merged into the main EDK2 repository, making it available to RISC-V developers worldwide. Furthermore, several open-source code submissions for OpenSBI and the Linux kernel have also passed the review and are waiting to be merged.
Improved overall boot process for RISC-V servers
The team has also completed the "RISC-V CPU+TPU" intelligent computing fusion solution, achieving the world's first enterprise-level TPU BM1684X driver and optimization on SG2042 CPU, and successfully running AI models like Stable Diffusion for image generation and large-scale inference models inferllm. This opens up a new path for deploying high-performance AI applications on RISC-V server platforms and promotes the development of high computing power by combining RISC-V and AI technologies.
During the research and development process, Professor Dai Hongjun actively practiced the concept of organized scientific research and integrated development. He led a joint research team composed of faculty and students from the School of Software, the School of Integrated Circuits, and the Institute of Intelligent Innovation. They consistently adhered to "integration into the ecosystem" and "open source development." They joined organizations such as the RISC-V Foundation, UEFI community, OpenKylin community, OpenAtom Open Source Foundation, and established the China UEFI on RISC-V Working Group and the Shandong University RISC-V Open Source Club, making significant contributions to advancing key breakthroughs in RISC-V technology.
edk2 repository URL: