GIST Launches Construction of Korea's Largest HPC-AI Shared Infrastructure

Aerial view of the G-I-S-T Data Center. Photo by GIST.

[Asia Economy Honam Reporting Headquarters Reporter Cho Hyung-joo] GIST (Gwangju Institute of Science and Technology) announced on the 7th that it has been selected as the lead institution for the HPC-AI (High-Performance Computing-based AI) shared infrastructure construction project, organized by the Ministry of Science and ICT, the National IT Industry Promotion Agency, and the AI Industry Convergence Project Group.

Last August, the AI Industry Convergence Project Group (Director Lim Cha-sik) announced the largest HPC-AI shared infrastructure construction project in Korea as part of the AI-centered industrial convergence complex development project, and selected the operating institution through proposal and review processes.

The ultimate goal of this project is to contribute to the activation of the AI ecosystem in industry-academia-research-government collaboration through the establishment of HPC-AI based shared infrastructure.

It is expected to serve as a catalyst for domestic and international industry-academia-government-research collaborations and international joint research, showing interest in the AI complex infrastructure located in the Advanced 3 District of Gwangju Metropolitan City.

To successfully carry out this project, GIST plans to ▲ build HPC-AI shared infrastructure facilities that can be jointly used for AI convergence research, education, and corporate support ▲ realize operational independence through sustainable shared infrastructure service operation and maximization of utilization ▲ provide collaboration opportunities among large, medium, and small enterprises and various business opportunities by establishing a data and AI leading model production and distribution base.

GIST is currently actively working on the practical setup of the G-I-S-T data center facility with 100 racks in the AI Graduate School space on campus, aiming for opening early next year.

Additionally, it plans to complete the phased construction and operation of HPC-AI shared infrastructure equipment by 2023, supporting GPU nodes with a computing capacity of 6 PFlops tightly coupled and interconnected via a 200 Gbps ultra-high-speed fabric, storage nodes with a total capacity of 10 PBytes, and targeting 99.9% uptime based on automated integrated management.

The established HPC-AI shared infrastructure will be utilized to support AI computing for research and education through cooperation with the AI Convergence Graduate School and to provide customized HPC-AI computing services for domestic industry-academia-research and global partners requiring large-scale training.

Furthermore, to achieve operational independence through revenue model creation, by 2027, GIST plans to expand HPC-AI infrastructure capacity by attracting a national high-performance computing regional hub specialized center, advance demand-tailored services, and sustainably attract global partners.

Kim Jong-won, Dean of the AI Graduate School and Director of the Supercomputing Center at GIST, stated, “We will do our best to contribute to the emergence of a virtuous cycle AI ecosystem by creating customized services for domestic industry-academia-research-government companies,” and added, “We expect that the GIST Supercomputing Center will support basic research and industrial technology development for the growth of the national science network.”

Following the AI Convergence University project announced by the AI Industry Convergence Project Group this year, GIST’s selection for the shared infrastructure construction project practically contributes to the AI-centered industrial convergence complex development project, striving for regional innovation, job creation, and fostering high-level AI talent through artificial intelligence.

This content was produced with the assistance of AI translation services.