2025 Accelerator Programming Summer School
Accelerators such as GPUs and FPGAs, when combined with general-purpose multicore CPUs, create heterogeneous computing environments. These systems are gaining attention in fields that require high-speed computation, such as deep learning, big data processing, and scientific simulations. However, there is a steep learning curve to effectively utilize these accelerators. This summer school will focus on programming GPUs using the CUDA language. We hope this program will raise awareness of the importance of heterogeneous and accelerated computing in Korea and contribute to its broader adoption.
— Professor Jaejin Lee, Director of the Summer School (Seoul National University)
-Date: Monday, August 25, 12:00 PM – Friday, August 29, 1:00 PM
-Venue: Seoul National University Siheung Campus Hotel & Convention Center (173 Seoul National University-ro, Siheung-si, Gyeonggi-do, Korea (2554 Jeongwang-dong)) [서울대학교 시흥캠퍼스 연수동&컨벤션센터]
Shared accommodation (2 people per room) will be provided during the program.
-Hosted by: Next-Generation Supercomputing Center, Institute for Future Innovation, Seoul National University
-Organized by: Thunder Research Group, Seoul National University [THUNDER Research Group]|
ManycoreSoft Inc. [Manycoresof]
-Sponsored by: Moreh [Moreh] |
CHAMP (Center for Optimizing Hyperscale AI Models and Platforms) [초거대 AI 모델 및 플랫폼 최적화 센터]
-Program: The 5-day program includes lectures, hands-on sessions, guest talks, and team projects. Participants will gain a solid foundation in accelerator computing, covering topics such as accelerator architecture, CUDA programming, and various optimization techniques. Prerequisite: Proficiency in C programming.
-Hands-on Environment: The lab environment consists of 8 nodes, each equipped with 4 NVIDIA V100 GPUs. Nodes are interconnected via Infiniband for high-speed communication.
-Registration Fee General: KRW 1,500,000, Students: KRW 1,000,000 (Includes accommodation and meals during the program)
Visit our website for more details and to register!