HAIRDESC carries out its activities in line with the objectives of the Next-Generation HPC-AI Development Support Program promoted by MEXT, with a focus on the following initiatives:
- To support next generation HPC and AI research and development centered on GPUs, HAIRDESC provides multi-level education for a broad range of users. This includes training on GPU supercomputer hardware and software, GPU programming, performance tuning and understanding of advanced computational acceleration technologies.
- To promote GPU programming education, HAIRDESC centrally collects and organizes educational materials including slides, documents, and lecture or seminar videos, and delivers programming education at various levels, thereby fostering the next generation of researchers and engineers who will lead HPC and AI research and development.
- In collaboration with computational science application developers, HAIRDESC supports application development and human resource development not only for traditional simulation-based computational science that relies on supercomputer performance, but also for new approaches that fully leverage AI for Science methodologies.
- HAIRDESC supports the development of GPU-enabled applications for HPCI resources operated by national universities and national research institutes, including the next flagship supercomputer FugakuNEXT led by RIKEN and targeted for completion around 2030.
- HAIRDESC actively promotes exchanges with international HPC and AI research institutions and researchers, particularly in the EU and the USA. Through these efforts, the Center aims to enhance the global presence of GPU computing in Japan, facilitate the overseas use of GPU applications developed in Japan, and enable the smooth adoption of advanced GPU applications developed abroad on domestic GPU supercomputers, thereby contributing to next generation HPC and AI research with both international competitiveness and cooperation.
- Centered on the three core universities, HAIRDESC conducts advanced research on state-of-the-art GPU systems and programming methodologies. This includes promoting research on new programming frameworks beyond traditional coding approaches, the development of GPU computing libraries, and advanced user support methodologies.