Behind the Innovation of 1990s Korean Cars Was ○○○ [Reading Science]

The 4th Industrial Revolution Era: Ultra-High-Performance Supercomputers as Core Infrastructure
Increasing Utilization in Industry, Medicine, Science, Space, Military, and Bio Fields
Surging Demand for Supercomputer Resources in the AI Era
Competition Among the US, China, and Japan to Introduce 1-Exascale Supercomputers
South Korea Holds 6 Units in the Top 500 Group Combining Public and Private Sectors
The First Unit Introduced in 1988 Played a Key Role in Domestic Car Performance and Design Innovation
Units 2 to 5 Also Contributed Significantly to Basic Science Research
Soaring Demand and Performance Rankings from 5th to 10th Place

[Asia Economy Reporter Kim Bong-su] If highways and ports were the core infrastructure for national economic development during the industrialization era, in the 4th Industrial Revolution era, ultra-high-performance computers, i.e., supercomputers, take on that role in scientific research and industrial fields. Leading countries such as the United States have developed ultra-high-performance supercomputers with exascale capabilities?performing over 10^18 calculations per second?and utilize them extensively not only in military fields like cryptography but also in product design and development, vaccine research, climate change prediction, space simulation, and artificial intelligence (AI) development. South Korea has recently decided to introduce its 6th national supercomputer at the pre-exascale level to keep pace with this trend.

◇ What is a Supercomputer?

An ultra-high-performance computer is a large computer system that performs massive calculations and is increasingly essential as core infrastructure not only for computational science but also for AI technology utilization in the 4th Industrial Revolution era. Supercomputers are generally classified into three groups: the 'Top500,' which lists the 500 most powerful supercomputers internationally; ultra-high-performance computers costing over $1.5 million; and server-grade computers (HPC) that outperform general computers and are used in science, engineering, and data analysis. Their applications include large-scale computational fields, various scientific and technological areas such as pharmaceuticals, convergent and complex applied research through AI integration, AI models that autonomously think, learn, judge, and act by learning from large-scale data, and public life and crisis management sectors that predict and prepare for social issues using AI and big data.

The computational power of supercomputers has recently reached the exascale level, capable of performing over 10^18 calculations per second. Since the introduction of terascale computers in 1997, which performed one trillion calculations per second, performance has improved more than one million times in 25 years. In May, the U.S. Department of Energy's Oak Ridge National Laboratory's supercomputer 'Frontier' demonstrated a computing power of 1.102 exaflops, securing its position as the world's top computer. It can perform 110.2 quintillion calculations per second.

Until the 1990s, the U.S. dominated the supercomputer field, but competition has intensified with challenges from China and Japan. China first ranked number one globally in 2010 with Tianhe-1A, and its Sunway supercomputer was recognized as the world's strongest in 2016-2017. China is also known to already possess exascale computers. Japan's Fugaku, which demonstrated 442 petaflops (1 petaflop = 10^15 calculations) of computing power last November, also held the world number one spot.

According to KISTI, by next year, the U.S. plans to operate or possess four exascale supercomputers, including three exascale and one pre-exascale; China is estimated to have 3 to 6; the European Union (EU) and the United Kingdom plan 4 to 5; and Japan plans 3. The competition to own exascale supercomputers is even fiercer. By 2026, the U.S. plans to operate 7 to 9, China 6 to 10, the EU and the UK 8 to 10, and Japan 4. South Korea recently announced plans to possess one pre-exascale supercomputer of about 600 hecta-scale by next year and to develop an additional exascale supercomputer by 2026.

South Korea currently has six supercomputers listed in the Top500. Samsung Electronics' SSC-21 (25 petaflops), introduced in 2021, ranks 15th; the Korea Meteorological Administration's Guru (18 petaflops) and Maru (18 petaflops) rank 31st and 32nd respectively; KISTI's Nurion (14 petaflops) ranks 42nd; SK Telecom's Titan (6.3 petaflops) ranks 85th; and Samsung Electronics' SSC-21 Scalable Module (2.3 petaflops) ranks 315th. Currently, South Korea is evaluated to have a national computing support infrastructure ranked around 14th globally as of June 2019, based on Nurion, introduced in 2018 (which was ranked 11th at the time of introduction).

◇ How the National Supercomputer Has Been Used

South Korea introduced its first national supercomputer in 1988 and utilized it for domestic automobile design and manufacturing, as well as scientific weather forecasting. Domestic car manufacturers such as Kia, Daewoo, and SsangYong used the first supercomputer to simulate vehicle collisions, which significantly helped improve molds, engines, and tires. Previously, 200 test vehicles were required for new car development, but with supercomputer use, this was reduced to 100 vehicles, saving approximately 450 billion KRW in development costs annually. Crash simulations that once took 30 days per test were shortened to 1.5 days, reducing development time from five years to three. The first supercomputer also contributed to scientific weather forecasting by accurately predicting the paths of Typhoon Berry in September 1989 and Typhoon Mireille in September 1991, thereby reducing casualties and property damage.

The second supercomputer, built in 1993 with 16 gigaflops capacity, was used for basic scientific research. Seoul National University utilized it to study the physical properties of carbon nanotubes and their industrial applications, contributing to the development of South Korea's display industry. The Korea Aerospace Research Institute used the second supercomputer for combustion chamber simulations during the development of liquid engines, which formed the basis for the Nuri rocket, leading to improved efficiency of liquid rockets. The Korea Institute of Geoscience and Mineral Resources also developed technology for estimating underground resource deposits and selecting drilling sites through elastic wave simulations using the second supercomputer.

The third (introduced in 2002, 4.5 teraflops) and fourth (2008, 360 teraflops) supercomputers were mobilized in 2011 for the world's largest space simulation conducted by the Institute for Basic Science and the Korea Astronomy and Space Science Institute, contributing to uncovering the processes of space formation and evolution. The third supercomputer allocated 20% of its resources for 80 days to perform calculations that would have taken 60,000 years on a general PC. The fourth supercomputer has supported over 1,350 research projects since 2011, used annually by about 1,000 researchers, contributing to the production of 1,040 SCI-level papers. Among these were 17 high-quality papers published in the world's top three scientific journals, including the first empirical verification of Einstein's gravitational wave theory, the first elucidation of antimatter interactions, and the world's first measurement of neutrino mass differences, all benefiting from the fourth supercomputer.

Companies have also greatly benefited from using the national supercomputer. As of 2016, 44 companies utilized the national supercomputer for product development. Among these, 66.7% succeeded in developing new products, and each company hired an average of 1.8 new employees.

The fifth supercomputer, introduced in 2018 with a performance of 27.7 petaflops, is well known as 'Nurion' (model CS500). At the time of introduction, it boasted world-class performance ranked around 11th. It has supported various research projects in national strategic fields such as bio, energy, and materials, contributing to the publication of 275 SCI papers. It has been used in large-scale turbulence heat flow research jointly conducted by Yonsei University and KISTI, supercomputing-based cell-virus interaction research at Ulsan National Institute of Science and Technology (UNIST), and the world's largest cancer whole-genome big data analysis jointly researched by the National Cancer Center, KAIST, and Seoul National University Hospital.

Lee Sik, Director of the KISTI National Supercomputing Center, stated, "If we relied solely on overseas resources, it would cost about 1.8 trillion KRW, so by introducing the 6th supercomputer at around 300 billion KRW, we estimate saving about 1.5 trillion KRW. Since South Korea's R&D investment ranks fifth globally, strengthening ultra-high-performance computing capabilities is essential to at least maintain or improve our current ranking and secure scientific and industrial competitiveness."