Asian Scientist Magazine (Nov. 14, 2022) —In the spring of 2020, amidst a crippling pandemic, Japan’s Fugaku supercomputer, eponymously named after Mount Fuji, roared to life—thanks to the unwavering resilience of the teams at RIKEN, the country’s largest research institution, and Fujitsu.
Named after the highest peak in Japan, Fugaku symbolizes the nation’s aspirations for a supercomputer with towering capabilities. Indeed, shortly after its debut, the data-crunching beast scaled its way to the summit of the TOP500 list, making Asia home to the world’s fastest supercomputer at the time.
“Fugaku as a supercomputer has never been about ranking,” said Professor Satoshi Matsuoka, director of the RIKEN Center for Computational Science, during an interview with Supercomputing Asia in 2021. “Rather, our goal was to achieve application results and swiftly respond to important and difficult societal goals and problems.”
Still Well In Its Prime
Though Fugaku has recently lost its position as the world’s fastest supercomputer to Frontier of the United States’ Oak Ridge National Laboratory, it’s noteworthy that Fugaku has held its crown for two consecutive years—an impressive feat that will no doubt leave behind a legacy.
Above all, Fugaku offered a beacon of hope in Japan’s battle against the novel coronavirus. Made accessible almost a year ahead of schedule, the supercomputer turbocharged a swath of computationally intensive research to suppress the wrath of COVID-19. From deducing the recommended partition height to preventing virus spread, and from modeling the dispersion of droplets to assessing the effectiveness of face masks and more, Fugaku was instrumental in informing policy responses to keep the nation’s infection rate at bay during the early stages of the pandemic.
As the world moves towards an endemic phase, Fugaku remains active in a panoply of scientific and industrial ventures. Marrying powerful compute with intelligent algorithms, researchers have revealed the origin story for carbon-12—an essential building block for life—in the turbulent cosmos, as well as slashing the runtimes of high-precision aircraft simulations from hours to mere minutes, among many other accomplishments.
While Japan is keen to build on Fugaku’s achievements by grooming a successor, several countries in Asia are also the epicenter of exciting developments in high performance computing (HPC). Whether modeling molecular structures in quantum chemistry, assembling de novo genomes, evaluating the climate, or probing the depths of the universe, HPC is accelerating businesses and research workflows across research fields and industries.
HPC Giving Healthcare A Boost
From artificial intelligence (AI) algorithms that read X-rays to personalized treatments underpinned by affordable genomic sequencing and population-wide health interventions impelled by big-data analytics, HPC has played a pivotal role in the evolution—and revolution—of the medical technology and healthcare landscape.
“Medical and health practitioners now have an increasing ease of access to data, which houses a vast range of information and resources,” said Associate Professor Yeo Khung Keong, deputy group chief medical informatics officer (research) at SingHealth, in an interview with Supercomputing Asia. “The massive amount of knowledge globally exceeds the ability of any individual clinician to master, and technologies such as supercomputers will play a vital role in processing or interpreting information to derive meaningful knowledge to advance medicine.”
Such a sentiment is also shared by the annual Supercomputing Asia conference held in March this year. The international conference, conducted in a hybrid format combining an in-person event with online components, emphasized the growing ubiquity of advanced supercomputing resources in a myriad of fields.
Among the many formal agreements made at the conference, a new Singapore-based public private partnership aims to harness the power of supercomputers to solve complex healthcare issues. A tripartite collaborative effort, the project brings together the National Supercomputing Center (NSCC) Singapore, Singapore’s largest group of healthcare institutions SingHealth, and US-based chipmaker Nvidia Corporation. to develop and deploy a new supercomputer housed at the Singapore General Hospital.
As part of the project, Nvidia will breathe life into the compute engine by contributing their unique software development tools and pre-trained AI models. In addition, accessibility is on the project’s radar—NSCC’s Supercomputing Digital Sandbox environment will ensure researchers who don’t speak the HPC language can still leverage the system for their work.
“Having supercomputing infrastructures equipped with AI and deep learning capabilities will provide unprecedented support for SingHealth’s clinicians and researchers to better understand prevalent chronic diseases as well as gain insights into diagnosis and treatment,” said Yeo.
In medicine, there’s rarely a one-size-fits-all approach—two individuals with the same disease may respond very differently to the same treatment. This is where precision medicine comes in, treating patients based on individual genomic, phenotypic and lifestyle factors. In this regard, SingHealth seeks to further unlock the potential of the human genome with HPC. Using HPC-driven AI algorithms, medical scientists can better identify neoantigens, a repertoire of proteins that forms on mutating cancer cells that is specific to individual patients. Such discoveries will help to accelerate the development of potential personalized cancer therapies, which could directly destroy cancer cells with minimal to no side effects to the patient.
An Emerging Supercomputing Giant
Up north from Singapore, HPC has had its roots set in Thailand, dating back to the first Cray supercomputer installation in the early 1990s. Growing from its foundations, the country commonly known as the Land of Smiles is certainly offering more than just that—it has ambitious plans to expand its supercomputing network to take on new challenges that would drive progress across academia, government and industry.
Playing a vital role in realizing the country’s HPC goals, the National Science and Technology Development Agency has teamed up with Nvidia and Hewlett Packard Enterprise (HPE) to build a new system at the Thai Supercomputer Center (ThaiSC). Touted as the largest of its kind in Southeast Asia, the new “LANTA” supercomputer, powered by HPE Cray EX, is adorned with a whopping 704 units of Nvidia A100 Tensor Core graphical processing units (GPUs)—twice as many as those in Singapore’s latest ASPIRE 2A supercomputer, one of the fastest in the region.
“The system is created as an end-to-end, scalable HPC platform that supports a broad range of workloads,” said Dennis Ang, senior director of enterprise business for ASEAN and ANZ at Nvidia, in an interview with Supercomputing Asia.
According to Dr. Anek Laothamatas, Thailand’s Minister of Higher Education, Science, Research and Innovation, the new supercomputer is especially important for national disaster predictions. Being able to analyze vast quantities of climate data enables researchers to model and predict weather systems more accurately than what is achievable in a traditional set-up.
With adaptation now part and parcel of climate resilience, such research could arm industries from agriculture to architecture with panaceas in the face of adverse climate impacts. For instance, farmers can plant drought-resistant crops when the forecast calls for little to no rainfall. City planners can also adjust building codes for safer construction when destructive storms are anticipated.
“LANTA will provide HPC service on a national scale,” said Dr. Piyawut Srichaikul, director of ThaiSC to Supercomputing Asia. “It will enable our scientists to break through the current constraints in computing resources and pursue frontier science in many competitive research areas.”
Shaping The Future Together
Putting its pandemic woes behind it, India is set to become the world’s fastest growing major economy for two consecutive years. On the HPC front, however, the nation has historically struggled to garner a supercomputing advantage—but that’s about to change. The National Supercomputing Mission, announced in 2015, endeavors to catapult the nation as a frontrunner in HPC. Executed in three phases, involving assembly, manufacturing and design—all done locally—the mission will increase India’s homegrown supercomputing capability to 45 petaFLOPS.
With ten supercomputers installed so far, this year sees nine more ready for action at leading Indian institutes. This will be a tremendous boon to hundreds of local institutes and thousands of active researchers working through the National Knowledge Network, the country’s pillar for supercomputing systems.
Revamp to Reinforce
Rather than reinventing the wheel, some strategic upgrades and updates could do the trick to refresh and reinvigorate a HPC ecosystem too. Outlined in the 2021 National Research Infrastructure Roadmap, a government strategy in Australia is set to provide the computing resources, digital tools, data governance frameworks and expertise needed to make best use of data. This includes establishing a more flexible, integrated computing ecosystem, where both national and commercial HPC resources are accessible to researchers.
“The aim of the roadmap is to encourage greater commercialization of research by allowing industry and researchers to engage more effectively,” said former Australian Minister for Science and Technology Melissa Price in a press release.
Meanwhile, the New Zealand eScience Infrastructure (NeSI) is upgrading its Mahuika HPC cluster. Since it came online in 2018, Mahuika has witnessed its users increase 50 percent year-on-year, empowering the nation’s scientists to tackle unique research ranging from simulating earthquake behaviors to managing biodiversity, and collaborating with the country’s indigenous people and organizations to advance the wellbeing of New Zealanders.
A partnership between the University of Auckland, University of Otago and Manaaki Whenua—Landcare Research, the NZ$2.1 million investment will double the performance and boost the computational capabilities of the HPC system. This will ensure the island’s national computing platforms remain responsive and highperforming to satisfy a growing array of data-centric and data-intensive research in biochemistry, physics, bioinformatics and more.
“We’re seeing motivation within domains to advance and grow their own digital capability,” said NeSI director Nick Jones, in an interview with Supercomputing Asia. “For example, there has been a dedicated drive within the genomics and bioinformatics fields to deliver and expand training and upskilling opportunities to its researchers.”
Not Without Its Challenges
Whether urgently addressing the climate crisis at hand, creating new medical innovations or fostering a more digitally woven society, the role of HPC is irrefutably central to navigating an unprecedented global transformation in many areas.
“Supercomputing in Asia has evolved along with the global HPC landscape,” said Ang. “Collaboration is a key enabler—technology providers team up with local experts and agencies to ensure the latest innovations are available to power trailblazing research.”
Nevertheless, advancing the frontiers of HPC in the region brings with it a set of equally challenging, and at times contradictory, problems that also require cross-border cooperation and collaboration.
“The unrelenting data explosion and the need for increasingly complex simulations call for even more computational power in the near term,” added Yeo. “This means HPC will have a growing carbon footprint due to the associated increase in energy consumption, which is an obstacle that must be surmounted in the current state of the global climate.”
“As a region we’re recognizing common challenges in climate change and our environments, and we’re inherently connected by our oceans,” said Jones. “Therein lies our potential for greater connection and joint ambition. How we achieve this is open for exploration, and we’re looking forward to learning more about how others are approaching these spaces.”
This article was first published in the print version of Supercomputing Asia, July 2022.
Click here to subscribe to Asian Scientist Magazine in print.
Copyright: Asian Scientist Magazine. Image: Shelly Liew