2019/05/16

Standing shoulder to shoulder, shooting for the cloud, and kicking start the future of cloud intelligence for Taiwan 
Taiwan Computing Cloud (TWCC) starts its trial run 

這是一張圖片

TWCC-Taiwan Computing Cloud officially debuts 
From left, Taiwan Mobile GM Jamie Lin, Quanta Computer Chairman Barry Lam, Ministry of Science and Technology Minister Minister Liang-Gee Chen, National Applied Research Labs President Yeong-Her Wang, ASUS Deputy Chairman Xu Shi-chang, and National Center for High-performance Computing Director General Shepherd Shi

Toady on May 16th, National Center for High-performance Computing (NCHC), one of 8 centers under National Applied Research Labs (NAR Labs), an affiliation of Ministry of Science and Technology (MOST) handed together with industry companies ASUS, Quanta Computer, and Taiwan Mobile to jointly host Taiwan Cloud Computing Conference (TWCCcon), where NCHC announced the trial run of an AI supercomputer that has reached new heights in the world's ranking. It is slated to officially open to applications for use by research and academic institutions while industries may start applying to join in the fourth quarter. Minister Liang-Gee Chen of Science and Technology thanked all the teams involved for their efforts to bring this service into being. He pointed out that NCHC is the platform for the R&D of AI in Taiwan that will strive to help nurture the all-around development in industries, governments, academia, and research institutions.

We also unveiled for the first time Taiwan Computing Cloud (TWCC) for AI computing in the future, a service platform that had been jointly developed by the AI team with members from NCHC, ASUS, Quanta, and Taiwan Mobile. It is hoped that this platform could speed up the development of AI-related technologies and services in Taiwan and facilitate the innovation, value addition, and applications of applied AI. Minister Liang-Gee Chen of Science and Technology indicated that TWCC is a milestone in forward-looking digital construction as well as an important strategic investment for the development of AI industries in the nation. He looked up all parties involved in industries, academia, and research, after the completion of this platform, to join forces to nurture AI into the next trillion-dollar industry in Taiwan.

Director General Shepherd Shi of National Center for High-performance Computing pointed out that the team running the TWCC platform aims to provide world-class functionality, speed, information security, and service. As the AI R&D platform of the nation, TWCC strives to become fertile ground for more AI development teams to nourish and grow into world-class organizations. We strive not just to become leading edge in scientific research but also to help industries put their roots down locally.

這是一張圖片

Minister Liang-Gee Chen of Ministry of Science and Technology

這是一張圖片

National Center for High-performance Computing Director General Shepherd Shi

TWCC runs on the strength of Taiwania 2, the supercomputer built by the national AI team that bested all the other Taiwanese supercomputers in the world's ranking. It uses 2,016 NVIDIA Tesla V100 32GB GPUs that deliver 9 PFLOPS of superior performance. Its GPU system is on par with those used on the fastest supercomputers around the world, offering optimal energy efficiency and providing the highest performance. Taiwania 2 was designed and manufactured under the supervision of the NCHC. Taiwan Mobile was responsible for the construction of the high-caliber, international standards compliant data center and the operations and maintenance of telecommunications-grade information security. Quanta Computer was responsible for the platform for AI computing and data storage. ASUS was responsible for the cloud service platform, cloud servers, and the integration of AI applications.

“Quanta is honored to have participated in the building of Taiwania 2,” Quanta Computer Chairman Barry Lam said. “We're particularly proud of the knowledge that it was primarily Quanta that built the Taiwanese AI supercomputer that possesses a super-fast computing capability, runs super energy-efficiently, and climbed in the world's ranking of supercomputers.” Lam continued: “After the national AI team has worked closely together for nearly a year, TWCC is now about to go online to provide services. We sincerely hope that this national-level installation will mark a historic milestone in the development of AI in Taiwan.”

ASUS servers business GM Jin Qing-bo said, “ASUS put the cloud R&D and software and hardware integration prowess of its server unit, cloud infrastructure software center, and its subsidiary ASUS Cloud to help the NCHC build in a mere seven months the TWCC, a comprehensive solution to Taiwan computing cloud from cloud mainframe to, platform software.” Jin continued: “We hope that the availability of the TWCC will provide strong backing for the R&D development energy of industrial, governmental, academic, and research institutions in Taiwan and will become the source of wisdom and intelligence to transform the industries in Taiwan.”

Taiwan Mobile GM Jamie Lin indicated that the Taiwan AI A Team--with members from Taiwan Mobile, Quanta Computer, and ASUS--was announced in 2018. The team has laid a solid foundation for the development of AI in Taiwan since its formation. The team has delivered distinctive results. Taiwan Mobile built the high-caliber, international standards compliant data center and the operations and maintenance of telecommunications-grade information security. Quanta Computer was responsible for AI computing and the data storage platform. ASUS was responsible for the cloud service platform, cloud servers, and the integration of AI applications. Taiwan Mobile will in the future integrate AI with 5G strategies to develop AI applications to be used in Taiwan Mobile and the Fubon Group, such as telecommunications, media, retailing, e-commerce, finance, and healthcare in order to quicken the speed of the transformation of Taiwanese businesses to AI and the speed of making AI an industry.

這是一張圖片

Quanta Computer Chairman Barry Lam

這是一張圖片

ASUS servers business GM Jin Qing-bo 

這是一張圖片

Taiwan Mobile GM Jamie Lin

The latest container techniques[1] help TWCC deliver its services. Through optimized AI software stacks, developers can quickly build their developmental environment at a 40% efficiency advantage over current methods. TWCC also supports large numbers of nodes affinity and GPUs, high-speed parallel computing across nodes affinity that provides 30% more service than the current cloud service. In addition to offering rapid computing capabilities, large storage spaces, and secured networks, the TWCC will also integrate AI programs, tools, and significant databases across the globe into the largest model store and data store in Taiwan to give users in industries, academia, and research more timely and more convenient computing services.

Aside from officially opening the TWCC platform, the TWCCcon also featured such activities as sharing of key trends and user experiences. Key trends were given by Quanta Computer Chairman Barry Lam while Nvidia Global VP and Taiwanese GM Qiu Li-meng shared AI trends and challenges in Taiwan and abroad. The ASUS spoke on behalf of the software development team about TWCC software application techniques. User experiences were given by users of the beta version: Umbo Light , aetherAI, HTC DeepQ, Taipei Veteran General Hospital, Tsing Hua University, Chiao Tung University, and Cheng Kung University.

The year of 2019 is the inaugural year of TWCC speeding Taiwan up toward AI. To improve the exchanges between developers and users, the NCHC of the NAR Labs will continue to host meetings that center on AI technique or AI R&D platform. It is our hope that lectures, teaching, experience sharing and such exchanges may spark imagination and energy and lead to bilateral exchanges and collaborations. AI may bring innovations and value to every professional endeavor in Taiwan. 


[1] Docker container is a lightweight virtual technique that starts and downloads faster than traditional virtual machines, takes up a smaller hard disk space, transplants quickly, and deploys easily. Currently, each container that we provide uses 1 - 8 concurrent GPUs.