Current location - Education and Training Encyclopedia - University rankings - Tianhe-1 is running at full capacity. What are the five supercomputers in China doing?
Tianhe-1 is running at full capacity. What are the five supercomputers in China doing?
Tianjin National Supercomputer Center (2009)

National Supercomputing Tianjin Center is the first supercomputing center approved by the Ministry of Science and Technology in May 2009. It is jointly built by Tianjin Binhai New Area and National University of Defense Technology.

Tianjin Supercomputing Center is located in Building 5 of Tianjin Economic and Technological Development Zone (TEDA) Outsourcing Service Park, covering an area of about 8,500 square meters. * * * There are two large computer rooms with an area of about 4,000 square meters * * * One of them is used to house Tianhe-1 supercomputer, and the second one is used for cloud computing and system expansion. In addition, there are substations and refrigeration stations. The power supply capacity of the substation is 13600 kva, and the refrigeration capacity of the refrigeration station is 9,600KW, which has strong supporting capacity.

The main business computer of Tianjin Supercomputer Center is "Tianhe No.1" supercomputer, which is supported by the major project "High Performance Computer and Grid Service Environment" of the 863 Program of the Ministry of Science and Technology. The first phase of the system was successfully developed by National University of Defense Technology and Binhai New Area in September 2009. Tianhe No.1 Phase II system was successfully developed on 20 10 and 10 after comprehensive optimization with independent CPU and independent high-speed interconnection communication system. Tianhe No.1 Supercomputer ranks first in the world in the top 500 supercomputers of 20 10,1world.

The main application fields of high performance computing in Tianjin Supercomputing Center include biomedicine, data processing of petroleum seismic exploration, animation and film rendering, new materials and new energy, design and simulation of high-end equipment, aerospace, fluid mechanics, weather forecast, climate prediction, simulation and analysis of marine environment, etc.

Shenzhen National Supercomputer Center (2009)

The project was approved in May, 2009, and was installed and debugged in June, 201,and the supercomputer center was put into operation in June of the same year 1 1. On June 20 12, the commercial service of high-performance computing services was officially provided to the society; By May of 20 14, the utilization rate of computing resources exceeded 55%, the number of high-performance computing users reached 1 0,056, the number of personal cloud computing users exceeded 1 0,750,000, and the number of institutions reached13,21.

The National Supercomputing Shenzhen Center (Shenzhen Cloud Computing Center) is a major science and technology infrastructure project with the largest single investment since the national layout and construction in Shenzhen. This project is a national 863 plan, a major project in Guangdong Province and Shenzhen City, and also a concrete action for Shenzhen to implement the Outline of the Pearl River Delta Reform and Development Plan (2008-2020) and the Comprehensive Reform Plan of Shenzhen City.

The mainframe system of Shenzhen National Supercomputing Center (Shenzhen Cloud Computing Center) passed the test of the World Supercomputing Organization in May 20 10, and its computing speed reached 1 27 1 trillion times per second, ranking second in the world. At the same time, it is equipped with massive storage up to 17.2PB and rich network bandwidth resources of major operators and educational networks.

National Supercomputing Shenzhen Center (Shenzhen Cloud Computing Center) is based in Shenzhen, facing the whole country, serving South China, Hong Kong, Macao and Taiwan, Southeast Asia, and providing various large-scale scientific computing, engineering simulation, animation rendering and other computing services, and providing cloud computing services for the society with its powerful data processing and storage capabilities. It will build a world-class high-performance computing R&D center and cloud computing service center with complete functions, rich platforms, high efficiency and energy saving.

Changsha National Supercomputer Center (20 1 1 year)

Changsha National Supercomputing Center 20 1 1 was put into trial operation, and 20141.4 was officially put into operation. The third national supercomputing center approved for construction after Tianjin and Shenzhen. Different from other supercomputing centers in China, Changsha Supercomputing Center is completely operated by universities.

Venue: On the campus of Hunan University, Tianhe No.1 high-performance computer of National University of Defense Technology was adopted, and the planned construction capacity was 1 1,000 trillion times per second, with a total investment of 720 million yuan. The first phase of Changsha Center is planned to have a construction area of 30,000 square meters, and it is planned to be completed by the end of 2065, with 438+065,438+0. After completion, the computing capacity will reach 300 trillion times per second, which will be operated by Hunan University, and the National University of Defense Technology will provide computing equipment and technical support, adhere to the principle of combining public welfare with commerce, and provide high-performance computing application services for the society and the public.

Since the trial operation of 20 1 1, the center has provided high-performance computing platform services for public service departments such as meteorology, land, water conservancy, health/medical care, transportation and so on. The Center has established strategic cooperative relations with domestic institutions in the fields of high-performance computing, cloud computing and animation rendering, and has tried it out in some large enterprise platforms inside and outside the province. After it is officially put into operation, it will provide large-scale simulation design services for national equipment manufacturing enterprises.

20 10,11June laid the foundation stone, and actively promoted the project construction according to the mode of "government-led, military-civilian cooperation, joint construction of provinces and schools, and market operation". The main equipment of the project was put into trial operation on the Internet in June of 20 1 1 year, and the main building of the project was completed in July of 201year. At the same time, the main equipment has been moved from National Defense University to the new site of Hunan University, and the installation and debugging have been completed, which has met the basic conditions for formal operation.

Jinan National Supercomputer Center (20 1 1 year)

Jinan National Supercomputing Center is one of the four trillion supercomputing centers approved by the Ministry of Science and Technology, with a total investment of 600 million yuan. The main body of construction is Shandong Academy of Sciences, and its subordinate unit, Shandong Computing Center, is responsible for construction, management and operation.

Jinan Center officially started construction on March 20 1 1, and was completed and unveiled on October 27 10 that year, providing external computing services. The successful construction of Jinan Center marks that China has become the third country to build a petascale supercomputer system with its own processor after the United States and Japan. Tested by the national authoritative organization, the continuous performance of Shenwei Blu-ray supercomputer system in Jinan Center is 0.796PFlops(PetaFlops, petaflops/second), the efficiency of LINPACK is 74.4%, and the performance-power ratio exceeds 74 1MFlops/W (petaflops/second watt). The assembly density, performance and power are more advanced than those in the world, and the comprehensive level of the system is now in the world.

Relying on Shandong Academy of Sciences, Jinan Center has set up a scientific research service team integrating technology research and development, computing services and technical support, working closely with experts and application units at home and abroad, providing computing and technical support services for marine science, modern agriculture, oil and gas exploration, climate and meteorology, drug screening, financial analysis, information security, industrial design, animation rendering and other fields, undertaking major scientific and technological or engineering projects at the national, provincial and ministerial levels, and providing platform support for China's scientific and technological innovation and economic development. Center positioning: public welfare high-performance computing technology and consulting service institutions.

Jinan National Supercomputing Center can analyze smog.

Although the smog problem still exists, the final result is not satisfactory, but in the face of smog, everyone should know a series of efforts behind it. Taking smog forecast as an example, today's technical level can accurately predict the air quality in the next three days or even longer. Dr. Tang Xiao, associate researcher of lasg State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, said at the 20 15 Intel High Performance Computing Summit held recently that all kinds of early warning measures released at present are based on the judgment of air quality forecast.

Although it seems that the smog forecast may not be as mature as the weather forecast, what you may not know is the difficulty behind the smog forecast.

Tang Xiao believes that air pollution is a very complicated process. It depends not only on the weather but also on the chemical reaction process. For a simple example, the air pollution in China in the 1990s was called soot pollution, and now it is called compound pollution. At that time, when burning coal, the primary emission concentration of sulfur dioxide was very high, which was harmful but not easy to be observed by the naked eye. Now, you will obviously observe the influence of air pollution. The important reason is that primary pollutants are converted into fine particles, namely PM2.5, by chemical reaction, which has a great influence on atmospheric visibility. Tang Xiao said that there are interactions between chemical changes and meteorological processes.

Because many factors are involved, the final result is unpredictable. You have to analyze how many of these changes have occurred, how many have changed, and how many have spread with meteorological conditions ... This process requires huge computing power on the one hand and accurate computing models on the other. So far, in terms of institutional model, after years of exploration, China has built a model that conforms to China's national conditions. At the same time, China National Environmental Monitoring Center cooperated with Dawning and Intel several years ago to establish a high-performance computing platform for smog prediction and early warning.

It is estimated that no one dares to say that such a system can meet the demand. As Tang Xiao said in an interview, with the increasing demand for scientific research, they need more detailed monitoring, which means that more data will be generated eventually, and it will grow in the form of orders of magnitude; On the other hand, the calculation of chemical reaction with one emission is very demanding, because it is not one or two independent chemical reactions at all, but hundreds of chemical changes that may react with each other. This is a considerable demand for the final calculation.

In fact, the challenges faced by smog forecasting can be understood as the epitome of various application difficulties that require ultra-large-scale computing power. The demand space is still very large, but the current technical level is still very limited. For example, Iflytek once said in an interview that even the fastest supercomputers can't meet their actual needs in terms of computing power. So the next best thing is to think of some optimization and improvement measures besides computing power. This can be summed up as an important purpose of Intel High Performance Computing Summit, "to create a communication platform for everyone and share any good suggestions and practices."

In addition to smog, this conference also includes practice sharing in celestial observation, deep learning, marine climate and other fields ... Undoubtedly, high-performance computing is increasingly affecting all aspects of our lives, and the author sincerely hopes that technology can make our lives better. For example, on the smog problem, we can not only make predictions, but also give some more meaningful and easy-to-implement solutions for each region and even each factory. After all, economic development is still the top priority.

Guangzhou National Supercomputer Center (20 13)

Guangzhou National Supercomputing Center, jointly built by Guangdong Provincial People's Government, Guangzhou Municipal People's Government, National University of Defense Technology and Sun Yat-sen University, is Guangzhou Science and Technology No.1 Project, which is a major strategic infrastructure to promote the development of strategic emerging industries and support the construction of a national innovative city and a smart Guangzhou. It will become a world-class supercomputer center integrating high-performance computing, massive data processing and information management services, providing a powerful engine for the economic and social development of Guangzhou, Guangdong and even the whole country. Guangzhou Center is located in the scenic campus of Sun Yat-sen University in Guangzhou University Town, with a total construction area of 42,332 square meters (5 floors above ground, about 32,332 square meters), of which the area of computer room and auxiliary rooms is about17,500 square meters, including the main engine room, storage room, high and low voltage distribution room, cooling equipment room and auxiliary rooms.

Since April 20 14, Tianhe-2 has been used by nearly 700 users, supporting more than 100 national projects, more than 20 applications with more than100000 cores and 9 applications with more than100000 cores. The output of available computing resources of "Tianhe No.2" in 20 15 years has exceeded the sum of available resources of two Tianjin supercomputers "Tianhe No.1 A" in full operation. The application scope covers materials science and engineering computing, biological computing and personalized medical care, digital design and manufacturing, digital design of energy and related technologies, astronomy and cosmology, earth science and environmental engineering, finance, economy, smart city big data and cloud computing. Especially in the fields related to the national economy and people's livelihood, such as space science research, pollution control, design and manufacture of large aircraft, design and manufacture of high-speed trains, assembly and sequencing of large genomes, biomedicine and drug screening in Qualcomm.