Judging from the number of supercomputers in various countries in the TOP 500 list of global supercomputers released in November 2021, China has a total of 173 supercomputers on the list. The number of supercomputers on the list has ranked first for 9 consecutive times, but the number has decreased by 13 compared with the previous period. . The United States ranked second with 150 units, an increase of 27 units from the previous period. The number of supercomputers on the list in China and the United States accounted for nearly 2/3 of the total number of the list. The TOP 10 list has not changed much. Japan's Fugaku won the championship for the fourth time. my country's fastest supercomputer Sunway TaihuLight is still at No. 4, and Tianhe-2 is still No. 7.
In recent years, my country's supercomputing has entered a stage of rapid development. my country's initially formed supercomputing industry chain consists of upstream, middle and downstream. The upstream mainly includes hardware resources (computing, storage, network, etc.), software resources (basic software, application software, etc.) etc.), supporting infrastructure resources (power distribution, refrigeration, etc.), the midstream is to integrate upstream resources, provide powerful supercomputing resources and provide supercomputing services and solutions for related industries, and the downstream is the application layer, including supercomputing Derivative industries and key application fields (see figure).
The content comes from " China Supercomputing Cloud Service Independent Market Research (2022) " . Start with the report on my country's supercomputing service and cloud service market, and analyze the market status, driving factors and future trends to further understand the supercomputing cloud service market and understand the pattern and future trends.
Download link:
Independent market research on supercomputing cloud services in China (2022)
Difficulties and breakthroughs of Chinese AI companies
2022 China DPU Industry White Paper
Triple Opportunities for AI Acceleration (2022)
2022 Xinchuang Industry Development Research (Soochow Computer)
Seven Technology Trends of Advanced Computing in 2022
A Series of Research on Chinese Neural Network Models
Supercomputing, also known as high-performance computing, refers to the use of centralized computing resources of multiple computer systems working in parallel to deal with extremely complex or data-intensive problems. Supercomputing has extremely strong computing power, and its floating point operation speed can reach petaflops per second. As one of the important commanding heights of the country's scientific and technological development, supercomputing is not only the embodiment of the country's scientific and technological strength, but also an important guarantee for the country's high-tech development.
Similar to traditional supercomputing services, the main service scenarios of supercomputing cloud also focus on industrial simulation, artificial intelligence innovation, biological information, meteorology, energy and other aspects. Different from traditional supercomputing, supercomputing cloud service sacrifices part of the computing performance of the cluster in order to achieve greater flexibility of computing resources. In the field of cutting-edge supercomputing that pursues extreme performance, supercomputing cloud services are more of a temporary supplementary computing resource, and the business preference is not high enough. Generally speaking, the core scenarios of the supercomputing cloud service market focus on general-purpose supercomputing and business supercomputing. In addition, artificial intelligence is also one of the core service scenarios of supercomputing cloud services.
The CPU and GPU market, the core hardware of supercomputers, shows a Matthew effect. The CPU is mainly led by Intel and AMD, and the GPU market is controlled by international giants such as Nvidia. In recent years, my country has attached great importance to the independent localization research of supercomputer core chips. Haiguang’s domestic x86 architecture CPU, DCU depth computing unit, National Defense Science and Technology Feiteng CPU, and Jiangnan Computing Technology Research Institute’s Shenwei processor have all obtained large-scale applications.
From the perspective of demand for computing power resources, high-performance computing can be divided into four categories: cutting-edge supercomputing, general-purpose supercomputing, business supercomputing, and artificial intelligence supercomputing.
Independent market research on supercomputing cloud services in China (2022)
By providing high-performance computing capabilities, the supercomputing industry empowers scenarios such as data recovery and analysis, and data simulation. From the perspective of service scenarios, the traditional supercomputing industry focuses on meteorology and oceanography, geological exploration, industrial simulation, and physical simulation. Most of them are colleges and universities, scientific and technological research institutes, and government-led scientific and technological engineering projects. With the advent of the era of the Internet, big data, and the Internet of Things, more and more companies have stronger and stronger demands for data analysis. In the face of massive data analysis and processing requirements, general computers and servers are no longer capable of meeting technical requirements, such as multimedia rendering, molecular model simulation, collision experiments, etc. In the future, under the empowerment of technology, more and more enterprise service scenarios will face significant demand for high-performance computing services.
The content comes from " China Supercomputing Cloud Service Independent Market Research (2022) " . The report analyzes China's supercomputing service market, China's cloud service market, China's supercomputing cloud service market, and the competition landscape.
Download link:
Independent market research on supercomputing cloud services in China (2022)
Difficulties and breakthroughs of Chinese AI companies
2022 China DPU Industry White Paper
Triple Opportunities for AI Acceleration (2022)
2022 Xinchuang Industry Development Research (Soochow Computer)
Seven Technology Trends of Advanced Computing in 2022
A Series of Research on Chinese Neural Network Models
Neural Network Series Research (2): Recurrent Neural Network
Neural Network Series Research (1): Convolutional Neural Network
Artificial intelligence industry series topics (full)
1. Artificial intelligence series (2): Application status and trends of the three major application scenarios of AI chips.pdf
2. Artificial Intelligence Series (3): Pain Points and Challenges of Deepening Application of China's Deep Learning Framework.pdf
3. Artificial Intelligence Series (4): Three Core Competencies of Artificial Intelligence Development Tools and Products.pdf
4. Artificial Intelligence Series (V): The Status and Value of Pre-Training Models in the AI Industry Ecology.pdf
5. Artificial Intelligence Series (1): Three Realization Paths of Intelligent Data Labeling Technology.pdf
Smart driving chips and software field combing
Artificial intelligence learning preparatory knowledge series empowerment
Chapter 1 Overview of AI
Chapter 2 Basics of Python Programming
Chapter 3 Basic Knowledge of Mathematics
Chapter 4 Introduction to Tensorflow
Chapter 5 Deep Learning Preliminary Knowledge and Deep Learning
Chapter 6 Overview of HUAWEI CLOUD EI
All the materials of this account are uploaded to Knowledge Planet. For more content, please log in to Intelligent Computing Core Knowledge (Knowledge Planet) Planet to download all the materials.