China's supercomputer development history:
1. 2010
In 2010, the "Tianhe-1A" gave China the world's fastest supercomputer for the first time, but because it was not followed by any further upgrades, it was soon squeezed down.
2. 2015
On April 9, 2015, the U.S. Department of Commerce issued a bulletin banning the sale of XEON chips to China's four national supercomputer centers, a decision that has hampered upgrades to the Tianhe-2 computer.
3. 2016
In June 2016, China had developed the world's fastest supercomputer, the Shenwei TaihuLight, which is now housed at China's National Supercomputer Center in Wuxi. The supercomputer's floating-point speed is twice as fast as the world's second-fastest supercomputer, Tianhe-2, also developed in China, at 930 million times per second.
4, 2019
In the newly released TOP500 list, the "Shenwei TaihuLight" supercomputer ranked third on the list with an HPL mark of 93.0 petaflops, and the "Tianhe-2" supercomputer ranked third with 61.0 petaflops. "Tianhe-2" supercomputer ranked fourth with 61.4 petaflops.
The total number of U.S. supercomputers on this list is 118, near an all-time low, but in terms of total computing power, it accounts for 37.8 percent of the list's total performance, with China accounting for 31.9 percent.
Predictions for the future development of supercomputers
China's supercomputer applications have gradually broadened from national security, nuclear weapons development, weather forecasting, oil exploration, such as national strategic areas, expanding to the Internet, big data, artificial intelligence, gene sequencing, film and television production, finance and other fields, benefiting a variety of different The industry is getting closer and closer to the national economic life.
In the future, the fusion of supercomputing and container clouds will also become a new product. This type of high-performance cloud has four major advantages, including: containerization. That is, with microservice architecture, rapid deployment of services to meet the needs of the enterprise cloud, lightweight and loosely coupled, more efficient and faster.
Simplifies the threshold of creating, managing, operating and maintaining application services, supports standardized visual orchestration template import and export, drawing that is to create the pipeline. Code submission that is deployed, automated management to reduce the complexity of the process, continuous integration, continuous release and continuous deployment of elastic scalability. Dynamically adjust the scale of application deployment without human intervention, improve the automation of service expansion, real-time contraction according to the threshold to save resources.