The next decade of intelligent computing is a golden period

Only by embracing intelligence can we calculate the future. At the International Conference on Intelligent Computing held in Shenzhen yesterday, academicians and experts believed that the current world pattern is undergoing profound changes, and intelligent computing is leading a new wave of electronic information technology innovation and reshaping the world pattern. Such a new era is precisely the supercomputer. golden age. New computer architectures will be launched in the next decade, and swarm intelligence will be the next breakthrough in artificial intelligence.

New computer architecture “explodes” in the next ten years

Li Guojie, academician of the Chinese Academy of Engineering, chief scientist of the Institute of Computing Technology, Chinese Academy of Sciences, and chairman of Dawning Information Industry Co., Ltd. delivered a speech entitled “Some Understandings of Intelligent Supercomputing”. In the report, Academician Li Guojie made some historical reviews of intelligent computing, expounding the difference between traditional computing and intelligent computing. Academician Li Guojie believes that the “Cambrian” explosion that will occur in the next decade requires the establishment of an intelligent supercomputing benchmark test program first.

Li Guojie said that the more complex the structure, the simpler the algorithm can be. The human brain is intelligent because the structure of the brain is very complex. In the future, artificial intelligence requires not only computing power, but also a more complex hardware structure, and new intelligent computers may be invented.

Regarding the future research direction of intelligent supercomputing, Li Guojie further predicted that the next ten years will be the golden age of architecture, and computer architects in academia and industry will usher in an exciting era. In the next ten years, there should be major architectural inventions like IBM360 and RISC, and Chinese scholars should make contributions worthy of the times.

Li Guojie also said that big data and artificial intelligence have higher and higher requirements for storage bandwidth, and the future computer structure should change the traditional von Neumann architecture. One is to implement neural network computing in the group variable memory, and to do deep learning in the memory, the power consumption can be reduced by 20 times, and the speed can be increased by 50 times. The other is to use 3D stacking, which is to stack more memory devices in a processor cycle. It is hoped that there will be a set of standards from supercomputing to big data and artificial intelligence, and there will be a ruler to measure technology, extending its influence from academia to industry.

Swarm intelligence is the next breakthrough

How did the microprocessor develop? Yang Xuejun, an academician of the Chinese Academy of Sciences and president of the Academy of Military Sciences, believes that the combination of analog and digital will be the “exciting point” in the future. In his speech “micros: The “New Brain” of Robots”, Yang Xuejun believed that science and technology have entered the era of “intelligence explosion”, and the cross-integration of artificial intelligence and other science and technology is triggering a “robot revolution”.

Yang Xuejun said that the robot operating system will become the engine of the “robot revolution” and the “new brain” for robots to achieve autonomous behavior, group collaboration and human-machine integration.

“Swarm intelligence that can adapt to the environment is the next breakthrough in artificial intelligence.” Yang Xuejun proposed to design a new technical software micros, which is an operating system for polymorphic intelligent swarm robots. He believes that the scientific problems of such an operating system are: autonomous behavior and swarm intelligence.

“I hope to break through problems and algorithms along these three stages.” Yang Xuejun said that they hope that the artificial intelligence algorithm of swarm intelligence will progress in several stages: one is to crack the structural self-aggregation mechanism of intelligence; the other is environmental adaptability and task adaptability , the self-organization mechanism; the third is the emergence of the self-evolution mechanism.

Shenzhen accelerates the layout of the next generation of supercomputing

At present, supercomputers are rapidly entering the era of intelligence. The global supercomputer market is expected to double from 2017 to $9.5 billion by 2022, according to research firm Hyperion Research. Feng Shengzhong, chairman of the conference, said that the current generation of intelligent supercomputers has become a benchmark for international competition in the computer industry.

At present, Shenzhen is also accelerating the deployment of the next generation of supercomputers. It is proposed in the outline of Shenzhen’s 13th Five-Year Plan to promote the development of the National Supercomputing Shenzhen Center. In the 2018 government work report, it was also proposed to start the construction of large scientific installations and promote the construction of the second phase of the National Supercomputing Shenzhen Center. According to reports, E-class computers are expected to settle in Shenzhen in 2022.

The Links:   3HNA011334-001 3HAC050363-001