How to make computing power easier?The academician of the Chinese Academy of Engineering, Wu Hexuan, responded to each scripture: not allowed the work of computing power to build only a very small number of enterprises. It is necessary to reduce the use threshold

Under the wave of digital economy, computing power has become a new type of productive forces after heat and electricity.

What are the obstacles to the transformation of computing power into productivity? How to solve it?

At the 2022 China Computing Power Conference held from July 29th to 31st, "Daily Economic News" reporters interviewed the academician of the Chinese Academy of Engineering and the former vice president of the Chinese Academy of Engineering.

Wu Hexuan said that although we have produced a lot of data, the proportion of data storage is not high enough, and the proportion of storage is even lower.

During the conference, the reporter interviewed the picture of Wu Hexuan.

Give full play to the ability of the data from the source

Computing power is regarded as the "base" of the development of the digital economy, but in reality, the computing power does not seem to be fully transformed into productivity. What are the obstacles to transform computing power into productivity? How to solve it?

Regarding the issue raised by the reporter of the Daily Economic News, Wu Heyi said that we hope that the computing power can be used as electricity, but in fact, the computing power is different from the power. The use of computing power requires a threshold. First of all, you must have sufficient data to model mathematical models through data modeling, and then you can calculate the decision -making of artificial intelligence.

Wu Hexuan said that China has a large population, complete industrial categories, and complete systems. For many years, it has maintained the status of the world's largest manufacturing power, which will generate more data.

"However, although we have generated a lot of data, the proportion of really storing data is not high enough, and the proportion of use is lower." He Heye said, so we must first give full play to the ability of the data, that is, the perception of perception, that is, perception, that is, perception, that is, perception, that is, perception is perception, that is, perception, that is, perception, that is, perception is perception. Data and storage data should be encouraged to go to the cloud.

He further explained that to promote high -quality development now, it is necessary to promote the digital transformation of enterprises. Of course, for most small and medium -sized enterprises, it is not to buy the IT devices and software required for digital transformation by themselves. In fact, through the cloud service, it can be used by cloud services. "We must let more enterprises use cloud services so that the computing power is useful."

"Secondly, there must be some industrial foundation for the development of grassroots computing power." He Heye said, including data centers, which actually calculated and stored the computing power, including chips and artificial intelligence chips. This is also our shortcomings. Essence

In addition, it must be extended to the front end. If the data is not available, it is directly calculated. It also needs to make the front -end pre -processing, including cleaning, labeling, desensitization, etc. These tasks require a lot of manpower to complete. Of course, it is more important to apply the back -end, including mathematical modeling and modeling software and algorithms. These still need to meet the needs of different applications by developing.

"These are actually for IT systems. In fact, as data centers are not only IT systems, but also electricity and air conditioning refrigeration, so how to reduce the energy consumption of data centers is also a matter of research." The Western calculation is the use of the more sufficient energy sources in the west, the cheaper electricity bill, and the renewable energy source to better meet the needs of low -carbon.

Wu Heyi said that the industry chain of the computing power industry is quite long. There are many jobs to do it. The data center is only a prominent link in the computing power industry.

To reduce the use of computing power

The theme of this conference is "Guide to Guide the Future", but there are various needs of various industries and each user for computing power, and there are obvious differences. How can we make computing power easier, easier, and more convenient to use?

Regarding the question raised by the reporter of the Daily Economic News, Xi Heyi said: "The computing power must be calculated first, and the calculation must have a model (mathematical model). How does the mathematical model come from? No, it can even be said that the threshold is relatively high. "

"Do not allow computing power to build models only in a very small number of enterprises, hoping to reduce the threshold of computing power." How to reduce? Wu Hexuan said that some companies have now opened the server's resources and opened some model platforms of the algorithm.

Wu He has further elaborated that there are already some open source big data basic computing power computing platforms on the Internet. Domestic Baidu, Ali, Huawei and other large companies have also opened some platforms. "This means that as long as small and medium -sized enterprises have their wishes and needs, they can use these platforms to do data training."

"Of course, this still has a certain threshold." He Heye admitted that some companies in China are now worthy of encouragement. Some companies have developed some models for the application of various industries. For example, a company has made a mathematical model of simulation gear. If SMEs want to simulate a gear, they can go to the cloud. Use the mathematical model of this company to enter the relevant data used by this company to get the relevant data used by themselves. Smart decision results. "Everyone has recognized this trend, to make the computing power easier and reduce the threshold for use of computing power."

However, He Hexuan pointed out that some of the most precise models still need to be trained, which requires the state to deploy some key areas to develop algorithms. "With the full layout of computing power resources, the threshold for calculation will be reduced. I hope that one day in the future, everyone can use computing power more easily."

Daily Economic News

- END -