lone012

  Hong Shihao: To catch up with advanced countries, it is necessary to have computing power or technology   Hong Shihao, Director of the Department of Information Engineering at National Taiwan University and Deputy Director of the National High-Speed ​​Network and Computing Center, has long been engaged in research in the fields of parallel processing, high-performance computing, and computing architecture. He used to work as a computer engineer at Sunrise in Silicon Valley, and is a well-known supercomputer expert in China. From his personal Facebook, we can see his focus on talent cultivation, industrial transformation, industry-university cooperation, etc. Every post is powerful and loud. "If you want to develop your own technology, you need two things, one is high-performance computing, and the other is to understand the software on the machine." Hong Shihao said that foreign countries have accumulated decades of experience. To enter an advanced country, it is bound to have computing power or technology. Almost all advanced countries in the world have supercomputers. Even Switzerland, which is a small country, has the world's top ten supercomputers for various scientific and engineering research and development. South Korea even enacted the Supercomputer Act ten years ago, expressly stipulating that the country should provide high-performance computing resources to support industry-university research and development, and even set up a budget to encourage the industry to develop supercomputer-related technologies and equipment. Hong Shihao pointed out that these countries all know the importance of high-performance computing. For example, simulation software is a classic application of supercomputers, including missile simulation and nuclear bomb design simulation. It is not necessary to detonate a nuclear bomb, and the nuclear explosion can be known through simulation software. after the situation. Especially in the era of AI, high-performance computing capabilities are becoming more and more important. The reason is simple, "Whoever has a strong computing power is more likely to win." Some people may think that if we don’t make the world’s most advanced products and only make ordinary consumer products, then there is no need to develop powerful AI technology? Hong Shihao believes that artificial intelligence is the crystallization of wisdom after a lot of training and data. These training results can be placed in electronic devices, including mobile phones, watches, or sockets around many street lamps, and then applied in many places, and the Internet of Things is one of them. However, the seemingly popular smart Internet of Things is actually difficult to make an impressive product. "The key is not the hardware, but whether the intelligence on it is really necessary?" Hong Shihao explained that many companies can make IoT devices. Today's industry can almost quickly produce many smart devices, but he has never thought about using them. How to use these devices and cooperation modes. In addition to the need for overall planning, there are also many software issues. Only a single networked device cannot be called a smart Internet of Things. Taking smart cities as an example, Hong Shihao pointed out that the most important thing is "data", including the processing and management of data. If the data is only placed in the data center, it will be meaningless, but the use of these data will require a lot of calculations Capability, that is, a very strong center, "an overall computing resource planning".  

Forum Role: Participant

Topics Started: 0

Replies Created: 0