You can search "Rebirth of the University Hegemon" on Baidu to find the latest chapter!
¡°It would be too risky to build it from scratch¡¡±
Tu Jianfeng said.
"There are risks, but the technical team we recruited from Yingjiang is still very experienced, and the risks can be controlled within a certain range. After all, we are about to launch a supercomputer with a speed of 2 billion operations per second. Specifically designed to aid nuclear fusion research."
Yang Jie said with a smile.
""
Tu Jianfeng and others opened their mouths wide¡ª¡ª
Huaxing Group Company has specially built a set of supercomputers for the purpose of researching nuclear fusion technology, and its computing speed is now the most powerful supercomputer in the world. What a luxury!
The amount of data that a supercomputer with such high computing speed can process in one day is equivalent to more than 90 years of non-stop computing by 6 billion people around the world.
The computing performance of the supercomputer built by Aihua Technology Group Co., Ltd. was still 5,000 trillion times. More than four years later, Huaxing Group Co., Ltd. has increased the computing power of the supercomputer to 40 times!
Inspur Group Company, a subsidiary of the Chinese Academy of Sciences, also launched a supercomputer with the same performance half a year later. Now the two companies have similar technical levels in supercomputers.
It¡¯s no wonder that Yang Jie himself studied under Academician Jin Minzhe, and part of the R&D team came from the National Parallel Computer Research Center. Many of the technologies are from the same origin.
Currently, Inspur Group and Aihua Technology Group Co., Ltd. are the best in supercomputers in China, but Aihua Technology Group Co., Ltd. is better at commercialization and has greater influence.
Simulating nuclear fusion requires extremely high computing performance of supercomputers. Previously, domestic supercomputers could only simulate more than 10 billion particles at most when conducting nuclear fusion simulations. One simulation would take at least several months, and It costs hundreds of thousands of dollars in electricity bills every day, and one simulation costs tens of millions.
But now that Huaxing Group has increased the computing speed of supercomputers by 40 times, the time for a simulation can be shortened to a few hours, and the number of particles that can be simulated can be increased to hundreds of billions, which basically meets the requirements of "artificial intelligence". Sun" science and engineering requirements.
Inertial confinement nuclear fusion is to compress deuterium and tritium particles that can produce fusion energy to only one thousandth of the volume of a liquid of the same mass, confine them into a core device similar to the size of a peanut, and then keep it from flying around Only when it is heated to a high temperature of 100 million degrees within a nanosecond of turning on can the fusion energy required for the scientific project of "artificial sun" be generated. For such an extreme reaction, it is difficult for existing experimental detection methods to penetrate deep into the fusion fuel to measure it. What is even more difficult is that these high-temperature plasmas are very difficult to control. The plasma not only has short-term interactions, but also long-distance interactions. The behavior of the interaction is close to chaos, so it is very difficult to control.
Supercomputers are used to simulate and calculate the motion behavior of these particles to verify the theoretical models established by scientists, and at the same time verify them through these experimental devices.
In the past, the computing power of supercomputers was not enough, so it was impossible to simulate the movement of so many particles. In the 1980s and 1990s, the computing power of the supercomputers in Yingjiang was only petaflops and petaflops, which were used to simulate nuclear fusion. It is completely insufficient. The simulated theoretical model of nuclear fusion is inaccurate, and the model can only be built by obtaining various data through a large number of experiments.
By using supercomputer simulation, each process can be simulated step by step and the physical details can be studied. The stronger the computing power, the higher the accuracy of the simulation and the faster the calculation speed. This saves a lot of time and is conducive to the improvement of the experimental device. It can be said to be a super tool for designing, analyzing and understanding experimental results.
It is certain that controllable nuclear fusion can bring nearly unlimited clean energy. Scientists have been making steady progress in this area. However, in fact, the difficulty of completely controlling nuclear fusion still encounters several very big problems.
The Southwest Institute of Nuclear Industry Physics currently does not have a supercomputer. The two supercomputer centers built by Huaxing Group are mainly used for self-use. The National Supercomputer Center has many applications, and the Southwest Institute of Nuclear Industry Physics also has to queue up. It took several years to wait.
This time Huaxing Group Company has established a supercomputer center specifically for nuclear fusion, which will be of great help to domestic nuclear fusion research.
Those scientists at home and abroad who are developing "artificial suns" all have dreams of using supercomputers to simulate them, and they need higher-performance and easier-to-use supercomputers.
One of the most serious problems of devices like tokamak is how to minimize or completely control the interruption of plasma flow. This requires scientists to build a predictive model that can predict the plasma interruption within 30 milliseconds before it occurs. Interruptions are predicted with an accuracy of 95%, so that the control system can adjust in a very short time to avoid plasma interruptions.
This is extremely difficult. It can only be achieved by using supercomputers with more than one billion billion calculations to simulate a very precise software control system and high-precision and high-sensitivity probes for real-time adjustment. Furthermore, nuclear fusion is very complex, and this control system software must have Intelligence and extremely high computing speed can solve complex problems quickly and accurately. This can only rely on deep learning of artificial intelligence.
The core technology of stellarators and tokamak is the same. The advantage is that the controlled plasma has no plasma current, so there is no risk of large rupture, and the operation is more stable. However, in order to completely control these unpredictable horizontal events, The collision of plasma is also an extremely difficult thing, even more difficult than a tokamak, because this device requires very touching requirements from processing to operation. It is a configuration for high-end players, but the advantages are also obvious.
Like tokamak, everyone needs to establish a very accurate mathematical model of nuclear fusion plasma turbulence before they can have a more accurate prediction of the collision route of these plasmas, but the stellarator is more difficult. .
We can¡¯t say who is right or wrong between these two technical routes. It mainly depends on how everyone chooses. Now Yingjiang has given up on the technical route of star emulator, leaving only two high-end players, Neon Country and Deutschland.
The technical team recruited by Yang Jie is engaged in stellarator technology, so they can only plunge into the nuclear fusion technology of stellarator.
To this end, he specially built a set of special supercomputers for this purpose, and also used artificial intelligence technology in order to speed up the progress of research in this area. (Remember the website address: www.hlnovel.com