Where did China stand in the race for the best supercomputers?

Not too long ago, they still dominated top500.org, but since then, nothing has been heard from the Eastern superpower.

A top500.org it still updates its ranking of the world’s supercomputers twice a year, most recently in the past few days, and it shows that the USA is practically in the lead. According to the official list, only the United States has an exascale supercomputer capable of at least 1 EFLOPS, while China is only good for the 15th position with a system called Sunway TaihuLight.

At the same time, there was a period when the first two places belonged to China. In the middle of 2016, for example, the USA was only a bronze medalist, and a year later it did not even reach the podium. Compared to this, Europe is still showing itself, but China is nowhere. So the question may arise: where did they go?

In fact, China is still building supercomputers, but they are not participating in the top500.org initiative. According to various reports, the Chinese big irons called Tianhe-3 and OceanLight are exascale systems, but it is not very possible to prove them, so they do not go beyond the rumor category. However, the supercomputers in question certainly exist, and their performance can be quite powerful.

However, the question is how strong it is. Currently, there is a consensus within the industry that the Tianhe-3 can be the fastest, and its theoretical computing power can be around 2 EFLOPS. With this, the system would probably not be good for the first place, but it could perhaps take one of the two lower steps of the podium. The real problem is not the speed, but the consumption, because there is also a rumor that the most powerful Chinese machine requires significantly more energy than the top systems in the USA, so in terms of energy efficiency, it is particularly bad. Here, of course, it is particularly bad luck for China that the USA has really started to regulate this factor, since an exascale system is not small in size, so it would not be very good if those involved were to reach the point where the sheer amount of computing capacity would require power plants designed for the given machine. So there needs to be a reasonable limit from this point of view, which may be a critical factor in the West, but not so much in China, so the great power in the East is much more inclined to solve problems with force and money, and reason is pushed into the background. This makes it particularly difficult for the Western world to assess where China is in terms of computing capacity, because their individual decisions do not really reflect the logic and principles that decision-makers in the West profess. And China does not want to make it easy to assess the capabilities of their systems, so they prefer to keep them secret.