Russian President Vladimir Putin says multiple new supercomputers may soon be installed across the country, as development lags behind the U.S. and China.
Speaking during a visit to the Cherepovets College of Chemical Technology last Tuesday, Putin suggested that plans were already in motion to expand a high-tech network of supercomputer and data centers that would no longer be limited to the capital city of Moscow.
AI Tool Created to Study the Universe, Unlock the Mysteries of Dark Energy
Russia has three of the top 500 supercomputer systems in the world, compared to China (228) and the United States (117), analysis suggests.
At last count, two of the country’s top supercomputersLomonosov-2 and Lomonosovwere housed at Moscow State University. The third is linked to a computing center of the Russian Federal Service for Hydrometeorology and Environmental Monitoring, or Roshydromet, also located in the capital.
In early January last year, a new AI research supercomputer dubbed “Zhores” was unveiled at the Skolkovo Institute of Science and Technology, but that is also located in Moscow. In the overall top 500 list, Russia’s top supercomputer is the Christofari, sitting at number 29.
The proposal to expand the infrastructure was made by AndreyYurchenko, who is a director of the Institute of Computational Technologies of the Siberian Branch of the Russian Academy of Sciences (RAS) in Novosibirsk. A transcript was published by the Kremlin.
Yurchenko suggested a “two-level” system, and noted Siberia provides 15 percent of high-quality scientific publications in Russia despite having less than two percent of overall resources.
He said: “[We should] develop a network of powerful supercomputing centers so that they are located not only in Moscow but also in St. Petersburg, Novosibirsk, Kazan, Vladivostok.
“Of course, not only super-powerful centers are needed, but also the second level, as we call them. They should also be located in the regions Tomsk, Tyumen, Khabarovsk, Yekaterinburg. Such centers can become a base… for the creation of networks of ‘data factories.'”
In response, Putin said that he agreed with a multi-tier system but said regulators still need to evaluate the plans. He said: “It is necessary to increase capacity and to create communication systems between these centers. And to evenly distribute these data factories.
“All this is in our plans. But it is necessary that the regulatory authorities monitor this very carefully and that appropriate monitoring of implementation be established. The whole system should be evenly distributed over territories…where there is intellectual potential.” Shortly after, the president added: “We will definitely pay attention to this, we will strive for this.”
The comments were translated into English using Google Translate. They were referred to in a media release from the Siberian Branch of the Russian Academy of Sciences.
As noted by DefenseOne in March last year, the Russian supercomputers, including Zhores, use technology that is imported from western companies, including Nvidia. Zhores was notable as it is an academic project not designed for military use, unlike the NDMC.
“Zhores is the first petascale energy-efficient supercomputer specially earmarked for solving machine learning problems and simulations based on data,” said Skoltech professor Sergei Rykovanov in a media release last year celebrating the installation.
“This computing system will help Skoltech researchers and its academic individual partners to make breakthroughs in a whole range of fields: digital medicine, Internet of Things, precision farming, image processing, processing new medications, smart cities, predictive technical maintenance, photonics, and the search for sources of X-rays and Gamma rays.”
Unlike a simple laptop or desktop PC, supercomputers are able to handle vast amounts of data, which is measured in floating-point operations per second (Flops).
Last November, Top500, which maintains a list of the world’s most powerful commercially available computer systems, confirmed that China and the U.S. dominated the industry.
“Summit and Sierra remain in the top two spots. Both are IBM-built supercomputers employing Power9 CPUs and NVIDIA Tesla V100 GPUs,” the most recent list said.
“Number three is the SunwayTaihuLight supercomputer, with an HPL mark of 93.0 petaflops. TaihuLight was [made] by China’s National Research Center of Parallel Computer Engineering & Technology and is installed at the National Supercomputing Center in Wuxi,” it added.
For now, a Russian presence at the top end of the list is largely missing.