- Статьи
- Economy
- Computing schedule: unified rules for the work of supercomputer centers will be created in the Russian Federation
Computing schedule: unified rules for the work of supercomputer centers will be created in the Russian Federation
In Russia, common standards for the functioning of supercomputing centers should be created and formalized, since in the near future their importance will only increase, and the number of such sites will grow. This was announced by Prime Minister Mikhail Mishustin during a meeting with the Deputy Prime Ministers. Izvestia analyzed which key tasks should be solved by uniform operating rules, as well as which branches and areas of science and economics will benefit most from the expansion of the network of supercomputing centers.
Replacing experiments
Supercomputers are very powerful computing systems designed to solve complex problems that require a huge amount of calculations. Unlike conventional computers, they consist of multiple processors running simultaneously, allowing them to perform billions and trillions of operations per second. Their productivity is measured in petaflops and exaflops, and the work is based on distributing one large task into many small parts that are processed in parallel.
Such systems are used where accuracy and speed of data processing are important: in scientific research, industry, energy, medicine, and the development of artificial intelligence. They are used to model the climate, create new medicines, design complex machines, and train neural networks.
Their operation requires a special infrastructure. It must include powerful cooling and a large power supply, so supercomputers are usually located in specialized centers.
As Felix Gadzaov, head of the RANEPA Digital School of Public Administration, noted in an interview with Izvestia, supercomputing capacities, along with grid technologies, are gradually becoming the basis of infrastructure that ensures the development of science and a data-driven economy. Their use makes it possible to replace expensive physical experiments with computational modeling and significantly accelerate the training of large artificial intelligence models.
— According to the November version of the TOP500 rating, the combined performance of the systems included in it reached almost 15 exaflops, that is, about 15×101⁸ operations per second. There are also Russian developments on this list, in particular Yandex's Chervonenkis, which demonstrated an indicator of 21.53 petaflops and took the 83rd position," he said.
The introduction of uniform rules presupposes the formation of a transparent mechanism for allocating computing resources, taking into account the priorities of tasks, as well as the establishment of unified requirements for information protection, data management modes and software compatibility, Felix Gadzaov emphasized. In addition, work is underway to standardize monitoring systems, capacity accounting, and lifecycle management of digital services.
In world practice, such approaches are implemented through formalized procedures for allocating computing time and peer review of applications, as is done, for example, in pan-European supercomputer infrastructure programs.
— A significant part of the regulation will inevitably be related to energy supply and project approval issues: according to the Soben study "Trends in Data Centers – 2026", in the United States, data center projects worth $18 billion were stopped during the year, and another $46 billion was postponed due to public pressure. - the expert clarified.
According to him, the expansion of the network of supercomputer centers can give the most noticeable result in areas where high calculation accuracy and data processing speed are critical. This applies to materials science and chemistry, pharmaceutical development, climate analysis, engineering design of complex systems, geological exploration and energy.
The use of computational models reduces research time and improves the quality of technical solutions, he noted. For example, Moscow State University reported on conducting seismic calculations on the MSU-270 system for a typical model of a hydrocarbon field, and in 2024 the United Engine Corporation commissioned a new supercomputer designed to develop digital counterparts of aircraft engines.
Clear order
There is a need in Russia to form unified approaches to the functioning of supercomputing centers, Marina Kholod, PhD in Economics and a leading researcher at the Laboratory of Artificial Intelligence, Neurotechnology and Business Analytics at Plekhanov Russian University of Economics, told Izvestia. Such rules should be based on standardized connectivity and well-established interaction between computing platforms, as well as a clear procedure for providing access to their resources.
—Regulation should consolidate common technical requirements and principles of data exchange, which will link disparate clusters into an integrated national system," she said. — Thanks to this, specialists and companies from different regions will be able to use facilities located in other parts of the country, regardless of geography.
According to her, special attention should be paid to unified approaches to infrastructure equipment, as well as information security and cybersecurity issues.
The development of a network of supercomputer centers and the streamlining of their work will have a significant impact on a wide range of high-tech areas, she noted. The most noticeable effect is expected in the field of artificial intelligence and software development, where the need for computing resources is particularly high.
"Access to capacity will accelerate the creation of domestic solutions and the development of machine learning technologies," the expert said. — The scientific environment will also benefit significantly, including research in physics, chemistry, biology, genetics and pharmaceuticals, where complex calculations and simulations are needed. This is especially important when developing new medicinal compounds, analyzing aerodynamics, and predicting climate processes.
Industrial sectors are also actively exploiting the potential of high-performance computing, including mechanical engineering, the aviation and automotive industries, as well as the energy and oil and gas sectors.
The proliferation of digital twins allows you to evaluate product characteristics in advance, simulate various operating scenarios, and optimize designs before the production stage, which reduces costs. Financial organizations use such resources to assess risks, automate trading, and identify suspicious transactions, which makes supercomputing power an important tool in this area.
Izvestia sent a request to the Ministry of Education and Science, but no response had been received at the time of publication.
Переведено сервисом «Яндекс Переводчик»
