The power of the Russian supercomputer "Sergey Godunov" has grown 2 times

The domestic supercomputer is named in memory of the famous Soviet and Russian world-famous mathematics Sergei Konstantinovich Godunov. It was created on the basis of the high-density and energy-efficient platform "RSK Tornado" with liquid cooling.

2024

Increase peak performance to 114.67 Teraflops

Specialists of the RSK Group of Companies have completed planned modernization of the supercomputer "Sergey Godunov" at the S.L. Sobolev Institute of Mathematics of the Siberian Branch of the Russian Academy of Sciences (IM SB RAS). As a result, the total peak performance of this computing complex increased from 54.4 Teraflops to 114.67 Teraflops (trillions of operations per second) - of which 75.87 Tflops on central processors and 38.8 Tflops on graphics accelerators. Representatives of the RSK Group of Companies reported this to TAdviser on September 12, 2024.

source = RSK GC

According to them, the modernization project was carried out at the expense of a grant to update the instrument base of leading organizations within the framework of the federal project "Development of infrastructure for scientific research and personnel training" of the national project "Science and Universities."

Director of IM SB RAS Andrei Evgenievich Mironov believes that the supercomputer has significantly increased the effectiveness of scientific research and will contribute to the development of new technologies. As he clarified, it became possible to solve multidisciplinary problems, simulate volumetric processes and predict the behavior of complex mathematical systems. The supercomputer performs calculations on critical problems and tasks facing the Russian Federation, including:

• construction of baric axial algebras and Grice algebras,
• optimization of parameters of mathematical models of turbulent currents with coupled heat exchange based on deep learning technologies and nature-like algorithms;
• wave; medical tomography
• modeling epidemiological, environmental, economic and social processes;
• modeling and building scenarios for the development of the biosphere-economy-society system, taking into account carbon-free and sustainable development and climate change;
• solving inverse problems of geophysics using the direct method based on the Gelfand-Levitan-Crane approach.

source = RSK GC

Supercomputer "Sergey Godunov" is the main tool for research and applied development in the Academgorodok of Novosibirsk and the creation of a technological platform under the auspices of the Scientific Council of the Branch mathematical sciences of the Russian Academy of Sciences on mathematical modeling of the spread of epidemics taking into account social, economic and environmental processes.

The high-performance system IM SB RAS is built on the basis of computing nodes, each of which is equipped with two processors families. Intel Xeon Scalable In addition, the composition, supercomputer as of September 2024, includes a node based on 4 NVidia A100 GPUs server and large memory visualization Intel (Xeon Platinum 8368 processors, 4 TB of RAM, 2 5000 NVidia RTX Ada graphics cards, 32 GB of GDDR6).

The official opening of the supercomputer "Sergey Godunov" with a capacity of 54.4 Tflops

The Institute of Mathematics named after S.L. Sobolev of the Siberian Branch of the Russian Academy of Sciences (IM SB RAS) held the official opening of the supercomputer "Sergey Godunov," which received its name in memory of the famous Soviet and Russian mathematics with the world name Sergey Konstantinovich Godunov. This was announced on February 26, 2024 to TAdviser by representatives of the RSK group of companies. The total cluster performance as of the end of February 2024 is 54.4 Teraflops (trillions of operations per second).

source = IM SB RAS

Installation and testing of the supercomputer system was completed by specialists of the RSK Group of Companies at the end of November 2023. As noted by acting Director of IM SB RAS Andrei Evgenievich Mironov, the supercomputer will help significantly increase the efficiency of scientific research and will contribute to the development of new technologies.

According to him, Russian scientists have the opportunity to solve multidisciplinary problems, simulate volumetric processes and predict the behavior of complex mathematical systems. The supercomputer performs calculations on critical problems and tasks facing the Russian Federation, including:

• computational aerohydrodynamics and problems of optimizing turbulent currents with coupled heat exchange based on deep learning technologies and nature-like algorithms;
• medical electroacoustic tomography;
• modeling epidemiological, environmental, economic and social processes;
• modeling and building scenarios for the development of the biosphere-economy-society system, taking into account carbon-free and sustainable development and climate change;
• solving inverse problems of geophysics using the direct method based on the Gelfand-Levitan-Crane approach.

source = IM SB RAS

In general, the supercomputer "Sergey Godunov" is the main tool for conducting research and applied development in the Academgorodok of Novosibirsk and creating a technological platform under the auspices of the Scientific Council of the Department of Mathematical Sciences of the Russian Academy of Sciences on mathematical modeling of the spread of epidemics, taking into account social, economic and environmental processes, specified in the RSK Group of Companies.

In 2024, with the help of a grant aimed at updating the instrument base of leading organizations within the framework of the federal project "Development of infrastructure for scientific research and training of personnel" of the national project "Science and Universities," a planned modernization of the supercomputer will be carried out, as a result of which the peak performance of this computer system will more than double and reach 120.4 Tflops (trillions of operations per second).

2023

Supercomputer details

Specialists of the RSK Group of Companies on December 6, 2023 announced the completion of the installation and testing of the supercomputer system at the S.L. Sobolev Institute of Mathematics of the Siberian Branch of the Russian Academy of Sciences (IM SB RAS) in Novosibirsk.

According to representatives of the RSK, the domestic supercomputer is built on the basis of the high-density and energy-efficient platform "RSK Tornado" with liquid cooling. Each of the six computing nodes that underpin the high-performance IM SB RAS system is equipped with two 3rd generation Intel Xeon Scalable processors (38 cores and a base frequency of 2.4 GHz each). At the same time, one of the computing nodes is a GPU node, which is equipped with two graphics accelerators and has a total memory capacity of 80 GB, specified in the IM SB RAS. Hybrid cooling enables efficient use of computer resources.

"The solution to the problem supercomputer is as follows: the user sends the task to server management, where it is processed and divided into smaller tasks. These tasks are then distributed among the supercomputer processors that perform them in parallel. After completing the tasks, the results are collected and sent back to the user, "- shared representatives of the IM SB RAS.

source = IM SB RAS

The project was implemented as soon as possible - according to RSK representatives, installation and commissioning work was completed in 3.5 weeks.

Thanks to a grant received by IM SB RAS within the framework of the federal project "Development of infrastructure for scientific research and training of personnel" of the National Project "Science and Universities," the institute will be able to continue the modernization of its equipment and attract the best specialists for research in various fields of science.

"The design of this supercomputer provides an extension to 34 computation nodes (CPUs). You can install 28 more servers and increase system performance and solve even more complex problems. It is also possible to expand the air field: the air conditioner allows you to cool 9 kW of air power. In this room, we have provided a place for installing another computer rack, "said Egor Kozlov, director of the RSK Technologies project management department[1]

Launching a 54tn operations-per-second supercomputer to solve complex physics and biology problems

At the end of November 2023, a supercomputer was launched at the Institute of Mathematics of the SB RAS in Novosibirsk, which will help scientists solve complex problems in the field of mathematics, physics, biology and other sciences. The performance of the computer is 54.4 teraflops, or 54.4 trillion operations per second. At the same time, 19.4 teraflops will be involved in GPU-nodes, the press service of the university clarifies.

The computer consists of six computing nodes equipped with two processors with 38 cores and a frequency of 2.4 GHz, 512 GB of RAM and an SSD drive. By 2025, the supercomputer is planned to expand to 234.4 teraflops, so that the most powerful computer in the Novosibirsk Academgorodok will be obtained. There are plans to install another 12 computing nodes, and another GPU node will also be added. Thus, capacity by 2024 will increase by another 89 teraflops. The total power consumption of this machine is expected to be approximately 41 kW.

source = IM SB RAS

We can solve problems of incredible complexity, simulate volumetric processes and predict the behavior of complex mathematical systems. It can be useful for developing new technologies, data analysis and education. It will help us teach students and young scientists modern methods of research and work with data. The supercomputer also plays an important role in the development of international cooperation, as it will allow us to exchange data and research results with other scientific centers around the world, - said the acting director of the institute Andrei Mironov.

This computer system was built within the framework of the federal project "Development of infrastructure for scientific research and personnel training" of the national project "Science and Universities." At the same time, the volume of investments in the project by the end of November 2023 was not disclosed.^[2]

Notes