The first world-class supercomputer center in Russia was opened at TSU

T-Platforms, the Program Systems Institute of the Russian Academy of Sciences, Intel and Microsoft announced the completion of the construction of the Regional Center for Sharing High-Performance Computing Resources of Tomsk State University. TSU, the oldest university in Siberia and the Far East, became the first Russian university equipped with a world-class supercomputer center: today it is one of the 15 most modern and productive computing centers in the world education system (according to the Top500 rating).

The TSU Computing Center is equipped with the SKIF Cyberia supercomputer based on 566 dual-core Intel Xeon 5150 series processors, which has become the most powerful computing complex in Russia, the CIS and Eastern Europe, as well as one of the hundred most powerful computers in the world. The peak performance of SKIF Cyberia was 12 trillion operations per second; system result on standard Linpack test was 8.945 trillion operations (74.5% of peak).

The basis of the supercomputer complex “SKIF Cyberia” is a computing cluster based on 282 computing and 1 control node in a 1U design based on 566 dual-core processors
Intel Xeon 5150. Thanks to Intel technologies, the power of the new SKIF is compared to
SKIF K-1000 has grown 5 times, while the number of used computing nodes and power consumption of the cluster remained at the same level. In addition to the standard set of control and system software, SKIF Cyberia uses the new operating system Microsoft Windows Compute Cluster Server 2021.

The complex solution of “T-Platforms” for TSU became the first commercial project in Russia in the field of high-performance computing, which included not only the supply of a computing system, but also the organization of the entire infrastructure of the supercomputer center on a turnkey basis.

The SKIF Cyberia uninterruptible power supply system with a maximum power of 160 kW provides voltage with specified characteristics to all components of the computing system, as well as protection against power failures. It is able to ensure the operation of a supercomputer for 10 minutes after a power outage. Modular system design allows for internal redundancy and hot-swappable components during operation.

As for the cooling system, the use of traditional air cooling turned out to be impossible due to the small area of ​​the TSU computing center. To ensure reliable operation of the equipment in such conditions, T-Platforms engineers designed a mixed air-water cooling system with N + 1 redundancy at all levels. Air conditioning units located in a row of enclosures provide hot air intake directly from the compute nodes. The hot air is then cooled by water in heat exchangers and supplied cold from the front of the computing racks. The heated water is cooled by freon in refrigeration units, the condensers of which are located on the roof of the computing center.

Comprehensive environmental monitoring of the atmosphere and hydrosphere, control over river flooding, the spread of fires and epidemics, rational use of forest and mineral resources, new competitive methods of oil and gas exploration, restoration of contaminated soils, design of rocket and space technology and safe mine equipment, creation of new types of rocket fuel and superhard coatings using nanotechnology – these are just a small part of the most difficult tasks that TSU scientists will solve with the help of SKIF Cyberia in close cooperation with enterprises of the region.

By admin

Leave a Reply

Your email address will not be published. Required fields are marked *