MTS Grom Supercomputer

Main article: Supercomputers

2021

Start MTS Grom

MTS PJSC on June 28, 2021 announced the launch of the MTS GROM supercomputer, whose performance reaches 2.26 petaflops.

MTS will use MTS GROM to develop the digital ecosystem. On its basis, the MTS Center for Artificial Intelligence plans to implement internal and external projects in the field of client service and increase business operational efficiency. The supercomputer will accelerate the creation and launch of products in the field of machine vision and industrial analytics, natural language processing, intelligent video surveillance. Universities, large scientific and medical centers will be able to use MTS GROM in research related to the modeling of complex processes.

The supercomputer plays a key role in teaching artificial intelligence algorithms. The computational power of MTS GROM allows you to create more complex models, test many times more hypotheses and train neural networks ten times faster than before. This directly affects the quality of products developed by the MTS Artificial Intelligence Center and the speed of their launch to the market.

In the future, high-performance MTS GROM capacities will become available to Russian companies in the# CloudMTS provider portfolio. This will allow large businesses to multiply reduce the time to develop and implement projects based on artificial intelligence and big data. Startups - hundreds of times lower the cost of high-performance computing for speech analysis, video processing and face recognition. For example, in Fintech, MTS GROM resources can be used to develop remote identification solutions or to assess credit risks.

The supercomputer is deployed by cloud provider# CloudMTS in its own MTS data center based on the comprehensive Nvidia DGX SuperPOD solution. This solution provides not only the speed of super-computing, but also scalable performance for the most demanding AI and HPC tasks.

Combining the capabilities of various areas of the MTS ecosystem to solve common technological problems allows us to implement large-scale infrastructure projects as soon as possible. So, in just a month, the cloud provider# CloudMTS built one of the powerful supercomputers in Russia for our tasks related to the development of products based on artificial intelligence, "said Alexander Khanin, head of the MTS Artificial Intelligence Center. - We are developing solutions not only to improve the efficiency of MTS Group's own business processes, we are also helping Russian business to follow the path of IT transformation. The speed of products launch and the quality of their work - parameters that are directly affected by the use of the supercomputer - are undoubtedly one of the key success factors in highly competitive digital services markets. Therefore, the launch of the MTS GROM supercomputer is an important step not only for the MTS Group, but also for the entire Russian IT industry, as well as for the scientific community, to which we will definitely provide access to MTS GROM resources to accelerate promising scientific developments.

About MTS Grom Supercomputer

As of June 2021:

• The supercomputer allows you to massively run many parallel calculations on all computing nodes (servers) at once. The solution is based on the Nvidia DGX A100 firmware platform with NVIDIA A100 graphics processors with a total memory of 320 GB
• Computing nodes are connected by a high-speed Infiniband network. This is a switched computer network that is used in high-performance computing and has high throughput and low latency.
• The MTS GROM computing cluster is equipped storage system NetApp with NVMe technology, which provides ultra-fast interaction of computing nodes with the system Data Storage to reduce the learning time AI of models.