Computers of the newest generation are in demand around the world.
Chelyabinsk is one of the leaders in the level of information technology development in Russia. The leader in implementing supercomputer technologies in the Urals and Siberian region is South Ural State University. A Center for Supercomputer Modeling, which is used for solving educational, scientific and industrial problems of not only our country but also international customers, has been created.
Supercomputers for science, industry and cinematography
By the end of 2016, within the frameworks of a project for forming research laboratories of the world’s level, the Laboratory for Problem-Oriented Cloud Computing Environments was established on the ground of South Ural State University’s School of Electrical Engineering and Computer Science (EECS) under directorate of Professor Andrey Chernykh, Director of the Laboratory of Parallel Computations of the Center for Scientific Research and Higher Education of Ensenada (Mexico).
“I met Andrey Nikolayevich at a scientific conference of the Russian Academy of Sciences dedicated to issues of cloud computations organizations and high-performance computing systems. At the beginning of the 1990s he was invited to the leading science and research center of Mexico where in 20 years he was able to organize a Center of high-performance computing which is well-known around the world by its research in the sphere of planning of distributed computational problems. After having met Andrey Nikolayevich, the idea of such joint laboratory had come up, says Director of the EECS Gleb Radchenko. Professor Chernykh already came visiting us in the last year’s December. He held a series of lectures for SUSU students and teaching staff, and now one of our postgraduates is undertaking an internship in his Center in Mexico. The main goal of the internship is conducting research works on the topic of planning and supervisory control of resources of cloud computational environments.”
Subject-matter of the Laboratory supervised by Professor Chernykh is oriented at solving problems connected with supercomputer calculations. Supercomputer is a very powerful, high-performance computational system which contains a series of unique developments of the world’s level, and which leaves behind the majority of computers existing in the world by its technical parameters and computation speed. There are less than a thousand of such systems around the globe.
At the SUSU Supercomputer Simulation Laboratory there are three supercomputers whose performance is at the level of the world’s leading computational systems. Two SUSU supercomputers (“Tornado SUSU” and “SKIF-Aurora SUSU”) are the most high-performance systems of the Urals and Siberia, which are placed the 8th and 15th among the most powerful supercomputers of Russia and the CIS countries.
“End user of such high-performance systems is required to have specialized knowledge in the field of high-performance computations such as: supercomputer architecture, experience of working in Unix-like operation systems, remote configuration, and ability to work with task queues. Within the joint laboratory with Professor Chernykh, we create solutions which provide a transparent provision of supercomputer resources to engineers, researchers, industrial customers by offering them cloud web-interface oriented at solving their specific classes of problems,” adds Gleb Radchenko.
An important area of supercomputer systems implementation at SUSU is big data analysis. This area formed back in 2000s, and it unites the totality of methods for processing unstructured data of large volumes and variety in order to obtain profitable results. Rapid development of this area associate with supercomputers coming to existence, as using them helps dealing with large volumes of data, analysis of which is impossible to perform by “classic” methods using personal computers.
Within the orientation at intelligent data analysis, SUSU chose solving problems for production sector and industry. The Cloud Industrial Platform project researches possibilities for creating a system of “digital twins” of manufacturing processes. Digital twin is a virtual image of a device or technology cycle in a form of computer physical or mathematical models. Digital twin in the real-time mode gets synchronized with manufacturing processes by data processing, the data obtained from intelligent wireless sensors. Such sensors allow monitoring characteristics of technological processes (such as equipment’s power consumption, velocity of liquids’ or gases’ flow in tubes, structural components of such flows, vibration in crucially important cross-points, etc.). Each second, hundreds of sensors transmit information to a supercomputer, where analysis of incoming data for assessing the efficiency of equipment’s operation and estimation of potential risks is performed constantly.
“We are developing a platform which will provide collection, interpretation and analysis of all this data, and provide the end user, for example, a chief engineer or owner of an enterprise, with the information that is useful precisely for him. For example, having analyzed operational efficiency of a pump system, supercomputer can provide recommendation about date and time to stop it for scheduled technical maintenance so that the cost of its stopping was minimal, explains Director of the SUSU School of Electrical Engineering and Computer Science. If a pump does not receive the scheduled technical maintenance it will operate less effectively and an enterprise will start suffering losses. From the other hand, if you switch off the equipment earlier, then it won’t work out its resources, and the company once again will suffer unnecessary losses.”
At the present time, supercomputer resources of SUSU are used by engineers and researchers in mining and processing industry; they are needed to engineering companies and scientific centers which work on modeling of physical processes, chemical phenomena and nano particles; to industrial enterprises which produce high-technology production; to aircraft industry and rocket-building company groups.
Another area where supercomputer power is used is visualization of 3D graphics. Modern movies and cartoons 90% consist of computer graphics; and before releasing a movie it takes to visualize it first. For visualization, extra-big computational resources are required. In our country, such resources are provided in particular by the SUSU Supercomputer Center. SUSU supercomputers were used for visualization of special effects for such movies as “He’s a Dragon”, “Ledokol” (Icebreaker), “Ekipazh” (Flight crew), the cartoon “Savva: Heart of the Warrior”, and many other.
“Such commercial projects can be realized, still the main intended purpose of supercomputers is scientific research and completing difficult manufacturing tasks,” adds Gleb Radchenko.
Three most powerful supercomputers of the Ural Federal district are located in SUSU. The first one, “SKIF Ural”, appeared at the university in 2008 and became the first SUSU supercomputer included in the ranking of top-500 the most high-performance supercomputers of the world (TOP-500). The “SUSU SKIF-Aurora” supercomputer was a joint development of the Program Systems Institute of the RAS, Russian company PSK SKIF, and Italian company Eurotech. In June 2011 it got included in the TOP-100 of the world’s most powerful supercomputers. The third supercomputer – “Tornado SUSU”, installed in 2013, became the first in Europe among universities’ computational complexes equipped with the newest Intel® Xeon Phi™ coprocessors, and one of the first systems switching over to liquid cooling. According to the Green500 world ranking, this supercomputer platform is one of the most power efficient platforms in Russia and the CIS countries.
Everybody to the cloud
Construction of one’s own data processing center at an enterprise is a very cost-based and responsible project which requires serious investment and high competencies. Supercomputer is a whole infrastructure for which you need a room, spaces for ventilation and cooling, a power system, system for data and power energy backup, a whole team of administrators and programmers. Not every company is ready to undertake such expenses, especially if there is a question about efficiency from using such resources. If average computational load of such machine is less than 50%, economically this project will be a failure.
“At the university we have our own power stations which provide power supply to the university’s building, including all supercomputer resources. Also, within the SUSU Center for Supercomputer Modeling, we have our own system of uninterruptible power supply which provides correct operation of all supercomputers in case of power black-out, explains Gleb Radchenko. In order not to purchase expensive equipment and not to elaborate the complicated infrastructure, enterprises address centers of data processing, such as our supercomputer center. We constantly have 80-90% of nods loaded: when one task is completed, there comes another one. This eliminates stoppage of computational systems and allows reducing the solution cost for a commercial consumer. We can use same analysis algorithms for different consumers, and this way enterprises can save sufficient money on routine operations, and focus on their major activity. Supercomputer can complete tasks not only for enterprises of the Urals; it can be used in any part of the globe. We can provide computational possibilities to our clients all over the world, and many colleagues from overseas universities, including Mexico, use our supercomputer for completing their tasks.”
Union of Mexico and Russia
Nowadays there are more than 150 scientific tasks from various fields are being solved using SUSU supercomputers: mechanical engineering, metallurgy and metal processing, fuel and energy complex, light industry, pharmacy, fundamental research in the field of physics, chemistry, computer sciences. It is planned that under collaboration with Professor Chernykh these areas will develop and supplemented with new areas of research.
“At the beginning of the 21st century we realized how important is international collaboration in the field of scientific research, continues Gleb Radchenko. For example, in Russia there are no more than 10-15 research teams who work directly over my mathematics. And when I publish a research article on my scientific area in Russian language the volume of target audience amounts to 100-150 people at best. But if I publish the same article in English in a high-ranking international journal, the number of researchers, to whom I can introduce results of my work, increases several times. All scientists are interested in such cooperation, it is for a reason that international collaborations are developing in the world, and scientists from different countries apply for joint grants. For example, the SUSU EECS takes an active part in PWs@PhD international project within the European grant Erasmus+ in software engineering, which includes the leading universities of Russia, Finland, Denmark, Great Britain and Jordan.”
From a student to a world-renowned scientist
SUSU students, starting from their third year of study, get involved into working in the Center of Supercomputer Modeling and Laboratory for Problem-Oriented Cloud Computing Environments. Those students who show their interest in this area together with their scientific supervisors prepare their own publications in Russian and English languages.
“Our students and postgraduates, who are working in the supercomputer Center and research laboratories, carry out research and development of software systems in such areas as data bases, cloud and high-performance computations, intelligent data analysis, neural networks, artificial intelligence, and many other, adds Gleb Radchenko. Students participate in all-Russian and international scientific conferences being held in various cities from Guadalajara to Vladivostok; they compete at championships in sport programming.”
Moreover, students have the opportunity to prepare projects and receive grants for realizing their own research projects. For example, in 2017 six students and postgraduates of the School of Electrical Engineering and Computer Science became winners of the all-Russian UMNIK contest and received financial support in the amount of 500 000 rubles for two years for development of their research projects. And if a student not only studies well but also publish his research articles in scientific journals of high level, he has the opportunity to compete for increased scholarship – up to 25 000 rubles per month. Therefore, they can not to waste their time searching for a part-time job but contribute to science.
For a student, research activity during his study is a guarantee for a successful career growth as the majority of the School of Electrical Engineering and Computer Science’ graduates get recruited in the leading companies of not only our country but the whole world right after their graduation.