GRID'2025 results: conference topics expanded, present and future of distributed computing The Joint Institute for Nuclear Research hosted the 11th international conference “Distributed Computing and Grid Technologies in Science and Education” (GRID’2025), organized by the Meshcheryakov Laboratory of Information Technologies. Almost 300 experts from various countries and scientific organizations worldwide discussed in Dubna the issues of the current status and development directions of information technologies, distributed computing, and grid technologies. At the opening, the conference participants were welcomed by JINR Director Grigory Trubnikov, JINR Scientific Leader Victor Matveev, MLIT Director Sergei Shmatov, and MLIT Scientific Leader Vladimir Korenkov. “I am very pleased that the number of participants is growing, as is the interest in this scientific field. We see that life sciences and computer sciences are now the most in demand from the point of view of university entrants, universities, and the needs of the economy. Our conference this year has brought together almost 300 people from different countries of the world. I am very glad that the giants of Russian computing and information technologies, those who create powerful computing and computer systems, algorithms, and programs for complex scientific and applied tasks, traditionally deliver talks on the first day,” Academician Grigory Trubnikov commented on the GRID’2025 opening, highlighting the large number of young specialists whose reports are included in the thematic sessions of the conference. “Computational sciences rightfully occupy the position of the queen of sciences. After all, not a single modern area of research can do without information technologies today,” JINR Scientific Leader, Academician Victor Matveev pointed out. “The GRID conference is the only specialized platform in Russia for discussing the entire scope of issues related to distributed computing and grid technologies. In 2025, our conference has brought together outstanding experts who are among the top ten most authoritative IT specialists in Russia to speak at the plenary session,” MLIT Scientific Leader Vladimir Korenkov emphasized. “It is highly important for us that the GRID conference has once again confirmed its status as an effective platform for enhancing international cooperation. Along with representatives of the JINR Member States, specialists from countries with which the Joint Institute has been establishing particularly intensive scientific cooperation in recent years came to MLIT,” MLIT Director Sergei Shmatov said. Grigory Trubnikov opened the GRID’2025 conference with an overview of JINR’s activities, namely, the Institute’s latest achievements, large-scale scientific projects, research areas, and basic facilities. JINR Director underlined the contribution of the Meshcheryakov Laboratory of Information Technologies to JINR’s activities and, in particular, serious efforts in training personnel and the work of MLIT specialists to create a digital environment at the Institute. The program continued with a talk by RAS Academician Arutyun Avetisyan, Director of the Ivannikov Institute for System Programming. Challenges that face modern society in the field of cybersecurity as a result of the widespread introduction of information technologies, including those using artificial intelligence, were outlined in the report. The speaker presented the experience and developments of the ISP RAS and other world specialized centers capable of providing directions to solve these issues, as well as the main trends in the development of IT technologies in Russia. In addition to the need to take into account socio-humanitarian aspects when implementing AI, Arutyun Avetisyan pointed out the importance of the problem of data reliability, emphasizing that even 1% of unreliable data is enough for an unsatisfactory result in the operation of AI-based systems. High-performance computing systems with a reconfigurable architecture were considered in the report of Academician Igor Kalyaev, Scientific Leader of the Direction of Southern Federal University. The basis of this approach is to provide the user with the ability to program the architecture of the supercomputer to the structure of the task it is solving. As the supercomputer’s computing device, it is proposed to employ not standard microprocessors, but a multitude of interconnected programmable logic integrated circuits (field-programmable gate arrays, FPGAs) of a high degree of integration. During the talk, Russian developments in this area were presented. RAS Corresponding Member Vladimir Voevodin, Director of the MSU Research Computing Center, identified supercomputer co-design as the central problem of modern supercomputer technologies and parallel computing. As the speaker noted, implying the joint development of the architecture and software of supercomputers, co-design can ensure high computing performance, however, fundamental research is required to reveal its potential. Vasily Velikhov, Assistant to the President of the Center for Information Technologies and Artificial Intelligence, NRC Kurchatov Institute, delivered a talk on the creation and development of a national computer network based on quantum communications. He spoke about the status of the implementation of the inter-university quantum network (IUQN) project and the first steps towards building a secure digital infrastructure of the new generation. Such a network will make high-performance distributed computing available to all universities. RAS Corresponding Member Ruslan Smelyansky, Head of the MSU CMC Department of Automated Systems for Computing Complexes, presented a vision of what the computing infrastructure of the future will be like. The speaker pointed out that the major driver of the computing technology development was applied needs. At the same time, speaking about requirements for the development of the distributed computing infrastructure, he especially highlighted the importance of deterministic communication and the provision of a predictable level of performance in networks. The report by Alexander Degtyarev, Professor of the SPbSU Department of Computer Modeling and Multiprocessor Systems, focused on the organization of resource-intensive computer modeling in real time, namely, on the creation of a virtual testbed or, in other words, a personal virtual supercomputer, which is a combination of a large number of various models for their subsequent expert selection. This will allow one to form in real time a computing complex capable of most effectively solving a specific task. The talk by MLIT Scientific Leader Vladimir Korenkov was devoted to the history of development and current state of distributed computing at JINR. Trends in the development of distributed computing for large-scale scientific projects were also considered. MLIT Deputy Scientific Leader Tatiana Strizh spoke about the path that the Meshcheryakov Laboratory of Information Technologies has taken over the past 60 years since its establishment. In her speech, she remarked that starting from the formation stage, the Laboratory was aimed at building a model for providing computing resources to users on demand. MLIT Deputy Director Dmitry Podgainy enlarged upon the tasks of the JINR scientific program solved using the “Govorun” supercomputer. He announced the completion of a new stage of the JINR supercomputer’s modernization, implemented jointly by MLIT and the RSC Group of Companies. As a result, the supercomputer performance for AI tasks has grown by more than a third. Using the example of the modernization of the “Govorun” supercomputer in 2024-2025, CEO and Co-founder of the RSC Group of Companies Alexander Moskovsky presented the hardware and software solutions of the RSC, a leading Russian developer and integrator of innovative supercomputer solutions. Sergey Plyusnin, Head of Architecture and Design at "E-Flops" LLC spoke about modern approaches to creating solutions for HPC, data centers, and clouds. In the following days of the GRID’2025 conference, a number of reports on a wide range of topics were made in plenary sessions. The conference participants discussed computing and software issues for the NICA megascience project in the BM@N, MPD, SPD experiments, as well as in the JUNO neutrino experiment. Konstantin Gertsenberger (VBLHEP JINR) talked about the distributed processing and storage of BM@N experiment data in the light of preparations for Run9. Arkadii Taranenko (VBLHEP JINR) presented the results of the full-scale modeling of the MPD experimental facility and information on the development of data analysis methods. SPD Computing and Software Coordinator Danila Oleynik (MLIT JINR) delivered a talk on the progress of implementing tasks on the development of the experiment’s software and computing. He noted considerable progress in elaborating a set of middleware responsible for the multi-stage, high-throughput processing of experimental data. Slavomir Hnatic (MLIT JINR) spoke about the development of software for the MPD experiment. Yulia Andreeva (CERN) presented the status and plans of the WLCG (Worldwide LHC Computing Grid) project. For over 20 years, the WLCG project has been successfully developing and continues to solve distributed computing tasks for the experiments at the Large Hadron Collider. At the same time, the upcoming High Luminosity LHC (HL-LHC) project represents an unprecedented challenge in terms of scale, complexity, and long-term sustainability. Xiaomei Zhang ( IHEP, China) gave a talk on the distributed computing system of the JUNO experiment. Xuantong Zhang (IHEP, China) spoke about the organization of IHEP distributed computing for international physics collaborations. The report by Andrey Tsaregorodtsev (CPPM-IN2P3-CNRS, France) was devoted to the status and development of the DIRAC platform. Victor Toporkov (MPEI, Russia) talked about the strategy of planning multidisciplinary work processes and resource management in cloud computing. The topic was continued in the report by Nikita Balashov (MLIT JINR) on the status of cloud infrastructures of JINR and its Member States. He highlighted that the cloud infrastructure served as a universal and flexible component of the JINR Multifunctional Information and Computing Complex. Over the past two years, operating systems and basic software have been updated within the cloud infrastructures of JINR and its Member States. Significant efforts were also directed at the implementation of new applications and services. Two methods of population annealing implemented in the MPI/CUDA hybrid architecture were presented in the report by Lev Shchur (Landau ITP and HSE University, Russia). Vladimir Sukhomlin (MSU, Russia) devoted his talk to modern standards and trends in training highly qualified professionals in the field of information technologies. A report on high-performance computing in radiation biology was made by Director of the JINR Laboratory of Radiation Biology Aleksandr Bugay. Modeling radiation-induced DNA damage using the Monte Carlo method, modeling DNA reparation and cell survival, and calculating the effects of radiation on tissues and organs were among LRB tasks outlined in this regard. To solve these and a number of other tasks, LRB colleagues employ the capacities of the “Govorun” supercomputer, including using deep machine learning methods. Computations related to radiation-induced brain disorders and tumor responses to irradiation are among the future goals of LRB theorists. Oleg Belov (VBLHEP JINR) introduced the conference participants to the research program of the ARIADNA collaboration, created to implement applied research at the NICA accelerator complex. Among the major areas of research at NICA beams, the speaker underlined radiation protection on Earth and in space, radiation effects in microelectronics, research into materials using ion beams, research into materials under extreme radiation doses, new technologies for accelerator-driven systems (ADS), research related to radiation therapy, radiation biophysics, and radiobiology. Alexander Bogdanov (SPbSU, Russia) spoke about the promising Data Mesh architecture, which involves changing the approach to data management. Pointing out that the constant evolution of data storage requirements leads to the fact that organizations are forced to utilize increasingly adaptive and large-scale systems, the speaker emphasized that Data Mesh was suitable for solving world-class tasks, while having the flexibility of cloud systems. Nadezhda Bagdasaryan (BMSTU, Russia) gave a talk on institutional transformation in the era of AI and challenges facing modern scientific schools. Hybrid artificial intelligence, as a combination of high-performance computing, neural networks, and precise mathematical models, was the topic of the report by Leonid Sokolinsky (SUSU (NRU), Russia). The ideas and approaches presented aroused great interest among the audience and a lively discussion. RAS Corresponding Member, IITP Director Maxim Fedorov delivered a talk on supercomputer modeling and machine learning in molecular sciences. He noted that the digitalization of chemical information applying AI and hybrid AI methods had enormous potential for the development of a wide variety of areas in science and innovation, however, this in turn entails the use of supercomputer technologies and Big Data, which acutely raises the issue of training specialists in these directions. Bioinformatics, as informatics based on biocomputer technologies, DNA information storage and nanobioelectronics, and its applications were considered in the report by Victor Lakhno (IMPB RAS, Россия). Fedor Ratnikov (HSE University, Russia) devoted his talk to the features of surrogate models for particle physics experiments. He concluded that surrogate models were a considerable help in tasks requiring large-scale modeling to interpret results, which certainly applies to high-energy physics. However, requirements for such models are quite specific: the sufficient information content and variability of generated objects, the prioritization of physical indicators, the quantitative evaluation of systematic uncertainties of physical results caused by the bias and variability of the model are needed. Andrey Shevel (PNPI, ITMO, Russia) spoke about large language models in high-energy physics and future developments in this area. Oleg Semenov (ITER Center, Russia) made a report on the creation and development of the Russian Scientific Network for Fusion Research project, aimed at building a unified scientific network for research in the field of thermonuclear fusion. The project currently unites 13 research centers in Russia. This is a unique information network that provides participants not only access to the results of all scientific experiments, but also the opportunity to participate remotely in experiments on scientific facilities. Viktor Kotliar (IHEP, Russia) talked about a comprehensive monitoring, automation and analysis system for the computing cluster of the Logunov Institute for High Energy Physics – NRC Kurchatov Institute. The report by Aleksey Artamonov (NRNU MEPhI, Russia) focused on the methodology of the intelligent analysis of scientific and technical information. A modern view on building a horizontally scalable solution for the tasks of storing large data volumes was presented in the talk by Stanislav Bogatyrev (YADRO technology company, Russia). In total, 37 plenary talks were delivered during the five days of GRID’2025. The conference’s thematic sessions also featured 127 reports in the following topics: Computing for megascience projects; Distributed computing systems, grid and cloud technologies, data storage systems, high-performance computing; Application software in high-performance computing (HTC and HPC); Methods and technologies for experimental data processing; Artificial intelligence methods in life sciences. Two round tables were organized within GRID’2025. The first one was devoted to the organization of the work of the joint scientific and educational laboratory of St. Petersburg State University and JINR. The laboratory is being created to solve tasks in the field of information technologies in high-energy physics and the training of qualified personnel. The participants of the second round table discussed the creation and development of information and analytical platforms. In particular, the development of the JINR Digital EcoSystem as a complex digital environment integrating a multitude of information services and business processes was discussed in detail. At the closing of the conference, JINR Vice-Director Latchesar Kostov delivered a speech, noting that the GRID’2025 reports and discussions were held at the highest level, giving rise to a number of new projects and joint investigations. “For the Institute’s Directorate, this conference is one of the central events in JINR’s scientific life, to which we will continue to provide comprehensive support. It attracts not only young scientists, but also the most outstanding scientists in this field from the Institute’s Member States and partner countries,” Latchesar Kostov underlined. Head of the SUSU Department of System Programming Leonid Sokolinsky addressed the organizers of GRID’2025 with words of gratitude. “This conference is different: a large number of participants and a wide range of topics. Everything is highly relevant and profound. On the one side there is a hard expansion of neural networks, on the other side there is a soft power of distributed computing. And it was this soft power that welcomed us here. Thank you for the warm atmosphere!” Scientists and specialists from 16 countries, including Armenia, Belarus, Bulgaria, China, Egypt, France, Georgia, Kazakhstan, Mexico, Romania, Russia, Rwanda, Slovakia, South Africa, Taiwan, and Uzbekistan, participated in GRID’2025. The conference’s social program traditionally included a boat trip on the Volga river and a barbecue in Ratmino, one of the most picturesque corners of Dubna. The GRID’2025 conference was dedicated to the founders of MLIT JINR, namely, the 115th anniversary of the birth of M.G. Meshcheryakov and the 95th anniversary of the birth of N.N. Govorun. The conference was held on the eve of the celebration of the 70th anniversary of the Joint Institute for Nuclear Research and the 60th anniversary of the Meshcheryakov Laboratory of Information Technologies, which will take place in 2026. The RSC Group of Companies, Platformix and ITCost were conference partners. The presentations and video recordings of the talks, as well as photos of the GRID’2025 conference, are available at grid2025.jinr.ru. Selected proceedings of the conference will be published in the "Physics of Elementary Particles and Atomic Nuclei" journal.
