Supercomputer Pereslavl-Zalessky

At the end of November, Pereslavl-Zalessky again became the center of the Russian supercomputer industry, where the National Supercomputer Forum (NSCF) has been held for the second year in a row [1]. The forum itself was held within the walls of the Institute of Software Systems named after A.K. Ailamazyan RAS, whose employees bore the brunt of organizational concerns.

NSCF was carried out with the support of the Department of Nanotechnology and Information Technologies of the Russian Academy of Sciences, one of the organizers was the National Supercomputer Technology Platform (NSTP).

If last year’s NSCF-2012 was a two-day “trial balloon”, which nevertheless brought together more than 240 participants from 124 organizations, then this second Forum was attended by more than 400 participants from 170 organizations, including 12 academicians and corresponding members of the Russian Academy of Sciences. In general, the representation turned out to be quite solid.

In addition to participating in the discussion of plenary and sectional reports, Forum participants could attend trainings from Intel, Rosatom State Corporation and SVT Research Center, as well as an IBM master class. Intel representatives spoke about current trends in the development of high-performance applications with an overview of their company's latest software tools. Rosatom conducted a review of the Zubr supercomputer (SC Zubr) of average performance installed at the RFNC-VNIITF. The SVT SVT training was devoted to the use of supercomputer technologies in the interests of modern computational physics. Finally, the IBM seminar was dedicated to the company's flagship supercomputer architecture - Blue Gene/O.

The sections and plenary sessions discussed issues of both software and technological support for supercomputer computing, as well as the results of specific studies and their applications.

At MSTU. N.E. Bauman is conducting research on modeling gas-dynamic processes in the designs of hypersonic aircraft, National Research Nuclear University MEPhI is working on the construction of iterative algorithms for cryptographic data processing, oriented to implementation using supercomputer technologies. In Novosibirsk, at the ICM&MG SB RAS, experience has been gained in using supercomputers equipped with Intel Xeon Phi accelerators to simulate the dynamics of astrophysical objects, and MIPT specialists are trying to solve large seismic problems on high-performance computing systems.

In addition to purely scientific tasks, supercomputer computing is gradually beginning to be used for real applied problems:

• Volgograd State University presented work with the results of computer modeling of the dynamics of flooding of territories in the event of an emergency using TESLA graphic accelerators;
• a group of research institutes reported on the use of a supercomputer and a special software package to solve the problem of increasing the efficiency of the muzzle brake;
• TESIS LLC spoke about numerical modeling in high-performance sports;
• CJSC MCST introduced everyone to the current state of affairs and prospects for the development of the domestic supercomputer platform Elbrus.

We have not forgotten about the creation of quantum and exascale computers, which today is an advanced direction in the development of super-performance computing. Presentations were made on these topics by representatives of the Joint Institute for High Temperatures of the Russian Academy of Sciences, IPU RAS and the Federal State Unitary Enterprise "Research Institute Kvant".

This, of course, is only a small part of the 110 reports made at the Forum.

This year, the organizers of the NSCF also paid attention to grid computing. For the first time at the Forum, the section “Grids from Workstations and Combined Grids” was organized. The Russian branch of the International Federation of Personal Computer Grid Systems (IDGF) [2] was involved in preparing the work and conducting the section.

The section discussed issues of increasing the efficiency of high-performance computing using grid technologies, the practice of organizing and conducting real corporate and voluntary distributed computing (DC) projects, and the first results of their implementation. The peculiarities of organizing voluntary RV projects and the development of the voluntary volunteer movement [3] were discussed in the speech of a representative of the “cruncher” community.

It was noted that an interesting trend in recent years has been the increased interaction between scientists organizing their projects and the volunteer community participating in them. The beginning of such cooperation was laid during the launch of the SAT@home project, which was already described on the pages of TrV-Nauka. This process is now being expanded and a new project is being tested jointly and is expected to receive the status of the Netmax@home voluntary distributed computing project.

Based on the results of the Second National Supercomputer Forum, it can be said that it is becoming one of the authoritative events in the field of development of domestic high-performance computing technologies and significantly contributes to the exchange of ideas and information for the development of Russian supercomputer industry.