Interview: T-Platforms CEO Sets Sights on Growing HPC Market

Print Friendly, PDF & Email

Feature Interview with Vsevelod Opanasenko, CEO of T-Platforms

Founded in 2002, T-Platforms has been steadily building their multi-national company into a powerful HPC leadership organization. They made a big splash last year with its expansion into Western Europe, and at ISC10, they demonstrated their passion and energy on the exhibit hall floor with an amazing hospitality event on the CCH Roof Garden. We can’t wait to see what they have planned for ISC11.

insideHPC is pleased to present this exclusive feature interview with T-Platforms CEO, Vsevelod Opanasenko.

insideHPC: At 510 Teraflops, the Lomonosov supercomputer at Moscow State University has reaffirmed its status as number one in the Top50 list of most powerful supercomputers in Russia and Eastern Europe. What kinds of breakthrough science has this compute capability enabled in the region?

Opanasenko: The Lomonosov supercomputer really is at the head of the Top50 list with its declared peak capacity of 510 TeraFLOPS. However, the specialists at our company have recently carried out extensive modernization on this computer complex, and today its peak performance reaches 1.3 Petaflops, a record-breaking performance for Eastern Europe. This will allow University scientists to carry out numerous research projects in different branches of science at a new qualitative level. The main projects will include energy and energy-saving, aviation, medicine, biology and bioengineering, genetics, molecular physics, nanotechnologies and many other areas. And in the future, this supercomputer will serve as a basis for tasks such as designing structural units for the latest model planes, cars and nuclear reactors.

insideHPC: There was recently a rumor / expectation that T-Platforms would announce a significant collaboration focused on exascale technology development that was bigger than Russia-centric. Is there any validity to this rumor?

Opanasenko: At the end of last year we signed a memorandum of intention in the area of joint research and development with Lomonosov Moscow State University. This was to create a technological basis for a new generation of supercomputer with a capacity of 1 exaFLOPS. However, we still have to overcome a great number of technological barriers on the way to creating such a powerful system, and such cooperation with Russia’s leading scientific center will be of invaluable help to us. The rich experience of University specialists in the development of algorithms for parallel computing processes, combined with the unique expertise of our engineers in creating the infrastructure of HPC systems and environments will enable us to make considerable progress in this work.  Additionally, we support the HOPSA * project aimed at the enhancement of supercomputer efficiency, and also actively cooperate with the leading European initiatives – PRACE, STRATOS, EOFS – focused on the development of next-generation exascale supercomputer technologies.

insideHPC: In 2010, I believe the TB2-TL was the first supercomputer / blade to deliver an astounding 105TFLOPS per rack. Can you clarify this for us – is this accurate, and what has happened since then in terms of performance upgrades?

Opanasenko: The TB2-TL system, built on the basis of hybrid architecture using NVIDIA graphic accelerators, established a computing density record last year. It served as the basis for us for developing a supercomputer with a capacity of over 1 PFLOPS today. Moreover, we regard TB2-TL as the technological foundations for developing a new generation computer complex whose capacity will exceed 1 exaFLOPS (1 quadrillion operations per second). Its creation will allow us to advance to a new qualitative level in seismic data processing, the modeling of processes taking place in the earth’s atmosphere, developing new kinds of medicines, and also solving many other problems that contemporary science faces today.

insideHPC: Along with Lomonosov Moscow State University, you sponsored a contest for “Efficient Use of GPU-Accelerators to Solve Large Problems.” What prompted you to launch that contest? Are you in the Finals stage now? What will be the prizes and when will they be announced?

Opanasenko: We organized this contest project, we wanted to show the real potential of present-day graphic accelerators as part of hybrid computer complexes that solve complicated scientific problems. As a part of this event, and with the support of Lomonosov MSU, a number of Russian scientific organizations began long-term research, the results of which may have a considerable influence on the development of the national economy. Participation in the contest was open to any Russian scientific organization that uses software with the support of graphic processors in its work. The contest winners will be awarded as part of the fourth supercomputer conference that will take place at the end of November, 2011. The companies that present the most interesting projects will gain exclusive and encouraging quotas for access to the supercomputer resources of MSU’s Scientific and Research Computer Center. In addition, these projects will be recommended for publication in the third issue of the book, “Supercomputer Technologies In Science, Education And Industry”.

insideHPC: Your business push into Western Europe generated quite the buzz at ISC’10 last year. I think it’s fair to say that T-Platforms and Intel were the two ‘most talked about’ companies at ISC last year. What can we expect from T-Platforms at ISC11?

Opanasenko: The ISC 2011 exhibition is surely one of the most important events in the supercomputer industry, and we pay special attention to our participation in it. This year we are going to present a set of the latest generation T-Blade2 systems. Thanks to their high performance and computing density, they will serve as a basis for new-generation computer complexes whose capacities will be measured on a petascale. Also, exhibition visitors will be able to see the medium capacity T-Blade 1.1 systems that are currently used successfully by many Russian companies’ computer centers. Furthermore, we are going to announce new company products at the exhibition: the T- Mini high-performance workstation for engineering calculations, and also a new generation of T-Blade V-Class beginner level systems. Along with hardware, we will show a number of software solutions developed by our company. In particular, we will present a set of Clustrx software products, developed by T-Platforms to manage supercomputers. During the event, as most companies do, we will hold a number of negotiations with representatives from the leading western developers of solutions for the HPC industry and global supercomputer centers, and we hope to sign collaboration agreements with new solutions’ sales partners for the European market.

insideHPC: When China took the number one spot on the last TOP500 list, there was quite a stir in the U.S. supercomputing community. Has there been a similar response in Russia?

Opanasenko: In the latest TOP500 list, the Lomonosov supercomputer was placed 17 out of the world’s most powerful supercomputers. I think that in the new Top500 list this supercomputer will occupy an even higher position, marking the achievements of Russia quite noteworthy in the international HPC industry.

insideHPC: 2012 will mark your 10th year in business as an HPC vendor. What are the unique challenges of running a high-technology company in Russia, and how important is it to your organization’s growth that you expand into other countries?

Opanasenko: Next year we will celebrate a ten-year anniversary. Over these years, T-Platforms has changed from a small group of associates devoted to their work, into an international HPC company. And of course, just like any high-tech company working in this market, we face certain difficulties. At present, our business strategy is focused on the development of new technologies, creating products and services based on them, and also active expansion into international markets. It would be impossible to realize these plans without attracting investments from both the state and private capital. But last October, one of our projects in the sphere of supercomputer technologies was supported by Russian state corporation “Bank of Development and External Economic Activity” (Vneshekonombank). The bank’s participation in providing capital for T-Platforms will allow us to strengthen the technological leadership and competitive advantage of Russian developments on the global HPC market. The funds attracted in this way will be allocated to the development of new products, the development of international partner cooperation to create a new-generation supercomputer, as well as active development on the US and European markets. It will allow us to open up new markets, realize new projects both for scientific organizations and on the Enterprise market. It is also quite difficult to attract high-class specialists who can develop the architecture for hardware and software complexes. To this end, we cooperate with Russia’s leading technical universities and pay great attention to the training of employees, and in this way create a personnel reserve inside the company.

insideHPC: In many ways, the TOP500 is the metric used to measure world technology leadership today. Can we expect to see Russia continue to grow its presence at the upper end of the TOP500?

Opanasenko: Today we are actively working on the development of innovative technologies which will form a basis for new-generation computer complexes in the near future. These technologies will allow us to become good competitors to the world’s best-known HPC market players, and petascale and exascale supercomputers, developed with this new technology foundation will occupy higher and higher positions on the Top500 list.

insideHPC: Your web site lists an impressive list of partners on the hardware and software sides. How important are these business relationships for T-Platforms as it builds its business in new markets? Will your sales model moving forward be more of a direct sales model or that of distributors and system integrators?

Opanasenko: Our strategy for working with partners is aimed at attracting technological leaders to cooperate with us, which is to say those who develop the products adapted in the best way for HPC solutions, analogs of which we still cannot produce on our own. However, as the market is quite narrow, the number of such companies is not very high. As for strengthening our positions in new markets, we rely on developing cooperation with channel partners that are interested in presenting our solutions on local markets. Experience has shown that they have the best knowledge of local market needs, can evaluate the current and future needs of potential customers of our solutions and competently sell complicated computer complexes. We are planning to stick to this model when we enter new markets in the future.