August 13, 2009
Russian President Dmitry Medvedev thinks his country's supercomputing capabilities need a jump-start. In an address to Russia's Security Council in late July, Medvedev chided his fellow bureaucrats that the country has failed to invest in supercomputing or grid technologies, putting the nation's security and industrial competitiveness at risk. His speech began by laying out the case for these technologies:
It's no secret that the majority of the most developed and advanced nations are focusing on this. It is obvious that the large-scale use of high technology data processing increases the effects of research many times over, radically reduces the cost of designing the most advanced and complex types of products, naturally increases the quality of industrial products, and streamlines business processes. It is precisely for these reasons that the entire world is working on this. Any country that makes headway in relation to creating supercomputers has, of course, advantages in terms of competitiveness, increasing its defence capacities, and strengthening security.
Medvedev went on to complain that Russia ranks only 15th in the aggregate capacity of its supercomputers, noting that "476 out of the 500 supercomputing systems use computers manufactured in the United States of America." Although he didn't mention it, Russia's top system, a 71.3 teraflop (Linpack) HP machine at the Joint Supercomputing Center in Moscow, has less than 7 percent the Linpack performance of the top system in the world, IBM's Roadrunner supercomputer. Even the top 50 systems of the CIS states (the former Soviet Republics) currently have an aggregate Linpack performance of just 382 teraflops, or about one third the power of the single Roadrunner machine. Considering that Russia's 2008 GDP of $2.225 trillion (according to the CIA World Factbook) places it 8th in the world, the country is definitely underachieving in the HPC realm.
Medvedev also brought up the fact that commercial use of supercomputing in Russia is woefully behind the times:
[W]e have only extremely few aircraft (actually one airplane) created on a supercomputer, that is only one that exists in digital form. Everything else is done on Whatman’s drawing paper like in the 1920s and 30s using the old approaches. It’s obvious that here only a digital approach can have a breakthrough effect, lead to dramatic improvements in quality, and reduce the cost of the product.
If all of this sounds familar, you are probably recalling similar speeches delivered by high-level government officials and industry stakeholders in the US, Europe, and Asia over the past several years. But the fact that this HPC cheerleading came from the head of state rather than just a high-level bureaucrat probably bodes well for Russia.
Unfortunately, Medvedev's speech didn't offer much in the way of solutions, except to suggest a general commitment to "invest in the production of supercomputers" and "stimulating demand in every possible way." It gets even fuzzier. It's not clear to what extent Russia wants to rely on foreign HPC technology versus developing its own. As it stands today, IBM, HP and SGI own a good chunk of the Russian HPC server market.
In an ITAR-TASS report in July, Secretary of the Russian Security Council Nikolai Patrushev expressed willingness to cooperate with the US and perhaps other countries on supercomputing technology, but hedged on far those relationships could go. "[W]e are facing a task to use the existing experience, particularly that of other countries, as well as to create our own development base, and we will work on the issue," he said.
One element that has to be taken into account is the country's need to test its nuclear deterrent with supercomputers. I imagine the Russians would get a bit squeamish about depending upon systems or software developed in the West to support its nuclear weapons programs. So don't expect to see IBM shipping Roadrunners to Moscow anytime soon.
Fortunately for Russia, the country does have some critical pieces of an HPC ecosystem already in place, the most important of which is a well-trained cadre of native mathematicians, computer scientists, and engineers. Secondy, there's T-Platforms, Russia's homegrown HPC vendor, that currently supplies about a third of the domestic market. T-Platforms' latest HPC blade offering based on Intel Nehalem chips is capable of scaling up to petascale-sized supercomputers, and I wouldn't be surprised to see such a deployment as early as 2010.
Posted by Michael Feldman - August 13, 2009 @ 5:46 PM, Pacific Daylight Time
Michael Feldman is the editor of HPCwire.
No Recent Blog Comments
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?