TOP500 project lists the top 500 supercomputers of the world on the basis of a parameter called LINPACK benchmark. The biannual list is usually released in June and November. The benchmark was an idea conceived by Jack Dongarra. In simple terms, it’s the rate at which the supercomputer solves floating-point operations and is evaluated by making the supercomputer solve a dense system of linear equations.
There are two things that set the Tianhe-2 apart from the top ranking. Firstly, it is the first supercomputer in the world which can perform a sustained computing speed of 33.86 petaflops per second—which means it can perform a total of 33,860 trillion calculations per second. Secondly, it has almost twice the speed of the next best system on the list, which stands at 17.59 petaflops per second. The previous title holder was Titan, a Cray XK7 system installed at the U.S. Department of Energy’s (DOE) Oak Ridge National Laboratory, now in a distant second place.
The Tianhe-2 isn’t the first super computer to emerge from China. China’s foray into the super computer field started with the introduction of Yinhe-I in 1983 with a performance level of 100 megaflops. Almost ten years later in 1992 it introduced Yinhe-II, achieving a performance of 1 gigaflop. Yinhe-III, an upgraded version of Yinhe-II introduced in 1996 achieved the performance level of 13 gigaflops, but it was still far behind the top supercomputers of the world.
But in 2009 China introduced the Tianhe-I which was immediately ranked as the world’s fifth fastest supercomputer in the TOP500 list. In November 2010 China took the top ranking for the first time with the Tianhe-1A, with a computing speed of 2,566 teraflops. Though the system lasted on the top spot for only a short while, it was an important moment in China’s supercomputing history. Since then China has progressed much in this field and Tianhe-2 is likely to stay right at the top for at least another couple of years.
China considers technology as a core component of projecting a nation’s capability and prestige; national strength is allied to the ability to flex its muscles on the technological front. Supercomputing stands at the forefront of technology, with extensive use in practically every field, especially defence applications.
The USA began the initial surge in production of supercomputers and has until today stood at the top with reference to the number of systems in the Top500 list. Other well-represented countries include France, Germany, Japan UK and India. China’s growth in this field has been unprecedented. Its obsession with technology and related fields can be fathomed from the fact that in the list released in November 2001, China had just three systems in the TOP500. China now has over 65 systems in the June 2013 list, including the number one position.
There are several ways that China managed to achieve this growth. The foundation of this model rests on China’s economic and technology programs following the liberalisation of its economy under Premier Deng Xiaoping. As well as its five year plans, China has three core policies which have aided in the development of Science and Technology (S&T). These are the 863 plan formulated in 1986, the 973 plan (1997) and the MLP plan (PDF) of 2000.
Since China has always been heavily dependent on foreign technology, there have been calls to try and reduce the dependence and promote national innovation capabilities. Each of the S&T plans has its own goals and objectives, but the common factors can be summarised as: to help boost innovation capacity in China, to achieve breakthroughs in areas of national interest and to improve China’s overall development, R&D capacities and socio-economic conditions.
The field of supercomputing holds an important position in these policies, either directly or indirectly and the setting up of supercomputing centres in different provinces is a direct result of these economic and technological policies. The National University of Defence Technology (NUDT) and Supercomputing Center of Chinese Academy of Sciences (SCCAS) are the leading institutions responsible for the developments.
To prove Chinese rapid advancement in the field, China released the the Tianhe-2 almost two years earlier than the expected release date of 2015. Even though the system still runs on Intel cores, China has managed to develop its own interconnections, operating system, front-end processors and software. This is a big moment for China as it seems to be delivering on its promise of boosting national innovation capabilities.
Even though the actual applications of the Tianhe-2 haven’t been formally listed, it’s believed to be a multipurpose computer which can be expected to predict earthquakes, aid in the development of new drugs, help in designing automobiles and military platforms, aid in creating visual effects, calculations for nuclear simulations etc.
The next big step in the field of supercomputing is to breach the exaflops barrier (which is more than one billion billion calculations per second). The early delivery of the Tianhe-2 gives China time to work on the exa-system while still holding the top rank. It’s evident that China has, for the time being at least, knocked out the rest of the world.
Nabeel A Mancheri is a Research Fellow, Institute of Social Science, Tokyo University. Viswesh Rammohan is a Research Associate at the National Institute of Advanced Studies, Bangalore. Image courtesy of the NASA Goddard Space Flight Center.