Today’s supercomputers

The computing power of supercomputers is usually measured by the number of operations that they can perform on numbers with decimal (i.e., floating) points. This measure is therefore called “floating point operations per second,” or Flops, for short. For example, adding the number 6.3 to the number 2.6 to obtain the number 8.9 constitutes one computer operation. Your single-core microprocessor desktop PC can probably perform one billion of such additions in one second. Thus, your single-microprocessor PC will be said to operate at approximately 1 GFlops. In comparison, the most powerful computer in the world today can operate at 33,862.7 Gflops or approximately thirty four thousand times that of your single-processor PC. (A gigaflop is one billion – “nine zeros” – operations per second. A teraflop in one thousand gigaflops, or a million million – “twelve zeroes” – flops; a petaflop can be measured as one thousand teraflops – fifteen zeroes. A petaflop is also equal to one quadrillion floating point operations per second.) Now you can understand why your PC cannot solve the equations that govern weather patterns across the globe, or predict how Boeing 747-200 aircraft will react to turbulence in flight, whereas the Tianhe-2 (Milky Way-2) Computer at the National Super Computer Center in Guangzhou, China, which is today’s fastest computer in the world, can get the job done! Computing activities that are based on the use of supercomputers are often referred to as high-performance computing (HPC).
The Configuration of Supercomputers
The way they are built these days, supercomputers (SCs) are in a sense not that special relative to a standalone PC. This is because PC-based SCs are a direct combination of a large number of microprocessors (or computer chips). Most of the PC-based SCs are made from commodity (i.e., standard) chips, and are therefore not that esoteric. This makes one wonder why some ambitious Obafemi Awolowo University (OAU) student is not busy building one up in his apartment. Obviously, the task is not that trivial, but it is the case that SCs derive their massive power from the sheer number of microprocessors that they contain. Other than the attributes of “greenness,” energy efficiency, cooling requirements, and the sort, the number of microprocessors (actually, cores), GFlops, and the manufacturer of the microprocessors, are important metrics of a supercomputer deployment.
Key players
As of November 2014, according Top500 – an organization devoted to supercomputers, – the ten most powerful supercomputers in the world were built in China (1, or one supercomputer in top 10), Japan (1), Germany (1) Switzerland (1) and the United States (6). According a New York Times article in late 2011, the Chinese are about three generations behind the state-of-the-art chip making techniques used by world leaders such as the United States, South Korea, Japan, and Taiwan. Thus, even though some of the most powerful computers have been built outside of the United States, the chips for the top ten supercomputers are made by American companies, with the exception of the Tianhe-2 in China and the K Computer in Japan, whose chips are manufactured by NUDT and Fujitsu, respectively. (NUDT, or the National University of Defense Technology, is a strategic national key university based in Changsha, Hunan Province, China.) The K computer, with 705,024 cores, was the fastest supercomputer in the world in November 2011, but the fourth-fastest in November 2014. The fastest computer today, the Chinese Tianhe-2, has 3,120,000 cores, or four times the number of cores for the K Computer. Tianhe-2 has been on the top of the list since June 2013. The second and third fastest supercomputers in the world are in the U.S.: The Titan – Cray XK7 (Department of Energy, DOE) and the Sequoia – IBM BlueGene (also DOE), have 17,590 and 17,173 Tflops, with 560,640 and 1,572,,864 cores, respectively. Although the United States is still in the lead overall, its relative position is slipping.
Sample Applications of supercomputers
In Science/Engineering, the computational procedure of predicting the outbreak and spread of fires, aerodynamic or hydrodynamic performance of aircraft, automobile design, ships, or virtually any land, sea, or air-based vehicle, require the use of HPC because of the large size of the models. Because of the intensive computational requirements, HPC is usually the only option available for data analysis and data mining.  It is easy to understand why the field of meteorology and weather forecasting requires voracious consumption of computational resources. The model equations for weather forecasting need to be solved to details that allow accurate interpolation to cover small villages as well as large international regions. Finally, military strategies and scenario analysis depend on the power of supercomputers.
Supercomputing in Africa
There is not a single supercomputing facility from Africa that has made the Top 500 list. Between 2007 and 2009, I carried out a casual study of supercomputing facilities in Nigeria, and came to the conclusion that there was none! My attempts to change the situation were well received by the then Minister of Education, Dr. Igwe AjaNwachuku; but, like most initiatives in Nigeria at the time, when a new boss takes over, he comes with his own group of people and “projects.” Nigeria can still not boast of a fully functioning supercomputing facility. In the interim, we will probably have  much success with Ghana and South Africa.