What is Supercomputer?
Going back half a century, even the smallest computer of that time would have occupied a whole room. And today, thanks to microchips, more powerful computers than ever before have been pocketed. But have you ever wondered how powerful a home-sized computer would be if it were made with these tiny chips? Is it possible to make a super computer in this way? —And super computing power is a million times more than your or my laptop or desktop.
In general, computers that are much more powerful and larger than the small computers used in our homes or office are called supercomputers. These are much more powerful than ordinary computers and are much larger in size. The average cost of building a supercomputer is about 100 million and its one-year maintenance cost is 9 million !! One of the world’s largest supercomputers, the Titan, is about the size of a basketball court. Unlike ordinary computers, supercomputers are not available in the market and very few companies in the world have supercomputers. Mostly government-based and used in various research institutes for large scale accounting. These are very expensive to maintain.
History of Supercomputer
The world’s first supercomputer was built in 1960 at the University of Manchester by IBM 7030 Stretch and a few computers from the CDC (Control Data Corporation) called Atlas and was built by a man named Simor Cray. The IBM 7030 Stretch is the first computer to use a transistor. In 1974, Cray designed a supercomputer called the CDC-6600, which used a silicon transistor. Each computer sold for 8 million. In 1986, Cray developed an 80 MHz computer called the Cray-1 from his own company, Cray Research.
The ILLIAC IV model outperforms the Cray 1 with a maximum peak performance of 1gigaFLOPS. In 1975, another computer, the Cray 2, arrived at a maximum speed of 1.9 gigaFLOPS, but it did not overtake the Moscow-based M-13 supercomputer. The invention of the microprocessor revolutionized the way computers were built and speeded up.
In 1972, a supercomputer called LINKS-1 was built in Japan primarily for graphics work, using a total of 514 microprocessors. Construction of supercomputers with thousands or more microprocessors began in Japan and the United States in the 1990s. In 1998, a supercomputer with 2048 processors named Hitachi SR2201 was able to accelerate to 600 gigaFLOPS. Since then, supercomputers with extremely parallel processors have been built.
Countries like China, Japan, Russia, etc. became interested in building powerful supercomputers for themselves. Currently, five of the world’s top ten supercomputers are located in the United States, two in China, one in Japan, and one in Switzerland. Of the first three, however, two are from China and the third from Switzerland.
Supercomputers are far more powerful than ordinary computers. To measure the speed of a computer, we usually determine how many hertz processors it has and how many megabytes / gigabytes of data it can process per second. But a completely different unit is used to measure the speed / power of supercomputers.
The unit is FLOPS (Floating Point Operation Per Second). It is completely different from the conventional calculation method and is used only to measure the speed of large processing machines. Let me give you a little idea about the FLOPS unit – 1gigaFLOPS means you have to calculate every second for about 32 years !!
If a supercomputer could do this in one second, it would be capable of 1gigaFLOPS. Similarly TeraFLOPS, PetaFLOPS units are also used.
1000 gigaFLOPS = 1 TeraFLOPS
1000 TeraFLOPS = 1 PetaFLOPS !! Even if you calculate 1 per second, it will take you 30 million years to get 1 petaflops !!!
Currently, the most powerful supercomputer is Fugaku. This computer is capable of calculating 2.8 times faster than the Summit computer. The computer is valued at 1.2 billion. It has 150 k high-capacity processing units
|Name||Start Year||Performance(pFlops)||Cost(million USD)||Ranking||CPU/GPU vendor||CPU||OS|
|FugaKu||2020||415||1213||June 2020 (1st)||Fujitsu||A64FX|
Custom Linux-based kernel
|Summit||2018||148||200||June 2018-Nov 2018 (1st)||IBM,NVIDIA||POWER9, Tesla||Linux (RedHat)|
|Sierra||2018||94||200||Nov 2018-Nov 2019(2nd)||IBM,NVIDIA||POWER9, Tesla||Linux (RedHat)|
|Sunway TaihuLight||2016||93||380||June 2016-Nov 2017(1st)||NRCPC||Sunway SW26010||Linux (Raise)|
|K||2011||10||1048||June 2011-Nov 2011(1st)||Fujitsu||SPARC64 VIIIfx||Linux|
Features of Supercomputer
They can support hundreds of users at once. Machines are capable of performing large amounts of calculations that are beyond human ability, meaning humans are unable to solve such large calculations. Many people can access supercomputers at the same time. These are the most expensive computers ever made. CPUs can support extremely high computing speeds. They can work on lists of numbers rather than pairs of numbers. They were primarily used in applications related to national security, nuclear weapon design, and cryptography. But nowadays they are also employed by the aerospace, automotive and petroleum industries.
Use of Supercomputer
Supercomputers are not used for day-to-day work because of their superiority. Supercomputers run applications that require real-time processing. Uses are as follows: Used for scientific simulation and research such as weather forecasting, meteorology, nuclear energy research, physics, and chemistry, as well as for highly complex animated graphics.
They are also used to explain new diseases and to predict illness behavior and treatment. The military uses supercomputers to test new aircraft, tanks, and weapons. They use them to understand the effects of soldiers and war. These machines are also used to encrypt data. Scientists use them to test the effects of nuclear weapons explosions.