What is a Supercomputer and when was it created?

What is a supercomputer and when was it created?

What is a supercomputer? We all know about computer but do you know about supercomputer? In listening, it seems to be the super version of Computer. And it is also true that supercomputer is called such a device that performs better and faster processing of all existing computers. If we talk of earlier times, then vaccum tubes and transistors were used in computers and computers used to come in large rooms. But since the time of Integrated Circuit or microchips came, now the size of computers has reduced to a great extent.

But ICs are used in supercomputer, Microchips are used in large numbers so that there is no difference in their size. Therefore we do not see supercomputer small size.

But their processing speed is thousands of times faster than all other normal computers. Here today in this article, we will know what is called a supercomputer, how it works and what are its advantages compared to the rest of the traditional computer. So let's start without delay and get complete information about what a supercomputer is.

What is a supercomputer?

Before knowing what a super computer is, if we know what a computer is, then we will have some ease in understanding it. Talking about a computer, then it is a general-purpose machine that takes information (data) through the input process, stores them and then processes them as needed, and finally produces some type of output. is.

On the other hand, if I talk about a supercomputer, it is not only much faster and a much bigger computer: but it works completely differently, it typically uses parallel processing instead of serial processing like an ordinary one. Used in computer. So instead of doing one thing at a time, it does multiple work at one time.

A supercomputer is a computer that currently performs at the highest operational rate. After all, where is the supercomputer used? Traditionally, supercomputers are mostly used for doing scientific and engineering applications so that they can handle large databases as well as do a large amount of computational operations. Performance wise it works thousands of times faster and accurate than normal computers.

The performance of the supercomputer is measured in FLOPS, which means floating-point operations per second. Therefore, the more FLOPS the computer has, the more powerful it will be.

What is Serial and Parallel Processing?

Let us know what is the difference between Serial and Parallel Processing? In an ordinary computer, only one work is done at a time, meaning that another work is processed only after the completion of one work, such processing is called serial processing.
For example, a man is sitting in the grocery checkout of a retail mall and after picking up whatever goods comes in the conveyor belt, he scans the scanner and passes it in the customer's bag, it would do the work in a distinct series of operations. So it is called series processing. Here, no matter how soon you place things in the conveyor belt or fill things in your bag after the scan, the speed of this process depends on the scanning speed or processing of that operator, and which is always one item at a time. Happens in The best example of this is the Turing machine.

At the same time a typical modern supercomputer works at very high speed for which it splits the problem into small pieces and works in one piece at a time. Therefore this process is called parallel processing.
If at the same grocery checkout, a lot of friends divide the items among themselves and checkout together in different counters and later collect all the things in one place, then it will work very soon and will not take much time. Since the work was divided here, it did not take much time for processing. That's why Parallel Processing is much faster than serial processing.
The largest and powerful supercomputers use parallel processing. With this, they can do any process in fast and short time. When it comes to big and complex work like weather forecasting (weather forecasting), gene synthesis, mathematical modeling etc. then we need computing power in the right way. In this case, parallel processing to Supercomputer is more useful. Generally speaking, there are mainly two parallel processing approaches: Symmetric multiprocessing (SMP) and Massively parallel processing (MPP).

What is clusters?

If you want, you can create a supercomputer for which you have to put many processors in a giant box and instruct them to solve the complex problem for which they can use parallel processing.

Or there is another way in which you have to buy a lot of off-the-self PCs and keep them in the same room, interconnecting them with each other with the help of a fast local area network (LAN). So that they will work broadly the same way. This type of supercomputer is called Cluster. Google uses these cluster supercomputer for web searches of its users in its data centers.

What is grid?

Grid is also a supercomputer which is very similar to a cluster (which is a group of separate computers), but it contains computers in different places with each other through the Internet (or another computer network). are connected. This type of computing is also called distributed computing, in which the power of the computer is spread across multiple locations in exchange for a single place (centralized computing).

For example, the CERN Worldwide LHC Computing Grid, in which the data from the LHC (Large Hadron Collider) particle accelerator is assemble in one place, uses a grid supercomputer.

In Grids Supercomputer there are less chances of failure, as all the computers are connected to each other so they can overcome the problems caused by parallel processing, where break up is a common thing.

Which Operating System Does Supercomputers Use?

You may be surprised to know that ordinary operating systems are used to run supercomputers that we run on our PCs, but we know that more modern supercomputer actuals include off-the-self computers and workstations. Are clusters of.

Until a few years ago, Unix was used according to the operating system, while Linux is still used in its place. Which is open-source. Since supercomputers generally work on scientific problems, their application programs are written in traditional scientific programming languages ​​such as Fortran, or more popular modern languages ​​such as C and C ++.

How powerful are supercomputers?

If we talk about normal computers, then MIPS (million instructions per second) is used to measure their computing speed. By which fundamental programming commands such as read, write, store etc. are managed by the processor. To compare two computers, their MIPS are compared.

But the way to rate supercomputers is slightly different. Since most scientific calculations are done in this, they are measured by floating point operations per second (FLOPS). Let us see the list made according to this FLOPS.

Unit	FLOPS	Example	Decade
Hundred FLOPS	100 = 10 power 2	Eniac	~1940s
KFLOPS (kiloflops)	1 000 = 10 power3	IBM 704	~1950s
MFLOPS (megaflops)	1 000 000 = 10 power 6	CDC 6600	~1960s
GFLOPS (gigaflops)	1 000 000 000 = 10 power 9	Cray-2	~1980s
TFLOPS (teraflops)	1 000 000 000 000 = 10 power 12	ASCI Red	~1990s
PFLOPS (petaflops)	1 000 000 000 000 000 = 10 power 15	Jaguar	~2010s
EFLOPS (exaflops)	1 000 000 000 000 000 000 = 10 power 18	?????	~2020s

If you study the history of computers, then you will find that not one individual has contributed in this, but many people have given their contribution from time to time. Then we got to see such amazing machines. But when it comes to SuperComputer, then a big credit goes to Seymour Cray (1925–1996). Because his contribution is highest in Supercomputer. You can also call them the father of a supercomputer.When and who made the world's first supercomputer

1946: John Mauchly and J. Presper Eckert constructed ENIAC (Electronic Numerical Integrator and Computer), University of Pennsylvania. It was the first general-purpose, electronic computer, it was about 25m (80 feet) long and about 30 tons its weight. It was created to operate military-scientific problems and was the first scientific supercomputer.

1953: IBM developed the first general-purpose mainframe computer, named IBM 701 (also known as Defense Calculator), and sold about 20 machines to different government and military agencies. The 701 was the first off-the-shelf supercomputer. Gene Amdahl, then an engineer at IBM, later redesigned it and named the upgraded version IBM 704, a machine that had a computing speed of around 5 KFLOPS (5000 FLOPS).

1956: IBM then developed the Stretch supercomputer for Los Alamos National Laboratory. It has been the world's fastest supercomputer for nearly 10 years.

1957: Seymour Cray co-found Control Data Corporation (CDC) this year and pioneered in making fast, transistorized, high-performance computers of which CDC 1604 (announced 1958) and 6600 (released 1964) were the main ones who seriously challenged Done on IBM's dominance over mainframe computing.

1972: Cray left Control Data to set up its own Cray Research and built high-end computers — the first true supercomputer. Their main idea was how to reduce the connections within the machine so that the speed of the machines can be increased. Earlier Cray computers were often C-shaped, so that they could be kept separate from others.

1976: The first Cray-1 supercomputer was installed at the Los Alamos National Laboratory. Its speed then was around 160 MFLOPS.

1979: Cray again developed a faster model than before, which had eight-processor, 1.9 GFLOP Cray-2. In this, the wire connections were reduced from 120 cm to 41 cm (16 inches).

1983: Thinking Machines Corporation again manufactured massively parallel Connection Machine, in which around 64,000 parallel processors were used.

1989: Seymour Cray then founded a new company, Cray Computer, where they developed Cray-3 and Cray-4.

1990: Due to a cut in defense spending and the evolving of powerful RISC workstations, by companies such as Silicon Graphics, it was posing a serious threat to supercomputer makers.

1993: Fujitsu Numerical Wind Tunnel made the world's fastest computer using 166 vector processors.

1994: Thinking Machines filed a case for bankruptcy protection.

1995: Cray Computer also started sinking due to financial difficulties, so they filed a case of bankruptcy protection Simultaneously Seymour Cray died in a road accident on October 5, 1996.

1996: Cray Research (Cray’s original company) was purchased by Silicon Graphics.

1997: ASCI Red, a supercomputer built by Pentium processors by Intel and Sandia National Laboratories, became the world's first teraflop (TFLOP) supercomputer.

1997: IBM's Deep Blue supercomputer defeated Gary Kasparov in a chess game.

2008: Jaguar supercomputer created by Cray Research and Oak Ridge National Laboratory became the world's first petaflop (PFLOP) scientific supercomputer. Which were later defeated by the machines of Japan and China.

2011–2013: Jaguar was extensively (and expensively) upgraded, and renamed Titan, and later became the world's fastest supercomputer, later downgraded by the Chinese machine Tianhe-2.

2014: Mont-Blanc, a European consortium who announced that they are going to build an exaflop (1018 FLOP) supercomputer from energy efficient smartphone and tablet processors.

2017: Chinese scientists announced that they are making a prototype of an exaflop supercomputer, which is based on Tianhe-2.

2018: China is at the forefront of the fastest supercomputers race, the Sunway TaihuLight made by them is the fastest running supercomputer in the whole world.

Which is the world's Top 5 Fastest Supercomputers?

There is a lot of competition in all countries about computing power, which can be at the forefront, but the top position is the same. Peak performance in supercomputing always changes. Even in the definition of supercomputer, it is written that it is a machine "which always works at its highest operational rate."