Supercomputers were a massive race in the '90s, while the US, China and others all competed to get the fastest computer. Although the race is a little extinct, these monster computers have always been used to solve many problems of the world.
As Moore's Law (an old observation claiming that computational power doubles about every two years) pushes our hardware further, the complexity of the problems to be solved also increases. While supercomputers were relatively small, they can now occupy entire warehouses, all filled with interconnected computer racks.
What makes a computer "super"?
The term "supercomputer" implies a gigantic computer much more powerful than your simple laptop, but it could not be farther from the case. Supercomputers are made up of thousands of smaller computers, all connected together to perform a task. Each processor core of a data center probably runs more slowly than your desktop. It is the combination of all these factors that makes computing so effective. There are many networks and special hardware involved in computers of this size, and it is not as easy to connect each rack to the network, but you can imagine them that way and you would not be so far from the goal.
All tasks can not be paralleled so easily. You will not use a supercomputer to run your games at one million frames per second. Parallel computing generally speeds up calculations-oriented calculations.
Supercomputers are measured in FLOPS, or floating-point operations per second, which is essentially a measure of how quickly it can do calculations. The fastest is currently IBM Summit, which can reach more than 200 PetaFLOPS, a million times faster than "Giga", which most people are used to.
So what are they for? Mainly Science
Supercomputers are the backbone of computer science. They are used in the medical field to perform protein folding simulations for cancer research, in physics to perform simulations for large engineering projects and theoretical calculations, and even in the financial field to track the market stock market to stand out from other investors.
Perhaps the job that benefits the average person the most is weather modeling. To say with precision if you will need a coat and an umbrella next Wednesday is a surprisingly difficult task, a task that even the gigantic supercomputers of today can not do with a great precision. According to our theories, to be able to perform a complete weather modeling, we will need a computer that measures its speed in ZettaFLOPS, two more stages since PetaFLOPS and about 5000 times faster than the IBM Summit. We will probably not aim for this point until 2030, although the main problem that holds us back is not the material, but the cost.
The initial cost for the purchase or construction of all this equipment is high enough, but the decisive factor is the electricity bill. Many supercomputers can use millions of dollars in energy each year just to keep going. So, while there is theoretically no limit to the number of computer-filled buildings that you can link together, we only build supercomputers big enough to solve the current problems.
Can I have a supercomputer at home in the future?
In a sense, you already do it. Most desktops today compete with the power of older supercomputers, even the average smartphone offers better performance than the infamous ones. Cray-1. It is therefore easy to compare with the past and theorize about the future. But this is due in large part to the fact that the average processor has become much faster over the years, which is not happening as fast.
Lately, Moore's law has slowed down as we reach the limits of the minimum size of transistors, so processors are not much faster. They are becoming smaller and more energy efficient, pushing processor performance in the direction of a higher number of cores per chip for desktops and more globally more powerful for mobile devices.
But it is difficult to envision the problem posed by the average user to the growing computing needs. After all, you do not need a supercomputer to browse the Internet, and most people do not run protein folding simulations in their basements. Today's high-end consumer hardware far exceeds normal use cases and is generally reserved for specific works that benefit, such as 3D rendering and code compilation.
So no, you probably will not. The biggest advances will probably be in the mobile space, such as phones and tablets. approach the office power levelswhich is still a pretty good advancement.