If you are well acquainted with the technological world, you must have heard of supercomputers. However, people who are not so present in this world, have also questioned themselves regarding this equipment that is capable of unimaginable things, technologically speaking.
To the letter, the supercomputer is nothing more than a computer with a very high processing speed. In addition, another point worth mentioning is the large memory capacity it has. Generally, its application is given in research areas that use large data processing.
These researches can be directed to military, chemical, biological and nuclear areas. This is because, for them to be developed, it is necessary to solve very complex calculations and intensive tasks, such as problems involving quantum physics, mechanics, meteorology, climate research, molecular modeling and physical simulations.
Photo: Reproduction / Wikimedia Commons
A little of history
The origin of the first supercomputers is given in the 1960s. The person responsible for the creation was Seymour Cray, who founded his own company, Cray Research, in 1970.
It was also in the 1970s that the University of Illinois assembled together with Burroughs Corporation the ILLIAC IV, a supercomputer that became famous for its dimensions. Today supercomputers are manufactured by companies such as SuperMicro, Nec, Sun, IBM, HP, Apple Inc, among others.
How are supercomputers
One of the first characteristics that appear as a synonym for supercomputers is the processing speed. This speed is generally greater than 80 TFlops. This guarantees processing of trillions of floating-point operations per second.
Regarding the physical space it occupies, even by its name, you can already have an idea that they are gigantic, requiring special cooling systems. But, these supermachines are not available to anyone. They are mostly used by major research centers. So much so that, currently, less than a thousand are in operation around the world.
Types of supercomputers
Depending on the task performed, supercomputers need to load information from specific assemblies. See what are the most common types:
Parallel Vector Processors (PVP)
These systems are made up of few processors. The interconnection is made, in general, by a high flow switching matrix (crossbar). Memory is shared and systems can be classified as multiprocessors.
Symmetric multiprocessors (SMP)
These are systems made up of commercial processors connected to shared memory. They are easier to program than machines that communicate by exchanging messages, since it is close to that made in conventional systems. The disadvantage in use comes from the limitation of interconnections.
Massively parallel machines (MPP)
They are computers built with thousands of commercial processors connected by a high-speed network. High performance is achieved with the large number of processors.