Computers have been part of our society in one way or another since the activation of ENIAC in the mid-1940s.

The definition of a supercomputer has changed over the decades. In 1951, the UNIVAC could process only a few thousand computations, or instructions, in one second. By 1975, the Cray 1 supercomputer ran 160 million instructions per second (MIPS) on large amounts of data.

In the 21st century, supercomputers have a similar setup as their ancestors. They take up a large amount of data center floor space. The main difference is instead of a few processors, they run multiple instances for each device. Together, they calculate billions of billions (quadrillions) of instructions per nanosecond. This is known as exascale supercomputing, where measurements change from MIPS to floating point operations (FLOPS).

These types of devices aren’t performing simple internet searches. Rather, supercomputers carry out the compilation of Big Data in the business and science sectors. For instance, they might be used for modeling molecular structures. Or government organizations such as the National Oceanic and Atmospheric Administration (NOAA) might use them to forecast potential severe weather patterns.

There are two types of supercomputers: general purpose and special purpose.

Vector processing supercomputers rely on array processors. These are similar to a central processing unit (CPU) of a standard computer. However, they perform rapid mathematical operations on a large number of data elements. These were the basis of the supercomputer industry in the 1980s and 90s, and today’s devices still have some form of vector processing instruction.

Clusters refer to groups of connected computers that work together as a supercomputing unit. An example is a group that runs high-powered database programs that help produce results from the compilation of Big Data.