What is High-Performance Computing?

Solomon Branch

High-performance computing (HPC) is the use of advanced computer technology, such as supercomputers and parallel processing algorithms, to figure out complex computational problems. The technology is mostly used by those in in the field of science, the military, and academia. As business transactions become more complex, HPC is being utilized in the areas of data storage and transaction processing, as well as mining or using that information for statistical purposes.

Supercomputers are often used in high-performance computing.
Supercomputers are often used in high-performance computing.

The main goal in high-performance computing is to solve computational problems with accuracy, speed, and efficacy. Parallel processing, which uses more than one central processing unit (CPU) to calculate a problem and technically should make the calculation much faster, is key to HPC success. Other forms of increased calculation speed, including multiple computers linked together known as clustered computers, also contribute to high-performance computing.

IBM's Deep Blue was a supercomputer that exhibited artificial intelligence by playing chess against human Grand Masters.
IBM's Deep Blue was a supercomputer that exhibited artificial intelligence by playing chess against human Grand Masters.

Supercomputers are also used in the process of high-performance computing. The terms supercomputer and supercomputing can be confusing because sometimes they are interchanged with the terms high-performance computing and high-performance computer, but they can also be considered a specialized form of high-performance computing. For all intents and purposes, however, supercomputer refers to a computer than can calculate a high rate of efficiency.

Computations done by a supercomputer are rated at the speed in which it can perform floating point operations per second (FLOPS), which is the amount of calculations with numbers it can perform per second. Floating point refers to the decimal point in a number which, since it is “floating", can represent a number with a higher degree of accuracy than a fixed-point operation. The amount of FLOPS needed to be considered a supercomputer is a teraflop or more, a value expressed as 10 to the 12th power in scientific notation, or 1,000,000,000,000 FLOPS. There are supercomputers, however, that can go up to petaflops, which is 1,000,000,000,000,000 FLOPS.

Use of high-performance computing also extends to certain technical problems, typically in engineering fields, such as the petrochemical industry, and computational fluid dynamics. When used this way, HPC is referred to as high-performance technical computing (HPTC). There is overlap between all the fields, so it is not always clear what constitutes HPTC or HPC, and the terms are in many ways synonymos.

The top high-performance computers are stored in a list known as the Top500. This list was created to centralize a database as well as create criteria for defining high-performance computing standards. It is kept by scientists in Germany and the US, and is updated twice per year.

You might also Like

Discuss this Article

Post your comments
Forgot password?