SUPER COMPUTING EXPLAINED
Super Computing: refers to the use of parallel processing for running advanced application programs efficiently, reliably and quickly. It’s also known as High Performance Computing (HPC).
The term applies specifically for systems that can perform above a teraFLOP (1012 Floating Point) operations per second. It also refers to the ability of a computer to process data and perform complex calculations at high speeds with no latency.
For example while a desktop with a 3GHz processor can perform around 3 billion calculations per second it cannot match or compare to High Power Computing Systems (Super Computing Systems) that can perform quadrillion of calculations per second.
Why is Super Computing so important?
HPC or super computing is the foundation for scientific, industrial and societal advancement and breakthrough for discoveries. It is through analysis of data that groundbreaking scientific discoveries are made, game changing innovations are fueled and the quality of life is improved for billions of people around the globe. This cannot be achieved without the high data processing power offered by super computing solutions.
Components of High Power Computing Systems
They consist of multiple computer servers networked together, software programs and algorithms running simultaneously on servers in clusters. The clusters are networked to the data storage to capture the output.
Some Examples of Super Computers around the World
- IBM Blue Gene/L: was invented by IBM in 2012 for simulation of nuclear weapons.
- KRANKEN: for bio-fuel and weather forecasting and modelling. Used by Tennessee University and Oakland ridge Laboratory in the U.S.
- K-Computer: designed by the Japanese company Fujitsu for diagnosing climate and disaster prevention.
- IBM Roadrunner: was developed by IBM for modelling the decay of the U.S nuclear weapons. It was the fastest supercomputer as of 2009 with a top speed of 1.042petaFLOPS. Unfortunately the IBM Roadrunner was shut down in 2013 due to low energy efficiency among other reasons.
Characteristics of Super Computing Systems
- High Computational Rates: having multiple processing units, HPC systems can interpret program instructions and execute tones of arithmetic and and logic operations per second. For example while a desktop with a 3GHz processor can perform around 3 billion calculations per second it cannot match or compare to High Power Computing Systems (Super Computing Systems) that can perform quadrillion of calculations per second.
- Faster Input/Output and File Systems speeds: Super Computing systems have powerful and quick read/write access to data of up to 13GB/sec which allows for rapid storage and quick retrieval of stored data,
- High speed connectivity between nodes: the large number of nodes (servers, CPUs and storage units) in HPC systems communicate via switches that operate at higher speeds of up to a bandwidth of 200GB/s between nodes in a network.
- Multitasking Capabilities: HPC systems are capable of supporting up to 100 users at the same time.
- Efficient Cooling system: HPC systems employ innovative ways such as liquid cooling for cooling the processing units and its components. This helps the units to stay below their throttle temperatures thus maintaining their high processing capacity.
I hope you found this article informative.